Transport Economics

BUTTON 9781786435668 PRINT.indd 1 27/04/2022 11:07

BUTTON 9781786435668 PRINT.indd 2 27/04/2022 11:07
 Transport Economics
4th Edition

Kenneth Button

University Professor, Schar School of Policy and Government,

George Mason University, USA

Cheltenham, UK • Northampton, MA, USA

BUTTON 9781786435668 PRINT.indd 3 27/04/2022 11:07

 © Kenneth Button 2022

All rights reserved. No part of this publication may be reproduced, stored in a

retrieval system or transmitted in any form or by any means, electronic, mechanical or
­photocopying, recording, or otherwise without the prior permission of the publisher.

Published by
Edward Elgar Publishing Limited
The Lypiatts
15 Lansdown Road
Cheltenham
Glos GL50 2JA
UK

Edward Elgar Publishing, Inc.

William Pratt House
9 Dewey Court
Northampton
Massachusetts 01060
USA

A catalogue record for this book

is available from the British Library

Library of Congress Control Number: 2022931653

ISBN 978 1 78643 566 8 (cased)

ISBN 978 1 78643 568 2 (paperback)
ISBN 978 1 78643 567 5 (eBook)

Typeset by Cheshire Typesetting Ltd, Cuddington, Cheshire

EEP BoX
BUTTON 9781786435668 PRINT.indd 4 27/04/2022 11:07
 Contents

Prefacex
List of acronymsxii

1 Transport, Economics, and Economists 1

1.1 Some background 1
1.2 A brief history of transport economics 5
1.3 The subject matter of transport economics 8
1.4 The economic characteristics of transport 15
1.5 This book 17
References 20

2 Transport and the Economy: Some Numbers 22

2.1 Introduction 22
2.2 The global picture 24
2.3 Transport at the national level 31
2.4 Local transport 37
2.5 Emerging trends 44
2.6 Some comments on the short-term effects of Covid-19 on
transport50
2.7 Where are the numbers? 52
References 54
Exhibit Response of air transport to a pandemic46

3 Transport and Location 55

3.1 The desire for movement and mobility 55
3.2 Transport: the ‘chicken’ or the ‘egg’? 57
3.3 Industrial location and transport 60
3.4 Economic gateways and corridors 67
3.5 Output, market areas, and transport costs 70
3.6 Urban transport and land values 74
3.7 Urban wages 80
References 83
Exhibit The impact of parking policy on house prices in the
Netherlands79

4 The Demand for Transport 85

4.1 Demands for transport 85
4.2 Influences on travel demand 87

BUTTON 9781786435668 PRINT.indd 5 27/04/2022 11:07

 vi TRANSPORT ECONOMICS, 4TH EDITION

4.3 Pricing a transport service 90

4.4 Trip purpose 96
4.5 Levels and methods of charging 97
4.6 The time period 98
4.7 The absolute level of the price change 100
4.8 Income levels 101
4.9 The price of other transport services 104
4.10 Tastes, human desires, and motives 106
4.11 The notion of a ‘need’ for transport 109
4.12 The valuation of travel time savings 112
4.13 The demand for car ownership 119
4.14 What does ‘behavioral economics’ tell us? 126
References 130
Exhibit Demand shocks on airline fares produced by high-speed rail
transport91
Exhibit Fuel efficiency of United States cars following the 1973 and
1979 oil crises93
Exhibit Meta-analytical synthesis of demand elasticity results103
Exhibit The role of ancillary revenues in airline finances108
Exhibit When London Underground workers go on strike128

5 Direct Costs of Transport 134

5.1 Factors influencing the supply of transport 134
5.2 Fixed and variable costs 136
5.3 Economies of scale, scope, density, experience, and commonality 149
5.4 Specific, joint, and common costs 155
5.5 Problems of common cost allocation: the road and rail track
cases157
5.6 Transport user costs and the notion of generalized costs 164
5.7 The bunching of public transport services 170
5.8 Economic performance 172
5.9 Costs and the measurement of economic efficiency 175
References 182
Exhibit The container and world trade143
Exhibit Measuring perceived costs of driving167

6 The External Economic Costs of Transport 186

6.1 Introduction 186
6.2 Externalities 187
6.3 Transport’s implications for the environment 190
6.4 The valuation of externalities 193
6.5 The magnitude of the environmental externality problem 201
6.6 Energy use 214
6.7 Introduction to traffic congestion 219
6.8 The economic costs of congestion 225

BUTTON 9781786435668 PRINT.indd 6 27/04/2022 11:07

 CONTENTS ­ vii

6.9 Refinements on the basic congestion model 227

6.10 Some broad aggregate calculations 232
References 233
Exhibit The economic costs of CO2 emissions196
Exhibit The economic costs of the Amoco Cadiz oil spill198

7 The Pricing of Transport 236

7.1 The principles of pricing 236
7.2 Matching supply with demand 237
7.3 Marginal cost pricing 243
7.4 Difficulties of ‘second-best’ situations 246
7.5 Price differentiation, price discrimination, and yield
management249
7.6 Pricing with stochastic demand 265
7.7 The problem of the peak 267
7.8 Transport subsidies, operational objectives, and pricing 270
7.9 Market instability, suboptimal supply, and the empty core 273
7.10 Indirect pricing 276
References 278
Exhibit The issue of predatory pricing240

8 Containing the Environmental Costs of Transport 281

8.1 Introduction 281
8.2 The main economic approaches 282
8.3 Marketable and tradeable permits 283
8.4 The OECD’s ‘polluter-pays principle’ 287
8.5 More on environmental standards 294
8.6 Transport subsidies and the environment 298
8.7 Protecting the sufferers 303
8.8 Energy use 304
8.9 Safety and accidents 315
References 318
Exhibit Marketable permits for lead in gasoline286
Exhibit The ‘Ubernomics’ of app-based ride-hailing301
Exhibit ‘Boris bikes’311

9 Optimizing Traffic Congestion 320

9.1 Economics and optimal traffic 320
9.2 ‘Road pricing’ 321
9.3 Applications of urban road pricing 326
9.4 Some difficulties with road pricing 329
9.5 Impacts of road pricing 338
9.6 Parking policies 342
9.7 Congestion pricing at airports 344
9.8 Seaports congestion 356

BUTTON 9781786435668 PRINT.indd 7 27/04/2022 11:07

 viii TRANSPORT ECONOMICS, 4TH EDITION

9.9 Non-pricing options for reducing congestion 358

9.10 ‘Micromobility’ 362
References 363
Exhibit The initial London congestion charge scheme340

10 Economics and Transport Logistics 366

10.1 Introduction 366
10.2 Transport logistics 367
10.3 The costs of warehousing and inventory holdings 370
10.4 Consolidation and trans-shipment 375
10.5 Mode choice 376
10.6 Urban logistics 380
10.7 Green logistics 382
10.8 International logistics 384
10.9 Big data, supply chains, and economics 386
10.10 Security 387
References 395
Exhibit Costs and benefits of transport security389

11 Investment Criteria: Private and Public Sector Analysis 398

11.1 Transport and infrastructure 398
11.2 Basic theories of investment policies 400
11.3 Commercial and social approaches to investment 403
11.4 Public–private partnerships 407
11.5 The theory of cost–benefit analysis 412
11.6 Coping with network effects 417
11.7 Cost–benefit analysis in practice and variations on the theme 419
11.8 Comparability between appraisal techniques 428
11.9 Assessing the effect on national income  431
11.10 Some institutional considerations 434
References 436
Exhibit The Third London Airport Study421

12 Transport Planning and Forecasting 438

12.1 The development of transport planning 438
12.2 The ethos of transport planning 443
12.3 Traffic modeling and forecasting 448
12.4 Sequential travel demand forecasting 453
12.5 Disaggregate modeling 460
12.6 Interactive and stated-preference modeling 463
References 465
Exhibit Blue-print planning: L’Enfant’s transport plan for
Washington DC439
Exhibit Accuracy in traffic demand forecasting451

BUTTON 9781786435668 PRINT.indd 8 27/04/2022 11:07

 CONTENTS ­ ix

13 Transport and Economic Development 468

13.1 Transport’s Role in Development 468
13.2 Economic growth theory and transport 469
13.3 Transport infrastructure investment and economic productivity 476
13.4 The multiplier impacts of a transport investment 481
13.5 Transport economics in less developed countries 486
13.6 The transport policy of the European Union 492
13.7 Transport effects on regional and urban development 504
References 512
Exhibit Employment implications of the United States’ federal
highway system471
Exhibit Railroads and canals in the United States’ economic
development474
Exhibit Infrastructure investment and economic productivity478
Exhibit Some macro- and microeconomics of elevator travel506

14 Political Economy and Transport Regulation 514

14.1 Underlying issues 514
14.2 Regulation theory 517
14.3 Monopoly power 522
14.4 Prioritizing transport policies 530
14.5 Paths of regulatory reform 541
14.6 Studying regulatory reform 546
14.7 Disruptive innovation 551
14.8 Coordination via the market, or by direction? 554
References 557
Exhibit Regulated and unregulated airlines526

Name index559
Subject index567

BUTTON 9781786435668 PRINT.indd 9 27/04/2022 11:07

 Preface

The first edition of Transport Economics appeared in 1982. This, the fourth
edition, provides an update of the subject together with current data, case studies,
and examples.
A lot has changed in the study of transport economics over the past 40 years,
and changes continue. Many of these are the result of a better understanding of
economics in general. Others are more transport-specific and reflect social, tech-
nical, and political developments. An updating of Transport Economics is seen as
necessary to reflect both of these broad categories of change.
Improved and more complete data, greater use of mathematics and advanced
econometrics, the emergence of experimental economics, and advances in com-
puter technology have seen more technical rigor introduced into economics. And
the changes have been part of the way transport economics has changed. This
revised edition of Transport Economics reflects some of these developments. But
these are refinements rather than revolutions. Much of our core understanding
of transport economics remains intact. The increased recognition of the role of
behavioral economics which allows relaxation of some traditional assumptions
regarding human behavior is, for example, still seen by most as a refinement
rather than a major paradigm shift.
Although conforming to general economic rules, transport is also very
contextual and contexts change. There have been, for example, major technol-
ogy changes, even over the decade since the appearance of the last edition of
this book. New forms of local transport have emerged including transportation
network companies like Uber and Lyft and micromobility modes such as electric-
scooters. The development of the smartphone, and in particular Apple’s iPhone
in 2007, has led to almost universal access to global positioning systems and, with
this, real-time information for route guidance and tracking. This has affected
both the ways individuals plan and adjust their travel as well as the ways in which
freight-supply chains are designed and operated. These, and other technology
shifts, influence the specifics of individual markets and are considered in this new
edition.
Additionally, much of the drafting of this edition took place during the
Covid-19 pandemic. The impact of this, and the associated policy actions to
reduce its spread, had immediate and obvious implications for transport at all
levels of aggregation. In many cases, such as commuting, shopping, and tourism
it reduced individual travel greatly. In other cases, as with the transport of medi-
cation and the home delivery of food and other supplies, it increased the move-
ment of goods is specific types of markets. Supply-chain economics has taken
on new dimensions in our lives. These are certainly matters of the moment and

BUTTON 9781786435668 PRINT.indd 10 27/04/2022 11:07

 PREFACE ­ xi

require the immediate attention of politicians and practitioners as well as filling

the media. But their longer-term effects are less certain.
While, at the time of writing, there has been evidence of a movement back to
something like what may be considered ‘the norm’ that prevailed before 2020, the
precise details of how the post-pandemic world will look are still unclear. What is
assumed here is that the fundamentals of transport economics essentially remain
intact even if the qualitative and quantitative parameters change post-Covid. In
other words, the theories and methodologies underpinning transport economics
remain valid, even though the magnitudes of individual parameters in specific
markets may have changed. The basic principles of supply and demand still hold:
if the price of gasoline rises, less will still be demanded even if the effect is quan-
titively different to past experiences.
The book retains the 14-chapter format of recent past editions. This broadly
conforms to a chapter a week per semester. The chapters, although following a
particular path, can also be used independently as free-standing modules.
To assist in pedagogy, 20 or so free-standing ‘Exhibits’ now provide a variety
of case studies and narratives to supplement the text. These are designed to add
greater flesh to the economic theory. More up-to-date examples and illustra-
tions also make the understanding of economic principles easier and assist in the
assimilation of economic concepts.
Additionally, there are changes to the referencing style compared to the three
previous editions. I adopt the Harvard Manual (author–date citation) style in
place of the Chicago (footnote) style. This is more in line with usage in modern
economic writing. The referencing is still light and focuses on material useful
when following up ideas and events. The book remains a mixture of prose, tables,
figures, graphs, and equations, the aim being to help understanding rather than to
overpower or impress the reader with advanced technique.
As in the past, I hope that teachers, students, and other readers will find the
new edition of Transport Economics interesting and useful.
Finally, a large word of thanks for the comments on previous editions that I
have received over the years from students, academics, practitioners, and others.
I may not have taken them all on board, but they have all been tremendously
helpful in the production of this work.

Kenneth Button

BUTTON 9781786435668 PRINT.indd 11 27/04/2022 11:07

 Acronyms

AVI automatic vehicle identification

BAH British Airports Holdings
BAC blood alcohol concentration
BART Bay Area Rapid Transit (United States)
BOOT build, own, operate, and transfer
BTC British Transport Commission
BTU British thermal unit
CAA Civil Aviation Authority (United Kingdom)
CAB Civil Aeronautics Board (United States)
CAFC company average fuel consumption (Canada)
CAFE corporate average fuel economy (United States)
CBA cost–benefit analysis
CBD central business district
CBSA core-based statistical area
CFC chlorofluorocarbon
CIF cost, insurance, and freight
CO carbon monoxide
CO2 carbon dioxide
CORSIA Carbon Offsetting and Reduction Scheme for International
Aviation
Covid coronavirus disease
CRS computer reservation system
CTP Common Transport Policy (European Union)
C-TPAT Customs–Trade Partnership Against Terrorism
dB decibel
DBF design, build, and finance
DBFM design, build, finance, and maintain
DBFOM design, build, finance, operate, and maintain
DEA data envelopment analysis
DICE dynamic integrated climate change
EPA Environmental Protection Agency (United States)
EU European Union
EUAA European Union Aviation Allowances
EU ETS European Union Emissions Trading System
FAA Federal Aviation Administration (United States)
FAMS Federal Air Marshal Service
FFDO Federal Flight Deck Officer
FHWA Federal Highways Administration (United States)

BUTTON 9781786435668 PRINT.indd 12 27/04/2022 11:07

 xiii
ACRONYMS ­

FOB free on board

FOC flight operating cost
FUND framework for uncertainty, negotiation, and distribution
GATT General Agreement on Tariffs and Trade
GDP gross domestic product
GNP gross national product
GPS global positioning systems
HATS household activities travel simulator
HC hydrocarbons
HGV heavy goods vehicle
HOV high-occupancy vehicle
HSR high-speed rail
IAM integrated assessment model
IATA International Air Transport Association
ICAO International Civil Aviation Organization
ICC Interstate Commerce Commission (United States)
IPSB installation of physical secondary barriers
ITS intelligent transport system
LRMC long-run marginal cost
MAC marginal abatement costs
Maglev magnetic levitation rail system
MIDAS Motorway Incident Detection and Automatic Signalling
(United Kingdom)
mpg miles per gallon
MSA metropolitan statistical area
NAFTA North American Free Trade Agreement
NATA New Approach to Appraisal
NATS National Air Traffic Services (United Kingdom)
NEF noise exposure forecast
NHS National Highway System (United States)
NHTSA National Highway Traffic Safety Administration (United States)
NNI Noise and Numbers Index
NOX nitrogen oxides
NTT new trade theory
OECD Organisation for Economic Co-operation and Development
OPEC Organization of the Petroleum Exporting Countries
O3 ozone
PAGE policy analysis of the greenhouse effect
PAYD pay-as-you-drive
Pb lead
PBS planning balance sheet
pcu passenger car unit
PNdB perceived noise decibel
PPP public–private partnership
PSV public service vehicles

BUTTON 9781786435668 PRINT.indd 13 27/04/2022 11:07

 xiv TRANSPORT ECONOMICS, 4TH EDITION

PTA passenger transport authority

PZEV partial zero emission vehicle
RHTM regional highway traffic model (United Kingdom)
RRL Road Research Laboratory
RTA regional trade area
SCC social cost of carbon
SELNEC South-east Lancashire, North-east Cheshire (United Kingdom)
SO2 sulfur dioxide
SRMC short-run marginal cost
SUV sports utility vehicle
TEN Trans-European Network
TEU twenty-foot equivalent unit
TFP total factor productivity
TGV Train à Grande Vitesse (France)
TNC transportation network company
TPP Transport Policy and Programme
UPTPS Urban Public Transport Planning System
US DOT United States Department of Transportation
USMCA United States–Mexico–Canada Agreement
UNCTAD United Nations Conference on Trade and Development
VOC volatile organic compound
WTO World Trade Organization
ZEV zero emission vehicle

BUTTON 9781786435668 PRINT.indd 14 27/04/2022 11:07

 1 Transport, Economics, and Economists

1.1 Some Background

Clifford Winston (2013) begins his assessment of the state of United States trans-
port by summarizing exactly what transport1 is:

Transportation is a friction – a cost in both money and time – that must be incurred
by individuals and firms to complete almost any market transaction. An efficient and
extensive transportation system greatly enriches the standard of living in modern
society by reducing the cost of nearly everything in the economy; expanding indi-
viduals’ access to and choices of employers and employers’ choices of workers; ena-
bling firms and urban residents to benefit from the spatial concentration of economic
activities, referred to as agglomeration economies; reducing trade costs and allowing
firms to realize efficiency gains from specialization, comparative advantage, and
increasing returns; and limiting firms’ ability to obtain market power by locating in
geographically isolated markets with no competition.

In 1982 I opened the first edition of Transport Economics with a quote taken
from an address to the United Kingdom’s Chartered Institute of Transport by
K.J.W. Alexander (1975). He made the rather sober point that despite the impor-
tance of transport in the economy:

the number of academic economists who specialize exclusively in transport could

probably be counted on two hands. If one adds to these economists the applied
economists employed in the transport business and the specialist consultants working
exclusively in that field I would be surprised if the total number exceeded sixty or
seventy.

In other words, while transport is, as Winston highlighted nearly 40 years

later, a crucial part of any successful economy, the number of active transport
economists is relatively small. There are certainly more than 60 or 70 today, but
transport still tends to be dominated by engineers, and more recently by those
engaged in supply-chain management.
This paucity of transport economists is rather strange given the fascina-
tion that transport has held for very eminent economists. Indeed, John Maynard
Keynes, perhaps the most eminent and disruptive economist in the twenti-
eth century, once confessed that when young, ‘I find economics increasingly

1 The English term ‘transport’ is used throughout as opposed to the more common
American term ‘transportation’ to avoid any confusion with the penal sentences given
in the past to British criminals. Proper nouns are exempted.

BUTTON 9781786435668 PRINT.indd 1 27/04/2022 11:07

 2 TRANSPORT ECONOMICS, 4TH EDITION

satisfactory, and I think I am rather good at it. I want to manage a railway, or

organize a trust or at least swindle the investing public’.
We also find that some of the more contemporary economists began their
academic careers delving into transport-related questions. The Nobel Prize
awardee James Buchanan, for example, was early in his career interested in road
finance and indeed wrote his Master’s dissertation on ‘Gasoline tax sharing
among local units of government in Tennessee’. Another laureate, Vernon Smith,
had an interest in aircraft landing slot pricing, and yet another, William Vickrey,
in road congestion and public transport supply. Maurice Allais, the 1988 laureate,
gave his name to the European Economic Community’s 1965 report, ‘Choice of
pricing policy for transport’, perhaps a somewhat unexpected exercise for a physi-
cist turned theoretical economist. Drifting to the extreme periphery of transport-
related work, Richard Thaler, the 2017 recipient of the award, made significant
advances in behavioral economics by, among other things, studying the use of
male urinal pans at Amsterdam Airport!
The situation as seen by Alexander in the mid-1970s largely continued in the
1990s when the second edition of Transport Economics was published. There were
more transport economists in the public sector generally assisting in engineering
activities and investment appraisal, but still their number was not large. Most aca-
demic transport economists were likewise linked to civil engineering departments,
with a few in business schools. In the United States, it was often taught as part of
a public utilities program.
By the time the third edition of Transport Economics went into proofs in
2009 there were certainly more, but still relatively few economists specializing
in the subject. But there was one significant change. The field of transport
logistics had developed considerably and was making extensive use of econom-
ics. One reason for this was the widespread deregulation of transport markets
throughout the world from the 1980s. This had resulted in suppliers of transport
services being subjected much more to market forces. They were less restricted
by ­government-imposed constraints and essentially had to think for themselves.
They needed to maximize efficiency to survive in competitive conditions. Put
simply, transport economists could help a company’s bottom line by highlighting
opportunity costs.
Of course, the world had also moved on in other ways. Logistics involv-
ing the planning, implementing, and controlling procedures for efficient and
effective transport and storage of goods, including services and related informa-
tion, from the point of origin to the point of consumption, had increased in
importance. The world economy had also become more global both in terms of
person and goods movement and inevitably this threw up a variety of new trans-
port challenges. There were now mega global supply chains and manufacturers
dominating the value of trade. Economics offers a portfolio of models to tackle
some of these issues arising in the new world. It has become a core component of
professionally oriented degrees in managerial science.
A second important change by 2009 was that people had become seriously
concerned with the state of the environment, and notions such as sustainable

BUTTON 9781786435668 PRINT.indd 2 27/04/2022 11:07

 TRANSPORT, ECONOMICS, AND ECONOMISTS ­ 3

development had taken center-stage. Transport imposes a variety of environment

costs on society, ranging from the very local problem of urban traffic noise to
emitting huge amounts of gases that contribute to global warming. Economics
provides ways of assessing the potential costs and benefits of policies designed to
contain, if not always remove, the adverse environmental costs of transport.
In subsequent editions of this book, I therefore saw little reason to drop
Alexander’s quote, and, although there had been something of a minor upsurge
in interest in transport economics, the subject still forms a small branch of the
economic discipline.
Moving into the 2020s, the important role of transport logistics and the
issues surrounding sustainable transport remain, and while the growth in world
trade has been temporarily slowed by the Covid-19 pandemic, transport econom-
ics seems to have become more important. Accompanying this, the information
systems required to ensure networks work smoothly have grown in sophistication
and created their own economic challenges. Individuals, for example, use global
positioning systems (GPS) to plan and track their own movements and busi-
nesses, whether public or private, make extensive use of information systems in
tracking, controlling, and managing the transport of their own goods, or in the
case of passenger modes, their ‘human cargoes’. Network and information eco-
nomics now inevitably interfaces and overlaps with transport economics.
We also have new technologies and transport systems. While these often
involve greater mass movement by mega mode – for example, super jumbo-jets
and larger container ships – they also entail innovations at the local level with the
emergence of transport network companies such as Uber, and micromobility as
offered by electric-scooters (commonly known as ‘e-scooters’). The managerial
and operational models for some of these latter modes, and perhaps especially
ride-hail companies, have been developed with considerable input from the econ-
omist, while in other cases there have been economic interests on the regulatory
side (Button, 2020).
The Covid-19 pandemic of 2020–22, the spread of which was in part due
to the ease of mobility we now enjoy, brought about, at least in the short term,
a sea change in transport policy. It led to many forms of transport coming to a
standstill, or at least a crawl, as social distancing and quarantining became almost
universal. Economists were aware of the potential effects that ease of mobility
and contact can have on the spread of disease. Indeed, another Nobel economist,
Robert Shiller (2017), had already dwelt on the Kermack–McKendrick theory
of disease epidemics and their similarity to economic bubbles when giving his
Presidential Address to the American Economic Association in 2015.
Nevertheless, Alexander’s statement also highlighted the fact that many
transport economic issues have traditionally been under-researched and often
poorly understood by non-economists. This situation has, however, improved
somewhat since the mid-1970s. One indicator is the increase over the years in
the number of serious academic journals that regularly publish papers on trans-
port economics. Most notable of these are the specialist Journal of Transport
Economics and Policy, which first appeared in 1967, the International Journal

BUTTON 9781786435668 PRINT.indd 3 27/04/2022 11:07

 4 TRANSPORT ECONOMICS, 4TH EDITION

of Transport Economics, which came a little later, and more recently Research in
Transport Economics. But there are also publications such as the Journal of Urban
Economics, Journal of Regional Science, Journal of Transportation, Journal of
Transport Reviews, Journal of Urban Economics, and the Transportation Research
series of journals that often carry transport economics articles. In addition, more
mode-specific journals have emerged, such as Maritime Economics and Logistics
and the Journal of Air Transport Management, which often carry material focus-
ing on economic issues. But, nevertheless, compared to many areas of economic
study, transport remains remarkably neglected.
Everything is also relative, and there have been some surges of interest in
certain aspects of transport economics. One example of this occurred in the
­mid-1980s, when Clifford Winston (1985), in examining developments in trans-
port economics, was able to comment on ‘a current intensity [of interest in the
field] not witnessed for more than fifty years’, this surge being mainly related to
the analysis of the impacts of more liberal or, to use the American term, ‘deregu-
lated’ transport markets that were emerging. More recently, the role of transport
as a facilitator of economic growth has attracted attention both at the microeco-
nomic level and at the macro. Matters of transport security and the financing of
transport networks were themes attracting considerable attention during the early
part of the twenty-first century, as was the role of transport in international trade.
This relative lack of contemporary academic interest in transport economics
is surprising because transport problems have in the more distant past stimulated
major developments in general economic theory. This includes the development
of the notion of consumer surplus by French economist engineers, such as Jules
Dupuit, in the 1840s, the refinement of cost allocation models by John Bates
Clark, Frank Taussig, and others in the early part of the twentieth century, the
examination of the marginal cost pricing principle to improve the charging of
rail services, and work on congestion and the environment by Arthur Pigou in
the 1920s.
More recently, the advent of computers has encouraged work on applied
econometrics (for example, the development of discrete choice models in con-
sumer theory by Dan McFadden, and refinements to flexible-form cost functions)
and on mathematical programming (including the use of data-envelopment anal-
ysis to assess the relative efficiency of different transport suppliers) using trans-
port case studies as a basis for analysis. The advent of ‘big data’ sets has more
recently provided fertile inputs for the empirical analysis of transport economic
issues (Milne and Watling, 2019). Nevertheless, full-time transport economists
in universities remain thin on the ground, and their numbers in businesses and
government are not much larger.
In the past, the relatively small number of clearly identifiable specialist trans-
port economists may have been attributable to the diverse nature of the industries
involved in their work that range from taxi-cabs and bicycles to oil-tankers and
jumbo-jets with their various institutional and technical peculiarities. Non-
economists – lawyers, management scientists, and engineers – often have special-
ist knowledge of the industry concerned with a secondary background in some

BUTTON 9781786435668 PRINT.indd 4 27/04/2022 11:07

 TRANSPORT, ECONOMICS, AND ECONOMISTS ­ 5

key aspects of economics needed in their work. At the applied level, numerous
small firms that do not have the resources to employ full-time specialist transport
economists does not mean they do not have individuals that fulfil some economic
role in the organization.
The paucity of professional transport economists may also be explained by
a tendency for physical planners and transport engineers to dominate substan-
tive investment and policy decision-making within the sector. Where economics
is used in this context it is often within the engineering framework and carried
through by engineers: so-called ‘engineering economics’. This situation has only
gradually been changing with innovations in management and planning, and
in the way that political institutions have evolved. Transport is now seen in a
much wider context. But this still leaves the questions: What is modern transport
economics? How did it evolve and what are the intellectual and practical driving
forces underlying it?

1.2 A Brief History of Transport Economics

As with most academic ‘disciplines’, economics has never been easy to define.
Not too seriously, the Chicago economist Jacob Viner is meant to have said,
‘Economics is what economists do’. This is not exactly helpful to the ­non-transport
economists’ world. But it is generally agreed that economics is about human
behavior and resource allocation, and about the allocation of scarce resources. It
focuses a lot on the role of markets in doing this, but not exclusively so. Various
aspects of it border on a variety of other disciplines such as law, sociology, and
engineering, often making boundaries vague. It may also be essentially descrip-
tive in the sense of it approaching a value-neutral science (it tells us that in most
conditions the quantity demanded of something will fall if its price rises), but it
can also contain elements of prescription (it is a good thing for social welfare if
prices are low), the latter often being called political economy. All these elements,
and some others, become important when studying the economics of transport.
What was called Modern Transport Economics by Michael Thomson (1974)
began slowly to take shape in the late 1960s and early 1970s. Up until that time,
as Rakowski (1976) points out, ‘the field had essentially been in a state of semi-
dormancy since the 1920s’, and while considerable institutional studies had been
conducted, very little analytical work as we would understand it today had been
attempted. Indeed, the two standard American textbooks on transport econom-
ics in the late 1960s were D.F. Pegrum’s Transportation: Economics and Public
Policy and Phillip Locklin’s Economics of Transportation, both of which went to
numerous editions. These texts were highly institutional in approach and, as an
indicator of their divergence from textbooks of the 2020s, contained only five or
so line drawings between them and no equations.
If one is looking for a watershed in the history of modern transport eco-
nomics it would probably be the publication of John Meyer et al.’s Economics
of Competition in the Transportation Industries in 1959, which provided a

BUTTON 9781786435668 PRINT.indd 5 27/04/2022 11:07

 6 TRANSPORT ECONOMICS, 4TH EDITION

state-of-the-art review of resource allocation problems in the sector, coupled

with rigorous statistical analysis. There had been several seminal economic works
relating to transport that appeared prior to this, including James Buchanan’s re-
examination of congestion and Beckmann et al.’s set of studies from 1956, but
Meyer’s book really set the tone for the way transport topics are analysed.
Interest in transport matters expanded throughout the 1960s with, for
example, William Vickey’s analysis of road charging and the role of public transit
subsidies, Ian Little and James Mirrlees’ development of tools for investment
appraisal and especially of cost–benefit analysis (CBA), Quandt and Baumol’s
improvements in methods of transport demand modeling based on McFadden’s
work and the ideas of Kelvin Lancaster, Moses and Williamson’s development
of innovative methods of travel time evaluation, and so on. All these are topics
we shall return to later, albeit in their more modern manifestations, and in many
cases with some real-world examples of applications.
The changes were not all academic. This period also saw public ­policy-makers
become more interested in the role of economics in improving decision-making
across the board. In many countries specialist departments were established in
national, state, or municipal government to examine the economic implications
of various transport policy options.
The 1960s and early 1970s saw a slew of official reports appear. In the United
Kingdom these included ‘Road track costs’, the Smeed Report on ‘The econom-
ics of road pricing’, the Beeching report on ‘The reshaping of British railways’,
and the report of the Roskill Commission on the Third London Airport. The last
set out the framework for the way economics can help in mega-project investment
appraisal. The European Union, then just established as the European Economic
Community, produced reports and plans in the early 1960s for integrating
and coordinating economic approaches to transport provision across much of
Europe including the Allais Report on ‘Choice of pricing policy in transport’.
In the United States this was the period of the large land-use transport studies
including the Chicago Area Transportation Study from the late 1950s, the Puget
Sound Regional Transportation Study, and the Detroit Regional Transportation
and Land-use Study. These were largely engineering studies seeking to deal with
mounting congestion problems, but economic analysis was part of the assessment
process, although not always well applied (Meyer and Gómez-Ibáñez, 1983).
While it is seldom easy to explain the timing of change, the reasons for the
renewed research interest in the 1960s, especially in the United States, Rakowski
(1976) attributes to ‘the problems of physical distribution and the development
of a new field which has come to be called business logistics’, ‘expanded interest
in all phases of urban transportation’, and ‘a great deal of research in the areas
of transport in the developing countries’. Kenneth Gwilliam (1980) echoes these
themes but places emphasis on the growing problems of urban transport in the
1960s and the recognition that land-use and transport needed to be considered
together, and usually simultaneously, if the problems were to be successfully
tackled. As a result of this, he argues, ‘the boundaries between transport econom-
ics and urban and regional planning were obscured’.

BUTTON 9781786435668 PRINT.indd 6 27/04/2022 11:07

 TRANSPORT, ECONOMICS, AND ECONOMISTS ­ 7

While the broad underlying intellectual thrust of modern transport has not
significantly changed since the 1970s and early 1970s, namely the use of economic
theory to enhance transport efficiency in the broadest sense, the topics of focus
have shifted to some extent and the economic tools used have been improved
and fine-tuned. The late 1970s saw concern in many parts of the world with the
macroeconomic problems associated with ‘stagflation’: the simultaneous occur-
rence of high levels of unemployment and poor economic growth accompanied
by inflation. The economics that arose placed emphasis on improving industrial
efficiency, including that of transport. The failure of a large part of the United
States railroad industry became a catalyst for some of the major changes in trans-
port. The adoption of ‘Reaganomics’ in the United States and Thatcherism in the
United Kingdom, involving similar approaches towards what became known as
‘supply-side economics’, led to a homing-in on reducing costs in heavily regulated
industries such as transport.
The empirical analysis of researchers such as William Jordan and Michael
Levine provided evidence of excess costs associated with regulation, the theories
of Baumol and others on contestable market structures offered a new way of
looking at competitive forces, and Harold Demsetz’s work provided a basis for
enhancing the efficiency with which transport infrastructure is provided and
maintained. Work by a variety of Chicago economists, including George Stigler
and Sam Peltzman, began to focus on the motivations of those responsible for
framing and administrating regulatory regimes, arguing that they seldom served
the public interest. The outcome was what some have called the ‘age of regulatory
reform’ (Button and Swann, 1989), which focused on removing distorting regula-
tions and allowing greater market flexibility.
More recently public concern, combined with academic curiosity, has led
to economists switching more of their attention to matters pertaining to the
environmental implications of transport supply, the role that economics can
play in enhancing the transport logistics supply chain, and the financing of
infrastructure. Transport, as we will see, can impose considerable strains on the
environment and, although this has long been recognized, the scale and nature of
transport has changed, as has scientific knowledge on the implications of these
strains – and especially so in the context of global impacts.
Frances Cairncross’ term the ‘death of distance’ also came to the fore in the
early 2020s. While many aspects of this are associated with the rise of information
technology and the speed of information dissemination, transport also played a
significant interactive role. Globalization, internationalization, and domestic eco-
nomic growth within many countries have been, in part, the result of improved
logistics, including just-in-time supply-chain management. Although often not
appreciated, many of these improvements in logistics are the result of the applica-
tion of basic economic principles.
Finally, modern transport requires an extensive infrastructure of roads, rail
tracks, ports, air traffic control systems, bridges, and so on. The construction
and the maintenance of this infrastructure must be paid for. While in the past
transport economics has focused primarily on deciding which elements to build

BUTTON 9781786435668 PRINT.indd 7 27/04/2022 11:07

 8 TRANSPORT ECONOMICS, 4TH EDITION

or repair, with the tax-payer bearing much of the financial burden, there is now
concern with the methods of finance to use and the overall amount of national
resources that is being devoted to various types of infrastructure. This issue is
increasingly important in many mature economies as old infrastructure becomes
obsolete or needs maintenance, but is perhaps more critical in lower-income
nations lacking many key transport networks.
When considering this background, it is important to emphasize that trans-
port economics is not distinct from all other branches of economics. Indeed,
many of the seminal papers on the subject have appeared in the general econom-
ics literature and have often been produced by individuals with a broad interest in
economics rather than being transport specialists. Nevertheless, as in most areas
of study, as our knowledge has increased, it has become difficult, if not impos-
sible, for economists to follow developments in all branches of their discipline.
The Renaissance woman or man of economics is in the past. There has been an
inevitable increase in specialization. Transport economics has in general terms a
long history, but, as Rakowski and others suggest, it was only in the 1970s or so
that it became a major field of academic study within universities, and only sub-
sequently did a body of specialists emerge. So the modern transport economist is
a relatively new beast. But what is it that he or she studies?

1.3 The Subject Matter of Transport Economics

The scope of each of the various subdisciplines of economics (for example,

agricultural economics, development economics, and public sector economics) is
determined not so much by ‘schools of thought’ or philosophies, but rather by
the type of subject matter examined and the problems tackled. Transport econo-
mists are interested in the economic problems of moving goods and people; they
are not normally so concerned with either the industries producing the vehicles
and infrastructure (aircraft manufacturing, road construction companies, ship
building, etc.) or with some of the very wide implications of transport policy (for
example, on the balance of payments), although matters to do with the environ-
ment certainly attract increasing amounts of their attention. Of course, this does
not mean that transport issues are viewed in complete isolation from their wider
context but it does mean that the main emphasis and thrust of analysis is directed
towards the more immediate transport implications.
While much of the economic analysis of transport issues is at the micro level
(for example, looking at the decisions of individuals or firms) or at the meso level
(focusing on transport industries or the importance of transport on a specific
region), there is also some interest in the macroeconomic impacts of transport,
for example on its effects on national productivity, globalization of trade, or labor
force migration.
This book seeks to give adequate coverage to all levels of aggregation.
Initially, however, we step back a little and spend time considering the economic
institutions within which transport services are supplied and on individuals and

BUTTON 9781786435668 PRINT.indd 8 27/04/2022 11:07

 TRANSPORT, ECONOMICS, AND ECONOMISTS ­ 9

firms seeking its services. Much of the theoretical economics that we encounter
is rather abstract and largely assumes away the role of economic institutions,
although this is now changing gradually. Institutional structures are important.
Elementary economics, for example, talks about things like perfect competition,
but such markets could not exist without laws giving property rights to suppli-
ers or without contracts between sellers and buyers. As the Nobel Prize-winning
economist Ronald Coase (1992) said in the context of economic reforms in the
former USSR countries, ‘[t]hese ex-communist countries are advised to move to
a market economy, and their leaders wish to do so, but without the appropriate
institutions no market economy of any significance is possible’.
Traditionally, however, within economics, institutions have largely been
treated as exogenous, a factor that is to be taken as given, and forming part of the
ceteris paribus assumptions of the economic analysis. Few economists spent time
studying it, and those who did, such as Thorsen Veblen and Kenneth Galbraith,
were treated in their day as rather marginal to mainstream economics – some even
considering them ‘sociologists’! However, this view is changing, and as Oliver
Williamson (2000) observed, ‘[w]e are still very ignorant about institutions. …
Chief among the causes of ignorance is that institutions are very complex. That
neoclassical economics was dismissive of institutions and that much of organiza-
tion theory lacked scientific ambitions have also been contributing factors’.
The so-called ‘new institutional economics’ has moved the economic study
of institutions away from being a largely descriptive, legalistic, and traditional
historical way of viewing the world to one that offers microanalysis of such issues
as to why economic institutions have emerged in the way they have. It has also
moved away from a largely negative way of looking at neoclassical economics
to one with its own theoretical constructs and tools of analysis. This approach
importantly sets the neoclassical economics that underlies most of the work in
transport economics up until the 1990s within a larger context. Figure 1.1 is a
simplified diagram that sets out four levels of social analysis. It is self-explanatory
except that the solid arrows show constraints that come down from higher levels
of analysis, while the dotted arrows indicate the direction of feedbacks. The data
offered in the figure represent the rough frequency in years over which change
takes place.
Long-term issues tend to receive less attention in transport economics the
further one moves from narrow resource-allocation matters. Their importance
cannot be neglected, however. There have been, perhaps, two major societal
changes that have impacted on culture in the broadest sense since the end of the
Second World War, and have had important ramifications for transport.
First, there was the demise of the Soviet Union and the move away from cen-
tralized planning, together with the gradual uptake of capitalism in China. These,
and similar developments elsewhere, represent major shifts in the embedded atti-
tudes of these countries and with this has come new ways of thinking about the
role of transport in society and the ways in which it should be provided.
Second, there have been changing attitudes towards free trade, and with
these have come internationalization and globalization. The resultant approach

BUTTON 9781786435668 PRINT.indd 9 27/04/2022 11:07

 10 TRANSPORT ECONOMICS, 4TH EDITION

Resource allocation and

Neoclassical economics/agency
employment:
theory
continuous

Governance, especially
contracts: Transaction cost economics
1 to 10 years

Institutional environment,
Economics of property
formal rules:
rights/political economy
10 to 100 years

Embeddedness, customs
and traditions: Social theory
100 years or more

Figure 1.1 Institutions and economics

to tariffs and other barriers to international economic policies have had profound
effects on shipping, air transport, and many forms of surface transport. The
establishment of the League of Nations in 1920 laid the way for this, and it has
extended since the end of the Second World War. It is not just a matter of global
institutions such as the World Bank, the International Maritime Organization,
and the International Civil Aviation Organization being formed, but rather that
there is now a belief that they have a durable role to play in society. This forms a
basis for new economic systems and an embedded global view on how business,
including transport, can be conducted.
The institutional environment box in Figure 1.1 involves formal laws, regula-
tions, and rules, and at a higher level, ‘constitution’. It is also concerned with the
instruments of law – legislatures and bureaucracies – and mechanisms of enforce-
ment. For example, it relates to legal reforms such as the 1978 Air Transportation
Act that liberalized domestic aviation in the United States, or the Single European
Market Act that came into force in Europe in 1987 and led to a phased deregu-
lation of transport and other markets within the European Union. But it also
importantly includes the issue of property rights and their enforcement. So-called
‘Pigouvian externalities’, such as pollution, and ‘club good’ problems of traffic
congestion are the most obvious areas of interest for transport economists in this
regard. From an economics-of-transport perspective, the question becomes one
of getting the rules right to meet societal demands.
Governance – basically the way business is done – is important and here the
economic emphasis is on notions of contracts. Strictly, it is about crafting order,
and thereby mitigates conflict and realizes mutual gains. There are many factors –
cultural, structural, cognitive, and political – that influence the way these formal
and informal institutions develop. If we take shipping as a case study, governance

BUTTON 9781786435668 PRINT.indd 10 27/04/2022 11:07

 TRANSPORT, ECONOMICS, AND ECONOMISTS ­ 11

embodies such things as the evolving structure of coordination of shipping

activities that began with conferences – essentially cartels – in the 1970s and has
developed through consortia into strategic alliances, whereby changes occur dis-
cretely rather than continuously. These, like their brethren in the airline industry,
can involve agreements on such things as schedules and rates, with flexibility over
which vessels carry any individual assignment.
There are also contractual issues within the larger integrated supply chains
that, for example, tie, to varying degrees, shipping with other elements in the
logistics system, most immediately ports but extending beyond this. Governance
is different from government in the sense that governance involves informal agree-
ments within a legal structure. As an example, the United States deregulated its
domestic airline industry by an act of congress, whereas the United Kingdom
essentially got the same result because the Civil Aviation Authority reinterpreted
the existing set of laws. Because of high transactions costs – legal costs and the
like – the recent trend has seen a considerable amount of contract management
and dispute settlement action be dealt with directly by the parties involved, in
other words, within market frameworks rather than by legal actions.
Finally, there are the ongoing processes involving the day-to-day activities
that take place involving transport. This is essentially the traditional bread-and-
butter of micro- and mesoeconomic analysis taking into account the specific
nuances of transport. Resource-allocation issues have, since the days of Adam
Smith and more especially Alfred Marshall in the late 1890s, centered on getting
the marginal conditions right. To assess how this might be achieved in theory, and
how it materialized in practice, requires some notion of nature of the relevant cost
and demand conditions – the guts of most introductory economics courses. The
basic principles, although the nuances are still being refined, are pretty well under-
stood and much of the recent attention in transport economics has therefore been
on quantification and, in particular, quantification of networks.
Historically, however, this has not always been the case. Much of the early
analysis of market structures in transport, for example, paid scant heed to the
network nature of the industry and ipso facto to network externalities. Much
of the interest until the 1980s or so, as we have hinted at previously, had been
on traditional economic parameters such as market-concentration ratios, the
degrees of product differentiation, and the levels of vertical integration. While
all can be important instruments in the economist’s toolkit, the network nature
of the transport, and the service nature of its outputs, received much less direct
attention in the analysis. Much of the early work on transport demand modeling
was also either devoid of any of these network features, or embraced them in a
mechanical fashion. But network features are important, although they often
inconveniently add complexity to useful analysis.
All network service industries have essentially identical economic charac-
teristics. Their output is non-durable, although they normally require significant
amounts of infrastructure. There are also often very significant external benefits
that, tied with various forms of scale effects, produce network economies. From
the consumers’ perspective, these external economies are related to the fact that

BUTTON 9781786435668 PRINT.indd 11 27/04/2022 11:07

 12 TRANSPORT ECONOMICS, 4TH EDITION

the larger the number of links in a network, the greater the degree of connectivity,
and with this the larger the choice set available to them – often called economies
of market presence. On the cost-savings side, the ability, where this is most effi-
cient, to channel traffic through hubs, rather than carry traffic directly between
origins and destinations, creates economies of scope (cost savings by combining
traffic) and of density (the more intensive use of the mode). But complex net-
works can also create costs, and most especially when congestion develops on link
roads or flight paths, or at hubs such as sea- and airports.
So, what does all this mean regarding the day-to-day work of transport
economists? In summary, an understanding of economic institutions is now
generally seen as important for analysing broad transport issues, nevertheless the
main ‘tools’ of the transport economist are still taken directly from the toolkit
of standard micro- and mesoeconomic theory, although one should add that the
actual implements used have changed significantly over the years. The pre- and
immediate post-Second World War emphasis centered on the transport indus-
tries (that is, the railways, road haulage, shipping, etc.) and on ways in which the
transport supply could be improved within largely regulated structure so that
maximum benefit would be derived from public and private transport operations.
The situation was summarized by one geographer who felt that transport eco-
nomics at that time was concerned almost entirely with ‘matters of organization,
competition and charging, rather than with the effects of transport facilities on
economic activities’ (O’Connor, 1965).
To some extent, and especially in relation to international transport and,
to a lesser extent, inter-urban transport, this interest has remained. It has,
however, more recently been supplemented by concern with the wider welfare
and spatial implications of transport. Greater emphasis is now placed on the
environmental and distributional effects of the transport system and, in some
cases, market efficiency is seen as an undesirably narrow criterion upon which to
base major decisions. As Alexander argued, in the speech cited at the beginning
of this chapter, one of the most important roles for economists is to make clear
the overall resource costs of transport rather than just the accounting costs. It is
no accident, perhaps, that much of the early work in CBA was in the transport
field.
Transport economics has, like virtually all other branches of economics,
become more mathematical and quantitative in recent years. At the theoretical
level, it now often seeks to develop a sequence of formalized models that try to
express a complicated reality. The dominant pre-war idea that economics is con-
cerned mainly with establishing broad principles (for example, that the quantity
demanded rises, ceteris paribus, as price falls) has also given way, with the advent
of econometric techniques and the computer age, together with improved data
sources, to attempts at detailed measurement (that is, a rise of x tons in the quan-
tity demanded will, ceteris paribus, follow from a $y fall in unit price). Transport
economists are now heavily involved in trying to assess the precise quantitative
effect of different policy options and with forecasting likely changes in transport
demand.

BUTTON 9781786435668 PRINT.indd 12 27/04/2022 11:07

 TRANSPORT, ECONOMICS, AND ECONOMISTS ­ 13

At the public policy level, the increasing sophistication of transport opera-

tions, combined with both the long lead times that are required for full policy
implementation and the financial costs involved, place mounting strains upon
economists to produce useful, quantified predictions of future trends. The
increased appreciation of the commercial benefits of better forecasting and
costing as part of the supply-chain management process has led to greater
employment of numerate economists in the private sector as well as the public.
The rise of public–private partnerships has added to this trend.
From the relatively small base of the 1960s, transport economists, although
not abundant, have now become established in most areas of transport policy-
making at all administrative levels. Their increased interest in the overall welfare
consequences of different transport strategies, together with a willingness to
attempt some form of quantitative assessment, has led to transport economists
becoming closely involved in major transport planning exercises. They have an
established role in advising on appropriate actions at the national policy for-
mulation level, but this has also spread down to a more specific function at the
local and, where relevant, state planning level. At the strategic level, transport
economists, for example, have made a significant contribution to the urban
transportation planning in industrial countries since the late 1960s, and have
been involved in many detailed appraisals of traffic investment and management
schemes in developed countries. Multi-lateral agencies, such as the World Bank
and the African, Asian, and Inter-American Development Banks, play roles in
the latter.
In line with this, the toolkits of transport economists have changed. The
advances made in transport investment appraisal – most notably the development
of CBA techniques – as practical instruments of analysis led to the adoption of
economic criteria for the assessment of many large-scale investment projects in
the 1960s and 1970s (for example, the Third London Airport scheme and Victoria
Underground Line in the United Kingdom). Recent refinements have resulted in
rather more uniform CBA procedures being employed as a standard method of
small-scale transport project appraisal. As we see later, these include the COBA
and NATA packages developed in the United Kingdom for trunk road invest-
ment appraisal, and the procedures favored by the World Bank – most notably
those created by Little and Mirrlees in the Project Appraisal and Planning for
Developing Countries guide – for appraising transport schemes in low-income
countries.
The dichotomy between the wealth of the industrial nations and the poverty
of Third World countries has resulted in large-scale programs being initiated to
stimulate the economic development of the latter. Much of this aid has been in
the form of monies and resources to improve transport provision. Over 20 percent
of World Bank lending, for example, goes on transport projects, as does about
15 percent of Bank assistance – grants, expertise, etc. Although it has not always
been agreed that aid always stimulates growth (Baur, 1971), or that if it does,
transport investments are the most suitable projects to finance, it is n
­ evertheless
important that within the narrow confines of transport efficiency these monies

BUTTON 9781786435668 PRINT.indd 13 27/04/2022 11:07

 14 TRANSPORT ECONOMICS, 4TH EDITION

are spent wisely. Transport economists have become increasingly involved in the
Third World in transport project appraisal work.
Large private transport businesses often employ economists but more often
they, and smaller enterprises, make use of the many specialist consultancy compa-
nies offering expert advice. The growth in interest in supply-chain logistics, which
spans all movement and storage of raw materials, work-in-process inventory, and
finished goods from point of origin to point of consumption, and within that
just-in-time management, has led to a much more implicit incorporation of eco-
nomics into transport activities.
One way of looking at this is through the notion of ‘value chain’ initiated
by Michael Porter (1985). A value chain is the additional value added as more
inputs are incorporated into the production process for a specific final output. It
describes the full chain of a business’s activities in the creation of a product or
service – from the initial reception of materials through its delivery to market,
and everything in between. Products pass through all activities of the chain in
order and at each activity the product gains some value. The combined interactive
chain of activities gives the products more added value than the sum of the added
values of individual activities. Capturing the value generated along the chain is
the new approach taken by many management strategists. For example, a manu-
facturer might require its parts suppliers to be located near its assembly plant to
minimize the cost of transport or it may require regular delivery of components
to keep production going whilst holding minimum stocks of the component. By
exploiting the upstream and downstream information flowing along the value
chain, the firms seek to bypass intermediaries creating new business models, or in
other ways create improvements in its value system.
Figure 1.2 provides a simplified generic value chain. The key point about it is
the extensive number of linkages required through a production process from the
initial extraction of raw materials to the final delivery of goods to a market and
their subsequent servicing. Transport is, at various levels of aggregation and in
different forms, important at all stages. Just-in-time management is, as we see in
Chapter 9 when we look at the links between transport economics and transport
logistics, important in ensuring that few resources are tied up in the process at any
one time so that inventory holdings are optimized.

Primary activities

Inbound Outbound Marketing and

Operations Services
logistics logistics sales

Support activities
Procurement, human resource management,
infrastructure, technological development

Figure 1.2 The simple notion of the value chain

BUTTON 9781786435668 PRINT.indd 14 27/04/2022 11:07

 TRANSPORT, ECONOMICS, AND ECONOMISTS ­ 15

The remainder of this introduction is concerned with setting the scene for the
body of the book. Initially in these sections, some of the main economic features
of the transport sector are discussed. The intention is, however, not to point to the
uniqueness of transport but rather to highlight the characteristics of the sector
that pose special problems for economists. Recent trends in transport are then
reviewed and commented upon. To keep the subject matter manageable, much
of the focus will be on the American and British situations, although experiences
elsewhere will not be neglected. This is followed by a brief review of what appear
to be some of the longer-term factors that are going to attract the attention of
transport economists. Finally, a detailed contextual section explains the format of
the book and outlines briefly the rationale for the structure adopted.

1.4 The Economic Characteristics of Transport

Possibly the most important characteristic of transport is that it is not normally

wanted in its own right. As Denys Munby (1968) put it half a century ago, ‘[o]nly
the psychologically disturbed or inadequate want transport for its own sake’. In
other words, the desire for most transport is a derived one, but appropriate provi-
sion enables the benefits of a myriad of other, final benefits to be enjoyed. It is a
major facilitator for enhancing personal welfare and for such things as economic
development.
People wish, in general, to travel so that some benefit can be obtained at their
destination. The trip itself is to be as short as possible. Of course, there are ‘joy
riders’ and ‘tourists’, such as those who take cruises, and there are those at the
extreme, like William Sidis, who are peridromophiles. But these people tend to be
in the minority. Similarly, users of freight transport perceive transport as a cost in
their overall production function and seek to minimize it wherever possible. The
derived nature of the demand for transport is often forgotten in everyday debate
but it underlies all economics of transport.
Just as the demand for transport exhibits specific if not unique features,
certain aspects of supply are nearly unique to transport. More specifically, part
of the plant is ‘mobile’ – almost by definition – and is entirely different in its
characteristics to the fixed plant (for example, rail track, airports, etc.). The
fixed component is usually extremely long-lived and expensive to replace. While
most factories in the manufacturing sector may have a physical life expectancy
of 100 years at most, we still use ports and roads constructed in Roman times.
Further, few pieces of transport infrastructure have alternative uses: some former
waterways have been turned into leisure areas and disused railway lines into cycle
routes and footpaths, but these tend to be exceptions.
In contrast, most mobile plant is relatively short-lived and replacement
usually occurs with physical obsolescence rather than technical obsolescence
as with the fixed components. It is also cheap, with the prospect of alternative
employment if demand declines in one market; for example, a bus can be trans-
ferred to another route or another form of service – in technical terms transport

BUTTON 9781786435668 PRINT.indd 15 27/04/2022 11:07

 16 TRANSPORT ECONOMICS, 4TH EDITION

operators have few ‘sunk costs’. Also, unlike fixed plant, the mobile components
of transport are generally subject only to minimal scale economies. (Ships and
aircraft may be exceptions to this in some cases.)
The fixed component, on the other hand, is normally subject to quite sub-
stantial economies of scale. Once a rail track is laid the marginal cost of using
it falls until some maximum capacity is reached. This also generally means that
there is a minimum practical size below which the provision of transport infra-
structure is uneconomical. There are minimum traffic flows, for example, below
which it is not economically practical to build highways or airports.
As Michael Thomson (1974) pointed out, it is these features of the fixed
and mobile components of transport that have influenced the present insti-
tutional arrangements in the sector. The high cost of provision, longevity,
and scale economies associated with the fixed components create tendencies
towards monopoly control, while the ease of entry, flexibility, and lack of scale
effects tend to stimulate competition in the mobile sector. In common with
many other countries, with the notable exception of the United States, official
reaction in Britain to this situation has in the past tended towards the nation-
alization and public ownership of transport infrastructure and the regulation
of competition in the mobile sector. Nations differ in the degree to which fixed
transport assets are publicly owned (there are private railways in some countries
while several European states have privately operated motorways) and in the
types of regulation imposed on mobile factors. The overall impression, however,
is consistent.
While the rationale for the public provision of the fixed components of
transport can be linked to the containment of any monopoly exploitation which
may accompany private ownership (although British and American experiences
in the nineteenth century suggests that control might equally well be enforced
through price regulation), the need to regulate the mobile component stems from
another aspect of transport operations. Transport generates considerable external
effects (the most obvious of which are congestion and pollution); as Thomson
said in the 1970s, it is an engineering industry carried on outside the factory. It is,
therefore, felt to be important to at least contain the harmful effects of transport
and at best to ameliorate them.
Coupled with this is the imperfect knowledge enjoyed by operators and their
inability to foresee relatively short-term change in demand. Regulation is, there-
fore, often justified to ensure that excessive competition at times of depressed
demand does not reduce the capacity of the transport system to an extent that it
cannot meet higher demand during the upturn. This is clearly a matter of serious
concern when a major crisis such as a pandemic affects society.
This also ties in with political-economy arguments that many types of trans-
port should be seen as a service that should meet a ‘need’, in a broad sense, rather
than be supplied under the market forces of supply and demand. Hence, tradi-
tional market forces need to be supplemented with government interventions to
ensure that this wider, social criterion of transport operations is pursued rather
than the simple profit motive. A clear example of this is to be seen in a major

BUTTON 9781786435668 PRINT.indd 16 27/04/2022 11:07

 TRANSPORT, ECONOMICS, AND ECONOMISTS ­ 17

nineteenth-century Supreme Court ruling in the United States, Munn v. Illinois,

when it was ruled that transport is ‘affected with a public interest’.

1.5 This Book

The remainder of this book follows the general pattern set in previous editions.
Economics is boringly static in that sense, and this book is concerned with the
application of economic theory to the transport sector. Unlike other books,
which often concentrate on specific modes of transport such as the railways or
shipping, or specific sectors such as the nationalized transport industries, or have
a geographical bent such as the transport policy of the European Union or urban
transport problems, one of the main aims of this book is to show that many
problems in transport are common to all modes (albeit with minor variations)
and cover many different circumstances and spaces. Consequently, the approach
is to show how economic theory may be applied to improve the overall efficiency
of the transport sector; examples are, therefore, drawn from all forms of transport
and many contexts.
Also, while it is unavoidable, not to say desirable, that official transport
policy must be implicitly incorporated into the analysis, this is not a book explic-
itly about transport policy. Economic forces do not operate in a vacuum but
within the context of laws and governance structures. Indeed, economists of the
public-choice persuasion, such as James Buchanan and Gordon Tullock (1962),
put emphasis on this political-economy dimension. But there are underlying
‘laws’ of economics that transcend circumstances. As said earlier, if prices go up,
people want to buy less of that product.
It is felt useful, on occasions, to give brief details of institutional arrange-
ments since they inevitably influence the relevant type of economic analysis to
apply (for example, a thumbnail sketch of the historical and institutional frame-
work of urban transport planning is included for this purpose), but, again, this
is primarily for contextual reasons. The closing chapter is explicitly concerned
with policy and institutions, and with the matter of the economic regulation of
transport markets.
At the theoretical level, the discussion is couched largely in terms of verbal
and diagrammatic analysis. Mathematical expressions are not shunned, but
equations are included rather as reference points, permitting readers to look up
practical working models should they subsequently wish to undertake their own
empirical investigations. There are virtually no mathematical derivations, but
important equations are ‘talked around’ and the reader, with a few exceptions,
should find no difficulty in following the book even if his/her mathematical educa-
tion has been neglected or forgotten.
Indeed, the philosophy of the approach adopted here is very much akin to
that espoused by the great nineteenth-century economist, Alfred Marshall, in a
letter to a colleague, A.L. Bowley, in 1906:

BUTTON 9781786435668 PRINT.indd 17 27/04/2022 11:07

 18 TRANSPORT ECONOMICS, 4TH EDITION

But I know I had a growing feeling in the later years of my work at the subject that a
good mathematical theorem dealing with economic hypotheses was very unlikely to
be good economics: and I went more and more on the rules – (1) Use mathematics
as a shorthand language, rather than as an engine of inquiry. (2) Keep to them until
you have done. (3) Translate into English. (4) Then illustrate by examples that are
important in real life. (5) Burn the mathematics. (6) If you can’t succeed in (4), burn
(3). This last I did often.

Although several Nobel Prize-winning economists, including Paul Samuelson,

Jan Tinbergen, and Tjalling Koopmans, pointed out that much of econom-
ics deals with quantitative concepts – prices, output, income, and so on – and,
therefore, inherently has a strong mathematic component (indeed, Samuelson
was credited with saying ‘mathematics is the language of economics’), even they
accepted that there are limits to this.
Those interested in the mathematical extensions of the arguments on specific
topics in the book are referred to the references cited. These often set out the under-
lying theory being discussed in more technical terms. The latter are also designed
to direct readers to the more applied work in the field and the numerous challenges
that economists encounter in trying to quantify various aspects of the transport
system. An understanding of basic microeconomic theory is assumed and, while a
knowledge of intermediate economic theory would be helpful, most of the theory
does not go beyond that found in a standard introductory economics text.
For those new to economics, or in need of an aide mémoire to previous
study, there are several introductory microeconomics books to be recommended,
including those by Paul Samuelson and William Nordhaus (2009), Paul Krugman
and Robin Wells (2018), William Baumol et al. (2019), and Gregory Mankiw
(2020). In all cases the authors’ experiences extend well beyond academia to bring
a degree of reality to their writings.
Any divisions of written material are partly arbitrary and really designed to
make things digestible for readers. Chapters forming this book are no exception.
Where to put material on the economics of transport security? Under logistics
(after all much of it is a business problem)? Why not externalities (since part of it
affects the public at large)? Why not transport policy (since there is a large public
policy element involved in delivering secure supply chains)? Decisions have been
made, but inevitably the material could have been arranged in a variety of alterna-
tive ways. The important point is simply to remember that economics is a way of
helping to understand problems.
The book begins in Chapter 2 by looking at the nature and scale of transport
at the beginning of the twenty-first century, and its recent trends. It looks at it
in global, national, and local economies. It moves on in Chapter 3 to look at its
broad interactions with other sectors of the economy, especially the land market.
Such interactions are often neglected in the explicit transport literature but are
central to understanding the role of transport in society. This is followed by a
related chapter concerned with the demand for transport services, the benefits
derived from transport, and the methods of measuring these benefits, drawing
largely upon the tools of welfare economics.

BUTTON 9781786435668 PRINT.indd 18 27/04/2022 11:07

 TRANSPORT, ECONOMICS, AND ECONOMISTS ­ 19

While Chapters 3 and 4 are essentially demand-orientated, the three follow-

ing chapters are primarily concerned with cost and supply aspects (costs here
being seen both in terms of conventional financial cost problems and, also, in
terms of wider external costs). The accountancy costs of running transport are
now recognized by policy-makers as offering too narrow a picture of the overall
social costs associated with the movement of goods and persons. The wider costs
are those that are imposed by transport users on each other, and that reduce the
internal efficiency of the transport system, and the environmental costs that are
inflicted on wider society.
The remaining chapters concentrate on optimizing the size and use made
of the transport sector both in the short term and in the longer term. Methods
of pricing by transport-supplying companies are reviewed in the context of the
nature of their transport activities and markets being served, with, once again,
emphasis being directed to the wider social dimension as well as to narrow com-
mercial criteria. Chapter 8 looks at methods of optimizing the environmental
effects of transport on society in general, and Chapter 9 focuses on the challenges
of dealing with the problems of traffic congestion, not only on our streets but also
at airports and seaports. The importance of transport in the supply chain and
modern logistics management is reviewed in Chapter 10.
Longer-term planning and investment decisions are considered in some
detail in Chapter 11. This is because of both the complexities of the issues
involved and the scale of resources wastage if wrong decisions are made. The
sheer costs involved can be enormous, and particularly so where major pieces of
new infrastructure are under consideration – for example, an international airport
or a Channel Tunnel. The inconvenience and disruption of construction to the
existing transport networks is also often significant. The problems of financing
and funding such undertakings are becoming more challenging and the respective
roles and interactions of the government and private sectors are considered.
The long physical working life of many transport projects, let alone the lead
time often needed before they become operational, means that it is important to
be able to forecast the future demands likely to be placed on infrastructure, and
consequently, in Chapter 12, space is devoted to economic demand forecasting
techniques. The emphasis is, once again, focused on the economic assumptions
involved rather than the econometric and estimation problems that may be
encountered, although the latter are not entirely neglected.
The final chapters of the book take a much broader view of transport, con-
sidering in Chapter 13 the role that transport policy may play in general economic
development both in the wealthier nations and in less developed countries. The
question of appropriate regulations and the factors that have influenced transport
policy over the years is explicitly examined in Chapter 14. The discussion is con-
cerned with the relative merits of central coordination and direction of transport
vis-à-vis the use of market mechanisms and competitive processes.
Each chapter is referenced to permit readers to follow up specific points in
more detail should they wish. These references are light and are not intended to
cover all the literature. They are directed to help students and newcomers to the

BUTTON 9781786435668 PRINT.indd 19 27/04/2022 11:07

 20 TRANSPORT ECONOMICS, 4TH EDITION

field. The aim is that they should be references to material that is relatively acces-
sible to most readers, rather than obscure reports, working papers, etc., which are
often extremely difficult for those outside official circles or the academic sphere to
obtain. Additionally, many of the references are to classic papers that can easily
be found electronically. These are mainly linked to key ideas and concepts. There
is also quite a lot of ‘name-dropping’. For some people, including me, authors or
other important figures act as ‘hangers’ to suspend their ideas from.
Experience suggests that students often feel in the information age that any
article or book written more than five years or so ago is outdated. This is, of
course, a folly of inexperience. Good studies that reveal universal principles are
‘good studies’ whatever their vintage. The lists of key references are also kept
short so that the main items are immediately apparent to those interested. Many
are to collections or surveys that go into greater depth on topics than we can in
this short volume.

References

Alexander, K.J.W. (1975) Some economic problems of the transport industry,

Chartered Institute of Transport Journal, 36, 306–308.
Baumol, W.J., Binder, A., and Solow, J. (2019) Macroeconomics: Principles and Policy, 14th
edition, Centage Learning.
Baur, P. (1971) Dissent on Development, Weidenfeld and Nicolson.
Beckman, M., McGuire, C.B., and Winsten, C.B. (1956) Studies in the Economics of
Transportation, Yale University Press.
Buchanan, J.M. and Tullock, G. (1962) The Calculus of Consent: Logical Foundations of
Constitutional Democracy, University of Michigan Press.
Button, K.J. (2020) The ‘Ubernomics’ of ridesourcing: the myths and the reality, Transport
Reviews, 40, 1–19.
Button K.J. and Swann, D. (eds) (1989) The Age of Regulatory Reform, Oxford University
Press.
Coase, R.H. (1992) The institutional structure of production, American Economic Review,
82, 713–19.
Gwilliam, K.J. (1980) Review of ‘Transport Economics’, Economic Journal, 90, 677–8.
Krugman, P. and Wells, R. (2018) Microeconomics, 5th edition, Worth.
Mankiw, G.N. (2020) Principles of Microeconomics, 9th edition, Centage Learning.
Meyer, J. and Gómez-Ibáñez, J. (1983) Autos, Transit and Cities, Harvard University Press.
Meyer, J.R., Peck, M.J., Stenason, J., and Zwick, C. (1959) The Economics of Competition
in the Transportation Industries, Harvard University Press.
Milne, D. and Watling, D. (2019) Big data and understanding change in the context of
planning transport systems, Journal of Transport Geography, 76, 235–44.
Munby, D. (1968) Transport, Penguin.
O’Connor, A.M. (1965) Railways and Development in Uganda, Oxford University Press for
the East African Institute of Social Research.
Porter, M.E. (1985) Competitive Advantage: Creating and Sustaining Superior Performance,
Free Press.
Rakowski, J.P. (1976) Transport Economics: A Guide to Information Sources, Gale Research.
Samuelson, P.A. and Nordhaus, W.D. (2009) Microeconomics, 19th edition, McGraw-Hill.
Shiller, R.J. (2017) Narrative economics, American Economic Review, 107, 967–1007.

BUTTON 9781786435668 PRINT.indd 20 27/04/2022 11:07

 TRANSPORT, ECONOMICS, AND ECONOMISTS ­ 21

Thomson, M. (1974) Modern Transport Economics, Penguin.

Williamson, O. (2000) The new institutional economics: taking stock, looking ahead,
Journal of Economics Literature, 38, 595–613.
Winston, C. (1985) Conceptual developments in the economics of transportation: an inter-
pretive survey, Journal of Economic Literature, 23, 57–94.
Winston, C. (2013) On the performance of the U.S. transportation system: caution ahead,
Journal of Economic Literature, 51, 773–82.

BUTTON 9781786435668 PRINT.indd 21 27/04/2022 11:07

 2 Transport and the Economy: Some
Numbers

2.1 Introduction

Transport is important; some have described it as the blood of an economy.

It is, in both physical and financial terms, a major global economic sector, an
employer of large numbers of people, the consumer of vast quantities of raw
materials, and it takes up a lot of personal time in its use. Whether it is now
relatively more important than it was in the past is difficult to say. Certainly,
the major civilizations of yesteryear put considerable emphasis on supporting
their transport systems, and many, such as the Phoenicians, the Romans, the
Chinese, and the Dutch, centered their economies around their quite distinctly
different transport systems. The Chinese, for instance, have always been essen-
tially a land power, whereas the Dutch and the Phoenicians were maritime
traders.
To understand just why there is such considerable interest in transport and
its efficiency, it would initially be useful to look at a few facts and figures on its
role globally, regionally, and more locally. The descriptions here are far from com-
plete. Many of the implications and effects of transport lie outside of the sector,
including environmental effects and land-use interactions, and we shall deal with
these more explicitly in separate chapters.
The present chapter is largely about the quantifiable, although it does
provide some qualitative information regarding transport. But it is always helpful
to remember words supposedly written by Albert Einstein on a blackboard: ‘Not
everything that counts can be counted, and not everything that can be counted
counts’. Many aspects of transport affecting how, when, and why people travel,
or why cargo consignees choose a specific shipper, are not easily quantified. Some
of these factors are not easily encapsulated in some convenient metric, while in
other cases they are subjective. Nevertheless, they may be highly significant when
looking at how the transport system functions. This has been gradually recog-
nized by transport economists who are showing increasing interest in behavioral
economics.
In addition, we should also hold in the backs of our minds the words of the
nineteenth-century British Prime Minister and novelist, Benjamin Disraeli, who
allegedly cautioned that ‘[t]here are three kinds of lies: lies, damned lies, and sta-
tistics’. Basically, he was referring to the persuasive power of numbers, and the use
of statistics to support weak arguments, and the tendency of people to disparage
statistics that do not support their positions. We shall attempt to be as objective

BUTTON 9781786435668 PRINT.indd 22 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 23

as possible in setting out some of the more insightful transport data. Some of the
more qualitative and subjective aspects are considered in later discussions.
But even so, it is impossible to paint anything like a complete quantitative
picture of the nature and importance of modern transport. One could approach the
challenge by focusing on a single country or industry – say shipping or railroads –
but this would inevitably be partial. Also looking at any single country’s transport
system can make it appear as if that country is special or unique. Certainly, in some
ways any data will inevitably be context-specific, but there are broader trends and
features that transcend physical boundaries, and many of the economic challenges
that confront policy-makers are almost global in their i­ ncidence – traffic congestion
being the most obvious case. Nevertheless, the examples offered in this chapter are
fairly wide-ranging to illustrate more general patterns and features.
We begin by looking at the very large pictures, international trade in trans-
port services, and then gradually move down the levels of aggregation, with a
diversion at the end to consider some emerging trends and trend breaks. Because
local transport parameters can vary quite considerably, and because we deal with
many specific local matters in subsequent chapters, this part of the description is
less complete. There is also a bias in that there is a disproportionate amount of
American data and information provided, although this is supplemented in places
with data from the United Kingdom and other countries.
In addition, it is important to remember that there can be quite large margins
of error in transport data. Many statistics used are based on surveys that have
inevitable statistical confidence intervals associated with them. Many surveys
are also spasmodic and are not done often or at regular intervals. Questions
asked often differ over time and there are issues of definitions when making
comparisons across countries (for example, of high-speed rail services which
are defined differently by the European Union and the International Union of
Railways). These problems persist despite the efforts of the Organisation for
Economic Co-operation and Development (OECD), Eurostat, the World Trade
Organization (WTO), and other bodies to achieve greater standardization.
Much transport is provided by the private sector (including, in the case of
car drivers, inputs of the travelers themselves), and the type and quality of such
of data can be significantly different from that gathered by public agencies. Most
of the private sector is commercially oriented and thus focuses on data that help
in better business decision-making, rather than that needed for good public policy
analysis. This often means, for example, that it is shorter-term. Commercial
activities are largely interested in time frames involving cost recovery, and data
are seldom gathered consistently, because private companies are generally con-
cerned with projects of immediate interest to them. Additionally, most private
firms operate in markets where a need for a degree of commercial confidentiality
engenders a reluctance to reveal data.
Even when census data are used, this type of information is normally only
collected every decade with extrapolations or samples used to fill in the interven-
ing years. Added to this, most data that are collected on transport reflect physical
characteristics of systems, and this is not always ideal for economic analysis. This

BUTTON 9781786435668 PRINT.indd 23 27/04/2022 11:07

 24 TRANSPORT ECONOMICS, 4TH EDITION

latter problem has become more acute in recent years as transport industries have
become less regulated and there have been privatizations. With these institutional
changes, commercial considerations have mitigated incentives to provide public
data and financial information.

2.2 The Global Picture

International trade has existed since the formation of nation states, and indeed
was the raison d’être for the establishment of many. Classical economists,
however, when trying to explain the intensity and patterns of international trade,
paid little attention to the role of transport – it was largely ‘assumed away’. Their
focus was on the comparative advantages of the nations producing the commodi-
ties being traded. Recently the emergence of the ‘new trade theory’ (NTT), often
associated with Paul Krugman (2009), has changed things. The NTT recognizes
that a large part of trade is intra-industrial, involving countries both importing
and exporting very similar goods (for example, Germany both imports cars from
and exports cars to Japan). Transport costs can form an important, if seldom the
main, element in explaining some of the current patterns of trade.
According to the WTO, global recorded merchandise trade in 2018 amounted
to $19.67 trillion, and no doubt there was much more that fell out of the official
statistics. While much of this involved the more traditional industrial regions
of Europe (with Germany accounting for $1.56 trillion of the region’s exports
and $1.29 trillion of its imports), the United States ($1.66 trillion of its exports
and $2.61 trillion of its imports), and Japan ($0.74 trillion of its exports and
­$0.75 ­trillion of its imports), the new mega-economies of China (with $2.49
­trillion of its exports and $2.14 trillion of its imports) and India are also now
major players. Table 2.1 offers details in terms of merchandise imports.
Both aggregate spatial and temporal patterns in world trade in merchandise
are highly correlated with patterns and trends in national gross domestic product
(GDP) – the production taking place within a country’s borders. Figure 2.1
provides a graph of recent trends. Clear ups and downs are seen in the world
economy as it has experienced trade cycles. Nevertheless, the correlation, if not
the causality between GDP and merchandise exports, is clear.
While these financial trends are useful, and germane to many economic
debates, they do not offer comprehensive insights into the physical nature of
transport demands or the quantity of resources that are being used to meet it.
For example, while international shipping carries some 90 percent of global trade
measured by weight, air transport accounts for 35 percent of that moved by value
despite only carrying 1 percent of the tonnage. The value of what is moved is
often unrelated to its physical magnitude, and, in this case, air transport tends to
move high-value, low-volume commodities where speed and reliability of delivery
are at a premium. Table 2.2 provides some more general information concerning
the growth of international air transport in the movement of people and freight.
Again, a broad correlation is seen.

BUTTON 9781786435668 PRINT.indd 24 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 25

Table 2.1 
World merchandise imports by region and selected countries ($ billions at
constant prices)

1983 1993 2003 2018

Value world ($ billions) 1,833 3,805 7,694 19,394

North America 18.5 21.3 22.4 18.4
United States 14.3 15.9 16.9 13.5
Mexico 0.7 1.8 2.3 2.5
Canada 3.4 3.7 3.2 2.4
South & Central America
& the Caribbean 3.9 3.3 2.5 3.3
Brazil 0.9 0.7 0.7 1.0
Chile 0.2 0.3 0.3 0.4
Europe 44.1 44.5 45.0 36.9
Germany 8.1 9.0 7.9 6.6
United Kingdom 5.3 5.5 5.2 3.5
France 5.6 5.7 5.2 3.5
Netherlands 3.3 3.3 3.4 3.3
Commonwealth of
Independent States – 1.5 1.7 2.2
Africa 4.6 2.6 2.2 3.0
South Africa 0.8 0.5 0.5 0.6
Middle East 6.2 3.3 2.8 3.8
Asia 18.5 23.5 23.5 32.4
China 1.1 2.7 5.4 11.0
Japan 6.7 6.4 5.0 3.9
India 0.7 0.6 0.9 2.6
Australia & New Zealand 1.4 1.5 1.4 1.4

Source: World Trade Organization.

Transport supply has grown to cater for the demands of the internationalization
of supply chains and globalization. Since the development of unitization in the
1960s, container ships and trains have been a core input into the international
movement of finished goods and components. For example, in 2019 there were
5,150 large container vessels engaged in that market. Added to this were 11,562
large bulk carriers and 7,444 crude-oil-tankers. Because of the finances needed
to invest and operate modern shipping fleets, the emergence of large shipping
companies has accompanied this. Table 2.3 details the major operators together
with their relative market shares. Four companies dominate the market in con-
tainer services, although that does not mean they operate the largest vessels; for
example, ships like HMM’s Copenhagen, Le Havre, and Gdansk can carry up to
23,964 TEUs.1

1 The TEU is an inexact unit of cargo capacity used to describe the capacity of container
ships and terminals. It is based on the volume of a 20-foot-long shipping container,
a standard-sized metal box that can be easily transferred between different modes of
transport, such as ships, trains, and trucks.

BUTTON 9781786435668 PRINT.indd 25 27/04/2022 11:07

 26 TRANSPORT ECONOMICS, 4TH EDITION

6 Merchandise exports GDP

Annual % change 4

0
2011 2012 2013 2014 2015 2016 2017 2018
Year
Source: World Trade Organization.

Figure 2.1 The volume of world merchandise trade and GDP (2011–18)

Table 2.2 International and global revenue air traffic

Passengers Passenger-km Freight tonnesa Freight tonne-kma

(millions) (millions) (millions) (millions)

2009 2,488 4,561,413 40.0 155,819

2010 2,705 4,924,229 47.6 186,631
2011 2,870 5,248,140 48.7 187,191
2012 3,004 5,528,880 48.0 185,239
2013 3,138 5,832,564 49.1 185,975
2014 3,316 6,181,177 50.7 194,633
2015 3,556 6,644,666 51.0 197,131
2016 3,794 7,135,773 52.8 204,187
2017 4,062 7,707,118 56.6 222,996
2018 4,322 8,257,635 58.0 230,967

Note: a. Excludes mail.

Source: International Civil Aviation Organization.

Unlike shipping, the freight operations of air transport overlap quite consid-
erably with passenger services. To meet the demand for air cargo services, for
example, there were 1,870 dedicated freighter aircraft in 2018, representing about
42 percent of the 700 billion available tonne-kilometer air freight capacity, the
remaining 58 percent being moved in the belly holds of passenger aircraft. But,
just like shipping, both the supplying companies – passenger and freight – of
international air services are huge undertakings. Although size can be measured
in several ways, in terms of fleets for example, American Airlines in 2019 had
957 planes, Delta Air Lines had 880, and United Airlines had 777. But, as with

BUTTON 9781786435668 PRINT.indd 26 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­27

Table 2.3 
Leading container service operators (2020), based on the number of ships
and TEUs

Company Headquarters Ships Total TEUs Market share (%)

APM–Maersk Denmark 676 4,051,436 16.90

Mediterranean Shipping Switzerland 574 3,815,410 16.00
COSCO Group China 488 2,978,334 12.50
CMA CGM Group France 532 2,824,214 11.80
Hapag–Lloyd Germany 241 1,726,578 7.20
ONE Japan 211 1,548,426 6.50
Evergreen Marine Taiwan 196 1,268,361 5.30
HMM South Korea 69 651,884 2.70
Yang Ming Marine Transport Co. Taiwan 91 608,072 2.50
PIL Singapore 107 342,405 1.40
Zim Integrated Shipping Services Israel 64 305,343 1.30
Wan Hai Lines Taiwan 105 285,070 1.20
Zhonggu Logistics Corp. China 115 168,581 0.70
Republic of Korea Marine South Korea 68 162,276 0.70
Transport Co.
IRISL Group Iran 48 152,419 0.60

Source: Alphaliner.

shipping lines, the importance of single fleets of ships or planes does not reflect
the market power they can exercise by combining their capacity and schedules.
Something returned to later.
Transport requires infrastructure as well as vessels, planes, and vehicles,
to move goods and people around the globe. This infrastructure is generally
­physically large, and inevitably expensive and immobile. It involves investments
that seldom have uses beyond those for which they are designed – in economic
terms, most infrastructure is a ‘sunk cost’. This is a problem we see later when
looking at technology improvements, or new regulations, which result in existing
infrastructure becoming economically obsolete well before it is physically worn
out.
To give an impression of the scale of some of the transport infrastructure
in the world, Table 2.4 lists the largest seaports in terms of what they handled,
together with changes in their rankings, between 2015 and 2018. The increased
importance of China is seen in terms of the number of its ports and their move-
ment up the table over time. Table 2.5 offers broadly comparable data on inter-
national passenger airports, in this case measured by the number of international
passengers they handled.
It is more difficult to isolate international road and rail infrastructure,
although land modes play a significant role in international trade not only as
the trunk-haul carrier across land borders but also in providing feeder services
to sea and air movements. Because the infrastructure of land modes is gener-
ally extensively shared with domestic transport, separating out the international
component is not easy. Indicators of the scale of selected national major road
networks, which are used for both domestic and international movements, are

BUTTON 9781786435668 PRINT.indd 27 27/04/2022 11:07

 28 TRANSPORT ECONOMICS, 4TH EDITION

Table 2.4 
The world’s largest container seaports by million TEUs (ordered by size, 2018)

Port Country 2015 2016 2017 2018

Shanghai China 36.54 37.13 40.23 42.01

Singapore Singapore 30.92 30.90 33.67 36.60
Shenzhen China 24.20 23.97 25.21 27.74
Ningbo-Zhoushan China 20.63 21.60 24.61 26.35
Guangzhou Harbor China 17.22 18.85 20.37 21.87
South Korea South Korea 19.45 19.85 20.49 21.66
Hong Kong Hong Kong 20.07 19.81 20.76 19.60
Qingdao China 17.47 18.01 18.30 18.26
Tianjin China 14.11 14.49 15.07 16.00
Jebel Ali Dubai 15.60 15.73 15.37 14.95
Rotterdam Netherlands 12.23 12.38 13.73 14.51
Port Klang Malaysia 11.89 13.20 13.73 12.32
Antwerp Belgium 9.65 10.04 10.45 11.10
Kaohsiung Taiwan 10.26 10.46 10.27 10.45
Xiamen China 9.18 9.61 10.38 10.13

Table 2.5 Largest airports by international passenger numbers (2019)

Airport Location Passengers

Dubai International Airport Dubai 8,888,536,700

London Heathrow Airport United Kingdom 7,530,693,900
Hong Kong International Airport Hong Kong 7,436,097,600
Amsterdam Airport Schiphol Netherlands 7,095,625,800
Incheon International Airport South Korea 6,767,614,700
Charles de Gaulle International Airport France 6,638,349,400
Singapore Changi Airport Singapore 6,489,000,000
Frankfurt Airport Germany 6,177,466,300
Suvarnabhumi Airport Thailand 5,086,884,600
Ataturk International Airport Turkey 4,897,877,000
Taiwan Taoyuan International Airport Taiwan 4,615,216,400
Kuala Lumpur International Airport Malaysia 4,353,174,100
Madrid Barajas International Airport Spain 4,185,712,500
London Gatwick Airport United Kingdom 4,147,685,800
Munich Airport Germany 3,654,578,700
Barcelona–El Prat International Airport Spain 3,654,578,700
Haneda International Airport Japan 3,530,007,600
Hamad International Airport Qatar 3,449,507,800
John F. Kennedy International Airport United States 3,348,507,800
Lester Pearson International Airport Canada 3,161,034,800

Note: There are larger airports that primarily handle domestic traffic – for example, Hartsfield–
Jackson Atlanta International Airport handled 110.53 million passengers in 2019 and Beijing
Capital International Airport some 110.01 million.

Source: Airports Council International.

BUTTON 9781786435668 PRINT.indd 28 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 29

Table 2.6 Motorway (freeway) networks of higher-income countries (2018)

Country Total length 1,000 km Kilometers per 10,000 km2

United States 263,932 268

Canada 17,000 17
Mexico 15,044 77
Spain 14,701 291
Germany 12,917 362
France 11,465 209
Italy 6,726 223
Korea 4,044 404
United Kingdom 3,686 151
Portugal 2,988 324
Netherlands 2,646 637
Turkey 2,155 28
Sweden 1,927 43
Belgium 1,763 577
Austria 1,719 205
Switzerland 1,419 344
Denmark 1,128 262
Finland 810 24
Slovenia 770 380
New Zealand 183 7

Source: Organisation of Economic Co-operation and Development.

seen in Table 2.6. The final column puts these data in the context of the differ-
ing sizes of countries. The road orientation of American society is well known,
but only in absolute terms. Countries such as Korea, Germany, Belgium, and the
Netherlands have more kilometers of road per unit of area.
A similar set of statistics for railroads, again listed in order of size, is seen in
Table 2.7. Unlike roads that are almost universally used for the movement of both
freight and people, railway networks are often more specialized. For example,
the Canadian and the United States system, excepting Amtrak and some local
services around larger cities, is primarily a freight system, whereas that of the
United Kingdom, and many geographically smaller countries, are used mainly
by passengers. Russia has a more mixed system. High-speed rail lines are almost
entirely passenger systems. The concept of infrastructure and the way in which
it is measured can thus be contextual, whether one is looking at its importance
to the freight or the passenger sectors. As we see in Chapter 5, this can pose
problems when trying to allocate costs of infrastructure to various transport user
groups, and setting the fares and rates for using them.
Another and linked difficulty with this type of data is that of devising a
common unit of measurement – for example, freight moved from an airport is
measured in tons and containers from seaports in TEUs, whereas passengers are
measured in terms of individuals. When studying something like fuel use, it is rea-
sonable to reduce everything to tons carried. There is a high correlation between
the two. But this is hardly appropriate when allocating the full costs of a flight

BUTTON 9781786435668 PRINT.indd 29 27/04/2022 11:07

 30 TRANSPORT ECONOMICS, 4TH EDITION

Table 2.7 World’s largest railway networks by kilometers

Country Kilometers Year

United States 149,407 2017

China 146,300 2020
India 99,235 2015
Russia 85,600 2019
Canada 64,000 2017
Germany 40,626 2017
Argentina 36,966 2014
Australia 33,168 2017
Brazil 29,817 2014
France 29,273 2017
Japan 27,311 2015
Mexico 23,389 2020
South Africa 22,387 2017
Ukraine 20,952 2016
Poland 19,209 2017
Iran 16,998 2014
Italy 16,798 2016
Spain 16,355 2017
United Kingdom 16,320 2017
Kazakhstan 15,530 2018
Sweden 14,180 2020
Turkey 12,740 2010

on which goods and people are carried. The on-board facilities needed for people
and cargo movement are entirely different.
Time-series comparisons can also be difficult. Output measures (for example,
ton-miles or passenger-kilometers), although appearing to circumvent problems
of comparability over broad categories of transport, in fact tend to be inad-
equate. Such measures ignore changes in the quality and costs of alternatives. For
example, flying with a low-cost airline today is very different to flying with the
full-service airlines of the 1970s. It is possible to count physical units (for example,
the number of cars or planes), but lack of homogeneity again prevents meaning-
ful, in-depth comparisons within individual modes. For example, an average car
in the early 2000s was very different from the average car today. Despite these
comments it is still possible to look at official statistics and obtain an overall feel
for the transport situation evolving in a very general sense.
While international trade, and transport with it, has grown globally, trade
between and within certain blocs of countries and regional trade areas (RTAs)
has tended to expand even more rapidly because of their new institutional struc-
tures. Most notable has been the creation of the North American Free Trade
Agreement (NAFTA), restructured since 2020 as the United States–Mexico–
Canada Agreement (USMCA), and the continued widening and deepening of
the European Union. (The latter is discussed in detail in Chapter 14.) Figure 2.2
shows just how important these trade blocs have become as markets in themselves.

BUTTON 9781786435668 PRINT.indd 30 27/04/2022 11:07

 31
TRANSPORT AND THE ECONOMY: SOME NUMBERS ­

Intra-RTA trade Intra-regional trade Rest of the world

100
90
80
70
60
50
40
30
20
10
0
A

A
EU

C
R

A
FT

A
SU

ES
A

CA

EM

W
SA

M
SE
A

M
CO

O
A

CE
N

CO
W

EC
ER
M

Source: World Trade Organization, World Trade Statistical Review, 2019; WTO, 2020.

Figure 2.2 Value of trade involving regional trading blocs (2017)

These relatively new institutional arrangements have both fostered greater

demand for transport services as barriers to trade in goods and services as trade
were diminished and made it easier to provide transport as cross-border regula-
tions were relaxed. In some cases, as with the European Union, they also allow
cabotage – the ability of a foreign transport operator to pick up and set down
in another country – which increases transport flexibility and efficiency further
(Button, 2011). For example, the creation of the European Single Internal Market
in 1992 has been estimated to have led to a 4.5 to 7 percent rise in the European
Union’s gross national product (GNP), with an increase of 30 to 50 percent in
trans-frontier lorry traffic.
As another example of open borders, the NAFTA, was a trilateral trade bloc
in North America created by the governments of the United States, Canada, and
Mexico, and was, in terms of the combined purchasing power and GDP of its
members, the largest trade bloc in world and the second-largest by nominal GDP
comparisons. From its creation in 1994, NAFTA provided a stimulus to trade
between its member states, including that between Mexico and Canada transit-
ing the United States. Table 2.8 provides some details of increases in the value of
trade from the inception of the NAFTA free trade agreement until 2015.

2.3 Transport at the National Level

Transport forms a major component of the national output as well as accounting

for a large part of national expenditure in most developed countries. For example,

BUTTON 9781786435668 PRINT.indd 31 27/04/2022 11:07

 32 TRANSPORT ECONOMICS, 4TH EDITION

Table 2.8 NAFTA trade volumes by value ($ millions)

1993 2015 Real increase (%)a

United States–Canada 199.18 518.22 63.50

United States–Mexico 85.22 481.54 255.00
Mexico–Canada 4.05 34.34 432.50
Tri-lateral 288.46 1,034.10 125.20

Source: a. US Bureau of Labor Statistics, Consumer Price Index.

in 2014 expenditure on transport represented 18.7 percent of total national

expenditure in the United States. In the United Kingdom it was 16.2 percent, and
in Japan, 10.0 percent. A substantial part of this, as seen for the United States in
Table 2.9, is taken up by final domestic demand, with private consumption being
the dominant element of this.
It is also clear that within the United States and many other countries there
is a strong reliance on the automobile in physical as well as financial terms. Air
transport, however, is also important in the United States compared to other
countries, mainly because of the high-income levels and the distances traveled
in North America. These features are reflected in data offered in Table 2.10
for the United States and in Table 2.11 for England, although in the latter it is
expressed in terms of the travel patterns of an average family rather than as an
aggregate. The English data are included here because there is more complete

Table 2.9 Final demand for United States transport ($ billions in 2009 prices)

2000 2008 2015

Gross domestic product 12,565 14,834 16,397

Personal consumption of transport 945 925 1,021
Motor vehicles & parts 346 341   419
Motor vehicle fuels, lubricants, & fluids 266 262   267
Transport services 333 321   298
Gross private domestic investment 213 167   335
Transport structures 9 10    12
Transport equipment 204 157   302
Exports (+) 218 269   321
Imports (–) 328 355   467
Government transport-related purchases 283 290   283
Federal purchases 25 38   n.a.
State & local purchases 246 233   n.a.
Defense-related purchases 12 20   n.a.
{Percent of GDP spent on transport}a 10.60% 8.70% 7.29%

Notes: a. Aggregate transport expenditure adjusted for the omission of some categories of
transport expenditures and rounding.
n.a. = not available.

Source: US Bureau of Transportation Statistics.

BUTTON 9781786435668 PRINT.indd 32 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 33

Table 2.10 United States passenger transport by mode (million passenger-kilometers)

2005 2010 2015 2018

Air 921,943 892,720 1,017,355 1,162,615

Highway 7,833,657 8,062,501 8,504,252 8,928,457
Transit 75,840 84,695 89,637 86,631
Inter-city/Amtrak 8,660 10,332 10,519 10,237
Walking      n.a.      n.a.      n.a.      n.a.
Cycling      n.a.      n.a.      n.a.      n.a.

Note: Passenger-kilometers in the number of trips taken by each mode multiplied by the
average length of those trips.
n.a. = not available.

Table 2.11 Average number of household trips by main mode in England

2005 2010 2015 2019

Private:
Walk 272 234 219 250
Walks of over a mile 74 65 68 65
Bicycle 15 15 17 16
Car/van driver 434 402 381 380
Car/van passenger 234 212 204 200
Motorcycle 4 3 3 2
Other private transport 8 7 7 7
Public:
Bus in London 19 25 20 18
Other local bus 43 42 41 32
Non-local bus 1 1 1 –
London Underground 9 9 9 12
Surface Rail 16 19 20 21
Taxi-/mini-cab 11 9 10 11
Other public transport 3 2 3 3

information on cycling and walking than in the United States. We also see a dis-
cernably greater use of public modes in the United Kingdom than in the United
States, a pattern for example repeated in most Continental European countries.
More recent trends have also been for the car to continue to increase its modal
share in most countries over time, although at a markedly slower rate in recent
years for many of the highest-income nations as vehicle ownership approaches
saturation – basically a situation where everyone able and permitted to drive has
access to a car.
As shown later, the rise in private transport use is closely related to higher
car ownership levels – although the question of cause and effect is a complex one
(Chapter 4). What should be remembered, however, is that while the relative use
of private transport is rising, the role of public transport, especially for commuter
trips into large cities, is still very important when discussing transport, and is
often rising in absolute terms and for specific journey purposes as urbanization

BUTTON 9781786435668 PRINT.indd 33 27/04/2022 11:07

 34 TRANSPORT ECONOMICS, 4TH EDITION

increases. Air transport increasingly dominates longer trips even outside of the
United States and, in particular, trips involving tourism.
What the tables do not show is the share of household expenditure by the
different income groups that is spent on transport. What is found from this break-
down of expenditure data is that the proportion of outlays devoted to transport
tends to rise with household income reflecting the ‘superior good’ nature of the
activity. What official data also show is the use made by different income groups
of the various modes. In the United Kingdom, for example, railways are used
primarily by those in the higher-income groups, as is the private car. In contrast,
bus transport is used disproportionately more by the poorer sectors of the com-
munity. We shall return to this when looking at urban transport trends.
Within countries there are also distinctions in terms of how freight is trans-
ported, and the trends these movements have been taking over time. As seen in
Table 2.12, in terms of ton-miles of inland freight transport, trucking has the
largest share in the United States although other modes, and especially railroads,
have been gaining in importance. This is a pattern, however, that is not reflected
in all higher-income nations. As shown in Table 2.13, the wider picture generally
shows a steady decline in the relative, but not always absolute, role of freight
railroads.
Table 2.12 also provides information on pipeline movements in the United
States, an important but often neglected form of freight transport. Physical limits
to the type of commodities which can be carried in this way – basically liquids,
including sewage and slurry, and gas – are likely to prevent pipelines from ever
becoming more than a minority mode of transport. Indeed, if society moves to a
carbon neutral economy, then some of their roles will diminish as the importance
of oil as an energy source declines.
This contrasts with the situation regarding trucking where the trend has been
clearly upwards. The explanation for this lies partly in the increase in unitization
of the goods shipped coupled with an uptake of just-in-time production and dis-
tribution. This has resulted in firms holding lower inventories, a subject dealt with

Table 2.12 United States ton-miles of freight (millions)

2000 2005 2015 2018

Total ton-miles 5,065,648 5,379,424 5,110,527 5,250,670

Air 14,983 15,746 13,190 15,969
Truck 1,971,087 2,210,106 1,985,827 2,033,921
Railroad 1,465,960 1,696,425 1,738,283 1,729,638
Domestic water transport 645,799 591,447 490,627 491,800
Coastwise 283,872 263,580 175,604 173,600
Lakewise 57,879 51,976 46,436 46,900
Internal 302,558 274,366 267,447 270,200
Intraport 1,490 1,525 1,139 1,100
Pipeline 967,819 865,700 882,600 979,343

Source: US Bureau of Transportation Statistics.

BUTTON 9781786435668 PRINT.indd 34 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 35

Table 2.13 
Freight moved by mode by country (2000 and 2017, billion tonne-kilometers)

Road Rail Coastal shipping

2000 2017 2000 2017 2000 2017

United Kingdom 153.7 156.1 181.2 172.0 66.2 23.0

Austria 17.2 18.4 16.6 22.3 – –
France 184.4 162.6 53.4 40.7 10.5 –
Germany 280.7 313.1 77.5 129.3 – –
Italy 158.6 119.7 25.8 22.3 33.4 64.9
Spain 148.7 231.1 12.2 10.5 37.8 49.7
Sweden 31.4 40.7 20.1 21.8 8.1 6.5
Czech Republic 39.0 44.3 17.5 15.8 – –
Poland 54.4 111.8 54.5 54.8 – –
Switzerland 13.6 17.2 11.1 11.7 – –
China 612.9 6,677.2 1,377.1 2,696.2 17.1 5,508.4
Japan 313.1 210.8 22.2 22.7 241.7 180.0
United States 3,396.7 2,954.2 2,257.6 2,523.4 414.4 250.7a

Note: a. 2016.

Source: Organisation for Economic Co-operation and Development.

in some depth in Chapter 10. To ensure that their stocks are adequate this means
more deliveries with inventories essentially being held in the transport system
rather than warehouses. Although rail and air transport have important roles in
this business model, trucks combine the flexibility with the cost structure required
for maximum efficiency over short and medium distances.
The public sector plays an important role in the direct provision of national
transport services in most countries, although the scale and nature of that role
differs. Table 2.14 shows the importance in a federal regime such as the United
States of state as well as central government expenditures. It has been rising over
time with respect to all modes. There are some caveats to be considered, however,
when using these data. The figures likely hide some transport outlays, for example
regarding the military as well as transport-related expenditures embodied in the
accounts for such items as law and order, environmental services, and health func-
tions of government. Additionally, public expenditure is subjected to cycles that
reflect the state of a national economy and macroeconomic government policies.
For example, following the attacks of September 11, 2001, the United States
considerably increased its direct expenditure on transport security, and after the
Covid-19 pandemic there began a phase of infrastructure expansion and renewal.
An added problem in looking at this type of aggrege data in large, and especially
federal, countries is that they can miss outlays at lower levels of government, for
example by local or municipal authorities.
Transport is a major employer in most countries, for example Indian
Railways’ 1.254 million employees in 2020 make it the world’s eighth-largest
employer. Considerable inputs of labor extends through the transport infra-
structure investment phase, construction of vehicles through to operations and

BUTTON 9781786435668 PRINT.indd 35 27/04/2022 11:07

 36 TRANSPORT ECONOMICS, 4TH EDITION

Table 2.14 Government expenditure on transport in the United States (current $ millions)

2007 2011 2014 2017

Highways 174,808 193,649 207,118 223,429

Federal 36,981 45,536 47,290 47,170
State & local 137,827 148,113 159,827 176,258
Air 34,107 39,200 38,576 43,004
Federal 17,895 19,610 20,022 20,165
State & local 16,212 19,590 18,554 22,839
Transit 46,183 53,042 59,699 66,861
Federal 9,199 9,620 11,520 11,728
State & local 36,984 43,422 48,179 55,133
Water 11,301 13,919 14,296 14,921
Federal 6,543 8,871 8,394 8,622
State & local 4,758 5,048 5,902 6,299
Rail 1,477 2,426 2,812 3,108
Federal 1,477 2,426 2,812 3,108
State & local 0 0 0 0
Pipeline 76 101 110 89
Federal 76 101 110 40
State & local 0 0 0 49
General support 891 1,447 1,389 873
Federal 878 1,426 1,367 850
State & local 13 21 22 23
Total 268,843 303,784 324,000 352,285

maintenance. Some idea of the importance of transport employment in the

United States economy is seen in Table 2.15 which provides time-series data
from 2000 to 2019. (By way of comparison, using slightly different definitions,
the European Union employed 11.3 million people in transport in 2016.) The
table combines not simply those directly employed in transport but also embraces
those engaged in related industries. Over the longer term, there has been a steady
relative decline in employment in transport in the overall American labor force as
people have switched from public transport to the automobile and as modes such
as trucking, the railroads, and airlines have improved their labor efficiency.
In technical terms there has been considerable scope for factor substitut-
ability in transport as technology has responded to the relatively high costs of
labor in the sector that has allowed for capital to replace labor; for example,
two-person cockpits on commercial aircraft, reduction of brakemen on railroads,
and automated metro trains. Transport has been what is often called ‘technologi-
cally progressive’ (Baumol, 1967). This may come about through scale effects (for
example, the use of larger ships with the same-size crew to carry more cargo) or
the substitution of labor functions (for example, navigation on aircraft by auto-
mated systems or the loading of ships by robots). Given the massive increase in
freight and passenger traffic over time that appears in the earlier tables, combined
with the small overall rise, or even decline, in the number of transport and related
workers, there has clearly been a very significant overall rise in labor productivity

BUTTON 9781786435668 PRINT.indd 36 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 37

Table 2.15 Transport employment in the United States (thousands)

2000 2005 2010 2015 2019

United States labor force 132,011 134,034 130,345 141,825 150,939

Transport-related labor force 13,901 13,395 12,054 13,542 14,815
Transport & warehousing 4,401 4,348 4,179 4,859 5,618
Transport-related manufacturing 11,660 11,076 9,305 10,777 11,492
Postal service 880 774 659 597 607
Government employment 867 888 876 845 869

in the transport industries. The emergence of drones and autonomous taxis and
trucks is strengthening this trend.

2.4 Local Transport

Local transport can roughly be divided into urban and rural, although suburban
transport can have its own characteristics and has become a subject of specific
interest. Almost by definition there are only limited common features across
cities. Figure 2.3 gives an indication of the variations in the sizes of the largest. It
is difficult to argue that they are a homogenous group.
Local transport conditions and use depend on, amongst other things, the
geography of the area (for example, whether it is hilly, has waterways through it,
or is close to another urban area), its history (for example, whether it has inher-
ited a legacy of infrastructure more suited to a bygone era), its economic roles (for
example, whether it is a political or commercial center, or has a large manufac-
turing base), and the political attitudes of local politicians (for example, whether
they foster motor vehicle use or have a public transit bias). Similar factors influ-
ence transport in rural areas, including the nature of local agriculture or mineral
extraction, the historical provision of infrastructure, the density of population,
and the peculiarities of the area’s geography.
Some very broad general patterns are identifiable in at least cities in the
higher-income countries of the world, and to some extent in poorer nations.
We have fewer data on rural areas in part because they tend to be less densely
populated, especially in industrial nations, but also because traditional concerns
of transport planners, most notably those involving traffic congestion, tend to be
much less pronounced. Indeed, many rural areas suffer from depopulation.

Urban Transport

Increasing numbers of people are living in urban areas as agricultural efficiency

improves and as industrialization spreads. The World Bank estimates that
about 55 percent of the global population are now urban dwellers. Even in the
­long-industrialized countries there have been significant changes in land use
and in urban form more generally as incomes have risen, as the service sector

BUTTON 9781786435668 PRINT.indd 37 27/04/2022 11:07

 38 TRANSPORT ECONOMICS, 4TH EDITION

40

35

30
Population (millions)

25

20

15

10

hi
iro

i
ng

ka

ka
yo

li

ai

ty

ul
k
ba

ire
ul

in
r
eh

gh

ac
Ci

nb
Yo
ha

sa
iji
k

Ca

um
Pa

gq
A
To

ar
an

ta
Be

D
o

ew

on
os
K
ic

Is
o
Sh

Så

ex

en

Ch
N
M

Bu
Source: United Nations.

Figure 2.3 Fifteen largest cities by population

has increased in importance, and as there have been demographic changes.

Interventions by urban planners have brought about their own impacts, trends
highlighted 60 years ago by Jane Jacobs (1961). There has been widespread
urban sprawl from most cities as their populations have increased. This not only
poses problems of a practical nature, such as the provision of public services and
the delineation of appropriate areas for local government, but from a descrip-
tive perspective, it means that data collected on old zoning systems can often
obscure some of the newer trends in urban transport, such as long-distance
commuting and reverse commuting. There have been some efforts to overcome
this data problem in recent years. The United States, for example, has sought to
partially resolve it by providing data on ‘metropolitan statistical areas’ (MSAs)
and ‘combined statistical areas’; for example, the former combines data from
administrative units within a contiguous area of relatively high population
density.
Even in suburban areas, transport trends found within traditional city areas
are not replicated because of their lower densities of housing and population,
dominated by single-family homes on small plots surrounded at close quarters
by very similar dwellings. The planning authorities also tend to adopt zoning
patterns that separate residential and commercial development, as well as differ-
ent intensities and densities of development. At the simplest level this means that
daily needs are not within walking distance of most homes and reliance on cars
increases, and that over time the distance of the average commute rises.

BUTTON 9781786435668 PRINT.indd 38 27/04/2022 11:07

 39
TRANSPORT AND THE ECONOMY: SOME NUMBERS ­

To cope with some of these issues, ‘edge cities’ have emerged: concentrations
of business, shopping, and entertainment outside a traditional urban area in what
had recently been a residential suburb or semi-rural community (Garreau, 1991).
Most edge cities develop at or near existing or planned freeway intersections and
near major airports. They rarely include heavy industry. They are large geographi-
cally because they are built around automobile mobility.
As a very broad generalization, most major cities in the world now suffer
from serious traffic congestion problems, and this sometimes can extend well
beyond traditional notions of the ‘rush hour’. This is, however, a subject we shall
deal with in some depth in Chapters 6 and 9. Here we focus more on the general
characteristics of urban transport, and simply highlight some of the major differ-
ences that exist between cities.
Figure 2.4 provides details of the relative use that is made of various pas-
senger transport modes in several major cities across the globe. Very significant
differences can immediately be seen in walking, cycling, and more locally offered
services. There is also a general correlation, as one may anticipate, between the
levels of car ownership (Table 2.16) in cities and the use made of the motor car,
although this may be deceptive if part of the mileage driven is outside of the
urban area. The higher-income countries also generally see less use of public
transport modes. This is partly because of higher car ownership rates, but also
reflects that public transport, and especially buses, are often seen as inferior

Automobile Public transport Walk Bicycle Other a

100
90
80
70
60
50
40
30
20
10
0
Sa o
go

os n

s
ris

n
pe lo

N n
Is bi

Be ul
ng g
M ore

i
l
k

ba
ou
ire
g

en to

rli

Ba ijin
Sa Yor

Ca Os

ro
nb
ie
ia

Pa

To

um
s

al

Se
Be
A

ai
Bu Bo
D
nt

ta
ew

n
N

Note: a. Includes auto rickshaw, private two-wheelers, and taxis.

Source: Deloitte City Mobility Index, 2019.

Figure 2.4 Mode split in a selection of major cities (2017)

BUTTON 9781786435668 PRINT.indd 39 27/04/2022 11:07

 40 TRANSPORT ECONOMICS, 4TH EDITION

Table 2.16 Car ownership per 1,000 population in selected countries (2017)

Country Cars/1,000 Country Cars/1,000 Country Cars/1,000

United States 838g South Korea    475h Colombia 116g

Malta 768h United Kingdom 471 Cape Verde 101a
Australia 730g Denmark 438 Senegal 44e
Poland 719g Malaysia   433e Angola 32e
Canada 685g Bahrain 422 India 28f
Luxembourg 670 Slovakia 408 Pakistan 17g
Italy 655h Bulgaria 393 Burkina Faso 16e
Japan 649h Latvia 381h Mozambique 14d
Spain 648 Hungary 355 Ethiopia 9
Slovenia 541 Argentina 316e Niger 7b
Belgium 508 Israel 314h Guinea 5c
Norway 514 Mexico 297e Rwanda 5c
Greece 487 Uruguay 280e Bangladesh 4e
Libya 483 China 188i Somalia 3b
Netherlands 478 South Africa 174h

Notes: a. 2007. b. 2009. c. 2012. d. 2014. e. 2015. f. 2016. g. 2018. h. 2019. i. 2020.

Source: United Nations.

goods for which the perception of the quantity of the service – sharing with
others, wait time between services, walking to pick-up points, etc. – falls with
rising incomes.
The data also do not tell us much about a topic that we shall devote time
to later in the book. These are the economic policies that are being pursued in
various cities to influence the amount of travel undertaken and the modes that
are targeted. These are generally endogenous to the urban situation in that policy-
makers in cities that have large volumes of car traffic for their size are, because
of this, forced to encourage the use of alternative public transport modes. As
Michael Anderson (2014) has shown, even if public transport carries only a small
share of commuters in a city, it can have large effects on the overall transport
system’s efficiency and congestion levels. Road congestion does not rise linearly
with traffic flow.

Rural Transport

While there has been a global movement towards urbanization, there is still a
significant amount of movement outside of cities, which for convenience we call
‘rural transport’. Much of this is transit traffic in the sense that it involves the
movement of people and goods between urban areas and does not stop, except
for breaks of various kinds, as part of this movement. Pure rural transport has its
origins and destinations outside of urban areas, or involves movement between
cities and non-city areas. Globally, much of this transport involves the move-
ment of agricultural products, and those involved in their production, although
it increasingly can involve tourism. Further, although we focus mainly on

BUTTON 9781786435668 PRINT.indd 40 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 41

economically developed countries where urbanization is extensive, rural transport

is often much more a subject of study in lower-income nations that still have a
significant rural population.
Defining rural transport is not simple, and circumstances vary with countries
and regions within countries (Hough and Taleqani, 2018). For these reasons, only
one nation’s rural transport situation is described. Looking at the United States,
the Office of Management and Budget’s definition of metropolitan areas (core
counties with one or more central cities of at least 50,000 residents with a Census
Bureau-defined urbanized area and a total metropolitan area population of
over 100,000 or fringe counties that are economically associated with core coun-
ties) provides a residual of non-metropolitan counties located outside the areas
that have no cities with 50,000 or more residents. This is taken as rural. About
61 million Americans live in rural areas, according to data from the American
Community Survey. Figure 2.5 shows some of the key socioeconomic differences
between the urban and rural populations of the United States.
In terms of transport, 95.8 percent of households in rural areas had access to
a car in 2015, compared with 88.9 percent of urban households. The ‘carless’ rural
communities were characterized by persistent poverty and had high concentra-
tions of Black, Hispanic, or Native American residents. Rural residents without
access to cars are often dependent on public transport, especially in high-poverty
areas. This is also true of the percentage of people with a disability and the aged.
Median age and the percentage of the population aged 65 or older has increased
in both urban and rural areas since the year 2000, but the increase has been great-
est among the rural population.

Rural Urban

Medium income (% thousand)

Bachelor's degree (%)

Completed high school (%)

Black (%)

White (%)

Disability (%)

Over 65 (%)

Median age (years)

Male (%)

Household size (number)

0 20 40 60 80 100 120 140 160 180 200

Source: Mattson (2017).

Figure 2.5 Comparisons of the United States urban and rural populations (2015)

BUTTON 9781786435668 PRINT.indd 41 27/04/2022 11:07

 42 TRANSPORT ECONOMICS, 4TH EDITION

During the 1990s, public transport services grew, with non-metropolitan provid-
ers offering 62 percent more passenger trips, 93 percent more miles traveled, and
60 percent more vehicles available. Public transport is available in 60 percent of
rural counties, but with 28 percent of about 1,200 systems offering only a limited
service. About two-thirds of rural systems operate in single counties or are city/
town in scope; only one out of four rural transit providers operate in a multi-
county area. About 60 percent of rural providers are public agencies, and roughly
a third are non-profit groups; less than 5 percent are private companies or tribal
entities.
The number of American rural communities served by long-distance bus ser-
vices declined sharply in the years following bus deregulation in 1982. The inter-
city bus industry now serves about 4,300 locations, down from over 15,000 prior
to 1982, with many of the service discontinuations concentrated in rural com-
munities. Still, 89 percent of the rural population is served by long-distance bus
services; it is the dominant mode of scheduled inter-city passenger transport for
most rural residents. One reason for this is that fewer than 200 non-metropolitan
places in the United States are served by passenger rail services, and, following the
terrorist attacks of 2001, the airline passenger industry suffered a downturn, with
smaller communities especially hard hit. The number of flights to small, non-hub
airports dropped 19 percent between 2000 and 2003. The outbreak of Covid-19
in 2020 led to a further decline in services.
Many rural roads in the United States are in need of repair. In 2020,
13 percent of America’s rural roads were rated in poor condition and 21 percent
in mediocre condition. Additionally, 8 percent of the nation’s rural bridges
were rated in poor/structurally deficient condition exhibiting significant dete-
rioration to major components. These bridges were often posted for lower
weight or closed to traffic, restricting or redirecting large vehicles, including
agricultural equipment, commercial trucks, school buses, and emergency services
vehicles.

The Micro-micro Level: Households

At the lowest level of aggregation, it is households and individuals that use a

nation’s transport system and ultimately pay for it. Travel budgets vary consider-
ably between households, and across income categories. A plethora of factors
influence this: household income, locations of home and work, age of the house-
hold members, physical capability of household members, priorities in leisure
spending, climate, and so on. Added to this, the adoption of different definitions
and data collection methods make international comparisons challenging but
some general patterns emerge.
While there is much variability across households within any country, as
well as between countries, it is generally true that household transport expendi-
ture vies with housing and food as the largest item in households’ budgets.
From Table 2.17, which sets out details of the average British family’s weekly

BUTTON 9781786435668 PRINT.indd 42 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 43

Table 2.17 
Weekly household expenditure out of net income in Great Britain (2018/19, in
2003 prices)

Average weekly expenditure (£)

Food & non-alcoholic drinks 61.90

Alcoholic drink, tobacco, & narcotics 8.00
Clothing & footwear 24.40
Housing, fuel, & power 79.40
Household goods & services 40.08
Health 8.00
Transport 80.20
Communication 21.30
Recreation & culture 76.90
Education 5.70
Restaurants & hotels 51.90
Miscellaneous goods & services 46.70
Other expenditure items 77.20
Total 585.60

Source: UK Family Expenditure Survey.

expenditure, transport is clearly one of the larger items accounting for about 14
percent of a household’s budget. For comparison, and accepting the caveat that
definitions can differ between countries, for the United States in 2018 (in 2003
prices), an average household with expenditures of $788.61 spent $259.47 of it on
housing, $133.91 on transport (about 17 percent), and $99.00 on food. In Japan
it was 10 percent. Data are sparser, but the same three items dominate household
expenditure in poorer countries. In India, for example, in 2017 food accounted for
28 percent of household expenditure, housing and utilities 17 percent, and trans-
port 16 percent; while in Brazil housing accounted for 36.6 percent of expendi-
ture, transport 19.6 percent, and food 17.5 percent.
The pattern does seem to be changing slightly, or at least in some of the
higher-income countries, with the outlays on transport declining as a propor-
tion of household expenditure. In the United States, for example, it was about
19 percent in 1987. The explanation for this may in part be that real transport
costs have fallen, and therefore the same or more transport may be obtained for
a smaller outlay. There may be more substitutes for transport as information and
communications systems have improved. The increased use of substitutes such as
video conferencing, remote working, and online shopping being examples of this.
But it may be that the demand for transport has moved towards a point of satu-
ration as household members make nearly all the trips they want, and increases
in income are directed elsewhere. Transport may be seen in economic terms as
becoming an ‘inferior good’ in this case. In other words, the Engel curve, describ-
ing how the proportion of household income spent on a good like t­ransport
varies with income, has taken a negative gradient.

BUTTON 9781786435668 PRINT.indd 43 27/04/2022 11:07

 44 TRANSPORT ECONOMICS, 4TH EDITION

2.5 Emerging Trends

The world is not static and there are several emerging, as well as ongoing, trends
that are beginning to influence the sorts of issues that transport economists
address. Added to this, in the medium term, is the considerable uncertainty that
will accompany the world economy and social systems as the longer-term effects
of the Covid-19 pandemic begin to emerge. At the time of writing there is con-
siderable uncertainty about the exact form and magnitude of the effects of the
virus on markets and institutional structures. Here there is no attempt to join
the debates that are taking place, but rather we assume that some of the broad
changes that were being seen pre-Covid will have matured and will have implica-
tions for transport. We shall, however, add a few separate observations regard-
ing Covid in Section 2.6. Here the focus is on trends apparent at the beginning
of 2020.
First, in the established industrial world the long-term trend, despite peri-
odic downturns in the business cycle (including recessions in many countries
from 2007 to 2009) and periods of very rapid growth as in the late 1990s (‘irra-
tional exuberance’, to quote Alan Greenspan), would seem to be for continued
economic expansion. The recent past has already witnessed significant increases
in traffic and vehicle ownership in these countries and there are reasons to expect
this trend to continue in broad terms over the years. Globalization does not
seem to have fully been completed, although the economic success of the Single
Internal Market within the European Union and the NAFTA/USMCA are pro-
viding demonstration effects of the benefits of removing trade restrictions. The
natural growth in trade accompanying these developments will create demand for
more transport services, but this inevitably will take place in the context of limited
infrastructure capacity. Even if there are major investment initiatives, it takes time
to construct new ports, roads, and so on.
Looking to Europe, a lot has been written on impacts for transport of the
creation of the Single European Market in 1992, and certainly the new situation
with an enlarged membership has implications for transport not only within the
member states but also for countries of the European Free Trade Area involved
in the larger European Economic Space (see Chapter 14). The problem has been
how to cater for such traffic growth in a fully, economically efficient manner, a fact
recognized by the European Union in its transport policy statements.
Second, it is becoming clearer that the liberalization of Eastern Europe,
coupled with the new political geography that has emerged, is posing problems as
well as offering opportunities for the countries in the region. Liberalization meant
that the overall ‘transport market’ in Europe expanded considerably in line with
major new urban and industrial centers being brought within the market system.
It means, therefore, that many major transport links are now part of Europe’s
transport future. In many ways this has proved advantageous for the long-term
development of European transport because it creates something more akin to
a natural market for transport services than existed previously, but there are
­problems with defining priorities and financing investments in the system.

BUTTON 9781786435668 PRINT.indd 44 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 45

Short-term problems also remain because of the attitude regarding transport

that had grown up in Eastern and Central Europe since the early 1990s and the
impact this has had on the physical transport infrastructure now in place. The
transport systems of Eastern Europe were traditionally dominated by rail and
suffered from low productivity and over-staffing compared to Western Europe.
It had been developed from the late 1940s to meet the trading patterns of the
members of the Council for Mutual Economic Assistance. There remain chal-
lenges in filling the voids in infrastructure linking the Eastern economies with the
West that grew up in this period.
North America has begun, albeit slowly, to integrate its economies through
free trade arrangements that are shifting patterns of production and, with this,
the demands for transport services. Coupled with this, the general growth in
importance of international trade for the United States has put increasing strains
on its international border crossings, a situation worsened by the costs of meeting
enhanced security arrangements. As with Europe, there are pressures for seeking
innovative ways to finance new infrastructure, maintain and upgrade what is
already in place, and to make better use of existing capacity (Transportation
Research Board, 2003).
It is not only in the industrialized nations and in the post-communist states
that new conditions are emerging. If there is to be economic development in
the low-income countries of Africa, Asia, and South and Central America then
transport will inevitably change. There is ample evidence from China, India, and
some other economies that the economies of some of these states have expanded
rapidly since 2020, although this is not so true of many of the lowest-income
countries. Returning to Table 2.16, this shows very low levels of car ownership in
the poorest countries. Additionally, these countries tend to have some of the least
safe transport systems; Botswana, for example, has the highest road mortality
rate on roads at 30 deaths per 100,000 population.
Economic expansion, however, especially if it results in significant growth in
some larger countries, could lead to substantial demands for car ownership, more
mobility, and the need for major new infrastructure initiatives, as well as imposing
serious congestion and safety challenges. As can be seen from earlier discussions,
the link between per capita income and vehicle ownership is a strongly positive
one. It will also mean that there will be increasing strains put on the environment,
both within these countries and in terms of global effects such as emissions of
greenhouse gases.
A significant feature of most low-income countries in recent years has been
the secular drift of population into urban areas. Historically, the growth of car
ownership in the largest cities in Third World countries was already of the order of
7 to 15 percent in the decade 1960–70. Added to this, the actual cost of transport in
these cities has risen significantly, with households spending over 5 percent of their
income on transport, and in some cases 15 percent, and city governments spending
15 to 25 percent of their annual budgets on transport investments and operations.
Looking forward, the transport situation in the urban areas of the devel-
oping world is inevitably going to get worse in the short to medium term. The

BUTTON 9781786435668 PRINT.indd 45 27/04/2022 11:07

 46 TRANSPORT ECONOMICS, 4TH EDITION

Exhibit   Response of air transport to a pandemic

The 2020 Covid-19 pandemic seriously affected the demand for and supply of air transport.
Passengers were reluctant to fly for fear of infection and official policies of quarantining and
banning traffic from selected countries caused a catastrophic collapse in demand. For cargo,
the accompanying global recession meant contraction of supply chains.
The suddenness of the pandemic resulted in the hub-and-spoke system simply evaporating,
pushing down passenger load factors and pushing up costs. With no adjustment time, the
industry quickly experienced a collapse in domestic and international air passenger demand
that led to airlines having to park more than 17,000 passenger jets by May 2020. Two dozen
airlines had collapsed by the summer of that year, despite measures by many governments
to sustain at least a core network of services.
While air cargo rose 2 percent between February and July 2020 compared to the
comparable period for 2019 as initially supplemental airlift was required to relocate medical
equipment and supplies, later significant volumes of cargo relocated from passenger
aircrafts’ belly holds to dedicated cargo planes. The figure provides a clearer picture of
trends in transported cargo volumes.
15.0%

10.0%

5.0%

0.0%
Tonne-kilometers

–5.0%

–10.0%

–15.0%

–20.0%

–25.0%

–30.0%
Jan 18

Mar 18

May 18

Jul 18

Sep 18

Nov 18

Jan 19

Mar 19

May 19

Jul 19

Sep 19

Nov 19

Jan 20

Mar 20

May 20

Jul 20

Covid-19 affected markets differently, requiring diverse responses by airlines. Lufthansa, for
instance, advanced retirement of its A380 fleet, while many carriers reduced their fleets,
put aircraft into ‘long-term storage’, or cancelled orders. In other cases, including Austrian,
Swiss, and Icelandair, carriers reconfigured planes for cargo use. Others pulled out of
mergers, for example the withdrawal of LOT’s bid for Condor.
Governments sought to tide their carriers over. Some provided loans, for example
Germany ($6.75 billion to Lufthansa) and Korea ($970 million to Korean Air via state-
owned banks). Some, such as the United States, introduced payroll grants, while many
provided easy credit facilities. Airlines, in turn, made concessions, for example giving up

BUTTON 9781786435668 PRINT.indd 46 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 47

landing slots to competitors, withdrawing from short routes where high-speed rail was an
alternative, and modifying membership of their boards.
Given their strategic role, the oversight bodies could do little besides offering advice and
collating information as it became available. The International Civil Aviation Organization,
for example, developed a ‘COVID-19 Recovery Platform’ offering guidance for airports,
airlines, aircraft, crew, and cargo handlers on how to reduce the public health risk
while strengthening confidence among the traveling public, the global supply chain, and
governments.
See also: S. Albers and V. Rundshagen (2020) European airlines’ strategic responses to
the COVID-19 pandemic (January–May, 2020), Journal of Air Transport Management, 87,
No 101863.

mid-1980s saw eight of the largest cities in the world with populations of over
10 million located in low-income countries. Predictions are that this number will
have doubled by the end of the twenty-first century, while 18 further agglomera-
tions in the developing world will have populations of between 5 and 10 million.
A major difficulty is that the growth of urbanization and the level of motor car
ownership and use are closely linked, While this is partly due to the concentra-
tion of wealth in the urban areas it is also entwined with the geographical spread
which accompanies urbanization and the resultant increase in the average journey
length.
Comparing Nairobi and Mexico City, for example, shows average trip
lengths of between 1.5 and 2.8 miles for the former, while those for the latter,
which is much larger, are between 3.5 and 6.0 miles. Public transport is much
less efficient at serving a spatially dispersed market and hence the automobile is
used more often. How to plan and cater for the inevitable expansion in traffic as
urbanization continues in these countries will be a mounting problem for trans-
port economists, as will be finding resources. Some indication of the resources
that are currently being spent in China, for example, is that $4.25 billion was
invested in Beijing’s infrastructure in 2004, and another $22 billion leading up to
the 2008 Olympics, to reduce the city’s traffic congestion issues.
The emergence of the economies of China and India has seen further
urbanization trends and with this has come new demands for consumer goods,
including automobiles (see Table 2.18 for details of China). These countries com-
bined account for about 37 percent of the world’s population, and between 1980
and 2005 the real personal income of India more than doubled, and more than
quadrupled in China (Pucher et al., 2007). The outcome has been a tripling of
motor vehicles in India and a tenfold increase in China. The transport situation
has been exacerbated by the urbanization resulting from these economic trends;
the development area of Chinese cities, for example, tripled between 1985 and
2005 while the population only doubled, producing significant urban sprawl. A
similar pattern emerges for India. There are different patterns of travel in urban
areas between the two countries, however, with over half the trips in Chinese

BUTTON 9781786435668 PRINT.indd 47 27/04/2022 11:07

 48 TRANSPORT ECONOMICS, 4TH EDITION

Table 2.18 Durable goods for 1,000 households (2006 or most recent prior year)

China India Total

Urban Rural Urban Rural

Automobiles 4.3 – 4.0 0.7 1.7

Bicycles 117.6 98.4 51.9 57.2 55.7
Cameras 48.0 3.7 0.0 0.0 0.0
Computers 47.2 – 0.0 0.0 0.0
Microwave ovens 50.6 – – – –
Motorcyclesa 20.4 44.6 28.3 7.9 13.6
Refrigerators 91.8 22.5 30.8 4.8 12.1
Telephones 93.3 64.1 – – –
Mobile telephones 137.4 89.4 70.4 27.5 39.5
Televisionsb 137.4 89.4 70.4 27.5 39.5
Video disc playersc 70.2 – 8.2 1.7 3.6
Washing machines 96.8 43.0 12.5 0.9 4.1

Notes: a. Data for India include scooters. b. Data for China include color TVs; data for
India include all TVs. c. Data for India include VCRs.

mega-cities being non-motorized compared to only 25 percent in India. In both

cases, however, motorization is growing rapidly – especially in terms of motor-
ized two-wheeled vehicles that, for instance, grew in number from 200,000 to over
50 million in China from 1981 to 2002.
But it is not only the local transport systems that are finding it difficult to
cope with rising demands as incomes rise in many developing countries; other
elements of transport infrastructure networks are also coming under increasing
pressure. The amount and quality of infrastructure varies widely between devel-
oping countries. In many cases, the issue is simply that there is insufficient infra-
structure for the needs of a country. This can be a major problem for landlocked,
exporting countries which, in addition to moving their output over their own
territories, have to also rely on the infrastructure of intermediate nations (Arvis
et al., 2010). While this has been a long-standing problem for some countries, in
other cases, there is simply a shortage of capacity to meet a sudden increase in
demands for their products.
Countries like China, which have seen large increases in their international
trade since 1990, find themselves with pressures on their airport and seaport
capacities to meet the globalized patterns of activities they now engage in.
Table 2.19, for example, shows the increase in Chinese freight movements by
mode between1980 and 2018, and Figure 2.6 shows the rapid growth in air cargo
traffic at the country’s main airports from 1990 to 2004. This pressure on capacity
extends to national inter-urban transport systems as the domestic supply chain
expands not just to feed the gateway ports but also as increased production itself
puts pressure on the domestic freight networks.
In terms of surface transport, in 2005 China had a road network of
more than 3.3 million kilometers, although about 1.47 million kilometers was
‘village roads’; paved roads were 770,265 kilometers of the total in 2004; the

BUTTON 9781786435668 PRINT.indd 48 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 49

2,500,000
Beijing Shanghai Guangzhou

2,000,000
Cargo throughput (tonnes)

1,500,000

1,000,000

500,000

0
90
91
92
93
94
95
96
97
98
99
00
01
02
03
04
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
Year
Figure 2.6 Air cargo growth at China’s main airports (1990 to 2004)

Table 2.19 Annual freight movement in China by mode (million tonnes)

1980 1990 2000 2010 2018

Rail 1,112.79 1,506.81 1,785.81 3,642.71 4,026.71

Road 3,820.48 7,240.40 10,388.13 24,480.52 39,568.71
Water 426.76 800.94 1,223.91 3,789.49 7,026.84
Air 0.09 0.37 1.97 5.65 7.70
Pipeline 105.25 157.50 187.00 499.72 898.07

remainder were gravel, improved earth standard, or merely earth tracks. In

2003 they moved nearly 11.6 billion tons of freight and provided 769.6 trillion
­passenger-kilometers. Highways (some 130,000 kilometers) have been critical
to China’s economic growth, as they worked to mitigate a poor distribution
network. Massive new investments mean that all major cities are expected to be
linked with a 55,000-­kilometer ­inter-provincial expressway system by 2020. (By
comparison, the United States’ Interstate freeway system is about 47,000 miles.)
The highway and road systems carried nearly 11.6 billion tons of freight and
769.6 trillion passenger-kilometers in 2003.
Both bi-lateral and multi-lateral aid, most notably through the World Bank
and the United Nations, but also from blocs such as the European Union, has been
used in the past to help less economically advanced countries improve their trans-
port systems. Table 2.20, for example, shows the recent expenditures of the World
Bank, and future commitments are of the same order of magnitude, although

BUTTON 9781786435668 PRINT.indd 49 27/04/2022 11:07

 50 TRANSPORT ECONOMICS, 4TH EDITION

Table 2.20 Commitments by the World Bank for transport ($ billions)

Mode Commitments Commitments

1996–2000 Percent 2001–2006 Percent

Roads 13.0 73 11.9 73

Railways 1.5 9 1.3 8
Ports 1.2 6 0.5 3
Aviation 0.1 0 0.5 3
General transport 2.2 12 2.2 13
Total 18.0 100 16.4 100

Source: The World Bank (2007).

with a somewhat broader focus in terms of how the monies will be spent. Most of
these funds are committed to enhancing the transport infrastructure of recipient
countries, but there is increasing interest in ensuring that environmental and safety
considerations are brought to bear at the project decision-making level.
It seems unlikely that the poorer nations of the world will require less
assistance to maintain their transport systems in the future, although there are
now more efforts by the aid-giving agencies to ensure that the monies provided
are used effectively. Added to this, there is increased encouragement for these
countries to do more by way of self-help in terms of imposing user charges on
infrastructure to at least recover some of the costs and to limit its use to activities
contributing most to economic development.

2.6 Some Comments on the Short-term Effects of Covid-19

on Transport

The combined short-term market effects of Covid-19 in 2020 and the reactions
of policy-makers regarding transport were unprecedented in peacetime. A few
brief comments are provided with some data to highlight the scale and nature
of the effects of the pandemic within the first 18 months of its arrival. The aim
is not in any way to offer a full picture but rather to focus on several of the more
pronounced patterns, as well as indicating some recovery trends.
At the global level, air transport and international rail services were
impacted in two different ways. The demand for passenger services fell as inter-
national business and leisure declined as both individuals and businesses cut
back on their mobility when confronted with the threat of catching the virus.
Governments restricted access to their countries to contain the spread of Covid-
19 in the public interest. Table 2.21 offers an insight into the magnitude of the
impacts of Covid-19 on passenger seats being flown for major airline markets.
A wide variation of initial adverse effects is clear. Globally, seats fell by nearly
40 percent between 2019 and 2020, while countries such as the United Kingdom,
Singapore, Germany, and the United Arab Emirates saw losses of well over

BUTTON 9781786435668 PRINT.indd 50 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 51

Table 2.21 
Air seats June 2021, as a percentage of seats for comparable dates in
2019 and 2020

June 24, 2019 (%) June 22, 2020 (%)

Australia –36.2 336.3

Brazil –43.6 242.6
China –0.3 34.9
France –54.1 196.7
Germany –67.0 115.0
India –49.1 79.3
Japan –58.7 –23.2
Mexico –15.3 122.2
Singapore –82.9 209.7
South Africa –47.7 465.9
South Korea –46.8 16.6
Spain –50.1 230.8
UAE –59.4 156.7
United Kingdom –77.7 43.3
United States –18.8 137.4
Global –37.9 81.4

Source: Official Airline Guide.

50 percent. The third column of the table shows the pick-up in 2021 with traffic
up over 81 percent over 2020.
Regarding local transport, there were considerable variations on traffic flows
across cities. Given the substitutability of physical movement and electronic
communication, cities with larger service-sector employment saw larger falls in
vehicular traffic as people worked from home and made more use of various
delivery services. Some modes lost more traffic than others. As can be seen from
Figure 2.7, while the temporal patterns of automobile and metro (in London, the
Tube) are very similar between January 2020 and June 2021, the overall decline in
public transport was significantly greater. This was almost certainly due to both
concern with traveling with potentially affected strangers and the requirement to
wear face masks on public transport.
While person and goods movements were rapidly and radically reduced with
the onset of the pandemic, the movement of information was much less affected.
This was because of the greater uptake of electronic communication and, in
particular, of e-commerce and video meetings. This was certainly not a new phe-
nomenon. Use of video conferencing and various other forms of electronic infor-
mation exchange had been gradually growing since the emergence of the internet
and world-wide web in association with hardware developments of such things
as the smartphone. Considerable numbers of individuals were already working
at home prior to the arrival of Covid-19 (Figure 2.8). But after the onset of the
pandemic, video conferencing increased exponentially.
One indicator of these trends are the flows of revenues earned by the video
conferencing companies. Zoom, for example, saw a 169 percent first-quarter
increase in its year-on-year revenue for the financial year ending January 2021,

BUTTON 9781786435668 PRINT.indd 51 27/04/2022 11:07

 52 TRANSPORT ECONOMICS, 4TH EDITION

120%

100%

80%

60%

40%

20%

0%

120%

100%

80%

60%

40%

20%

0%

Source: UK Department for Transport.

Figure 2.7 Daily London automobile (upper) and Tube passengers (lower), March 3,
2020 to June 21, 2021 compared with same dates in 2019

which accelerated to 355, 367, and 369 percent in the second, third, and fourth
quarters. In terms of e-commerce, online retail sales in the United States increased
by 32.4 percent year-on-year in 2020 and were up 39 percent in the first quarter
of 2021. While some of this was the result of the ongoing growth in electronic
transfers, much came from reductions in physical interactions.

2.7 Where are the Numbers?

This chapter has made copious use of a wide range of data sources. Statistics of
this kind, however, can rapidly become outdated. There have been long-standing
concerns over the availability of timely economic data regarding transport, its
limited coverage, and its quality (for example, see Head, 1980), although this may
reflect the insatiable demand of economists as much as reality. The widespread
use of the world-wide web makes it easy to update data and to add new sets. It

BUTTON 9781786435668 PRINT.indd 52 27/04/2022 11:07

 TRANSPORT AND THE ECONOMY: SOME NUMBERS ­ 53

16

14

12

10

0
La a
G ia
H ece

itz K
ry

Sl ly
ia

Ire d
nd

G ain

Sw ny

m k
he g

s
N en

Es y

Fr a
Be ce
D ium

nd
i

ni

et ur
n

xe ar
a
an

tv

Sw U
ak
Ita
ga

an
ed

w
la
la

a
Sp

to

Lu nm

N bo
re

rla
m

lg
m

ov

or
er
un

er
Ro

e
Figure 2.8 The percentage of employed persons aged 15–64 working from home (2017)

also allows readers to critique the relevance and accuracy of the quantitative
information that is offered in this book. Here, a few words are offered as to the
sorts of public data sources that are openly available as well as some information
on the specific types of economic statistics that can be useful when looking at how
transport systems function.
Data on transport are widely available but differ in their quality and the ways
in which they are presented. The quote of Disraeli’s cited earlier should not be
forgotten. The units used, for example, need to be treated carefully; fuel use may
be expressed in kilometers per liter or in miles per gallon (and of course a United
States gallon is about 20 percent smaller than an imperial gallon to confuse things
further). There may be a big difference, as already noted, between the importance
of, say, various modes of transport when assessed in physical terms as opposed to
value terms. Money values themselves can be at current values, with no allowance
for inflation, or in constant prices set against a base year when they are adjusted
for changes in the value of money.
Governments at all levels publish data, as do trade associations, transport
undertakings, and research institutes. A few useful sites include the following.
For the United States, national data are collected by the US Department of
Transportation’s Bureau of Transportation Statistics (http://www.bts.gov/); for
the United Kingdom corresponding data are provided by the Department for
Transport (http://www.dft.gov.uk/pgr/statistics/); Eurostat provides data for the
European Union (http://epp.eurostat.ec.europa.eu); and for Canada a com-
parable site is https://tc.canada.ca/en/road-transportation/motor-vehicle-safety/
statistics-data. The OECD’s International Transport also offers compilations
of international data (https://www.oecd-ilibrary.org/transport/data/itf-transport-
statistics_trsprt-data-en). With modern search engines it is a relatively simple
matter to seek out national data for other countries.

BUTTON 9781786435668 PRINT.indd 53 27/04/2022 11:07

 54 TRANSPORT ECONOMICS, 4TH EDITION

Internationally, the United Nations’ International Civil Aviation Organization

(http://www.icao.int/) and its International Maritime Organization (http://www.
imo.org/) provide statistics on air transport and shipping respectively, and
its World Tourism Organization (http://www.unwto.org/) covers the transport-
intensive and growing tourism sector. The World Bank offers data and studies
relating to the transport and related conditions in developing countries (http://
www.worldbank.org/).
Major professional and trade organizations include the International Union
of Railways (https://uic.org/support-activities/statistics/), the International Air
Transport Association (https://www.iata.org/en/services/statistics/), and the
International Road Confederation (https://www.irf.global/statistics/).

References

Anderson, M.L. (2014) Subway, strikes, and slowdowns: the impacts of public transit on
traffic congestion, American Economic Review, 104, 15–33.
Arvis, J.-F., Raballand, G., and Marteau, J.-F. (2010) The Cost of Being Landlocked:
Logistics Costs and Supply Chain Reliability, World Bank.
Baumol, W.J. (1967) Macroeconomics of unbalanced growth: the anatomy of urban crisis,
American Economic Review, 57, 415–26.
Button, K.J. (2011) Transport policy, in A.M. El-Agraa (ed.), The European Union:
Economics and Policies, 9th edn, Cambridge University Press.
Garreau, J. (1991) Edge City: Life on the New Frontier, Doubleday.
Head, J. (1980) Transport statistics: sources and limitations, Transportation Journal, 20,
33–46.
Hough, J. and Taleqani, A.R. (2018) Future of rural transit, Journal of Public
Transportation, 21, 31–42.
Jacobs, J. (1961) The Death and Life of Great American Cities, Random House.
Krugman, P. (2009) The increasing returns revolution in trade and geography, American
Economic Review, 99, 561–71.
Mattson, J. (2017) Rural Transit Fact Book 2017, Upper Great Plains Transportation
Institute.
Pucher, J., Peng, Z., Mittal, N., Zhu, Y., and Korattyswaroopam, N. (2007) Urban
transport trends and policies in China and India: impacts of rapid economic growth,
Transport Reviews, 27, 379–410.
Transportation Research Board (2003) Freight capacity in the 21st century, TRB Special
Report 271.
The World Bank (2007) A Decade of Action in Transport: An Evaluation of World Bank
Assistance to the Transport Sector, 1995–2005, World Bank.

BUTTON 9781786435668 PRINT.indd 54 27/04/2022 11:07

 3 Transport and Location

3.1 The Desire for Movement and Mobility

The nineteenth-century Scottish novelist, Robert Louis Stevenson, once said,

‘For my part, I travel not to anywhere, but to go. I travel for travel’s sake. The
great affair is to move’ (Stevenson, 1879). As highlighted in the discussion
in Chapter 1, Stevenson was very much in the minority. Few people travel
for the sheer joy of it, although some modes of transport do arouse feelings
of excitement, romance, or sentiment. But even here, there are fine lines to
be drawn. Drifting down a Venetian Canal in a gondola, sipping a glass of pro-
secco as the sun sets with the right person sitting beside you can clearly provide
utility, but is this really from the transport? There are also, of course, people
who take touring holidays each year where sight-seeing from a bus or car is an
integral part of the enjoyment, but, again, is it the transport per se that provides
the utility?
Most individuals travel because they wish to benefit from the social, recrea-
tional, educational, employment, and other opportunities that become accessible
with movement. Similarly, freight transport opens opportunities for greater effi-
ciency in production, and it permits extensive geographical specialization with the
accompanying benefits of an increased division of labor. Larger markets can be
served. More simply, transport permits the spatial disadvantages of separation to
be reduced. The concept was taken even further in 1997 with the publication of
Frances Cairncross’ The Death of Distance, which argued that the internet was
becoming an enormously efficient way to transport digital data around the world.
Put simply, that electronic communication was replacing the personal and physi-
cal transport in many activities.
Almost five decades ago, however, Michael Thomson (1974) provided
a helpful and durable classification of seven main reasons why people in the
modern world still desire to transport either themselves or their property:

• The heterogeneity of the earth’s surface means that no one part of it can
provide all the products people wish for. An acceptable bundle of such goods
can only be obtained by either moving around collecting them, or having
them brought to you.
• The continuation of modern society and the high levels of material well-
being rely upon a degree of productive specialization. Industry requires a
multiplicity of diverse inputs which must be collected from wide-ranging
sources and, to permit the necessary level of specialization, extensive market
areas must be tapped and served.

BUTTON 9781786435668 PRINT.indd 55 27/04/2022 11:07

 56 TRANSPORT ECONOMICS, 4TH EDITION

• In addition to specialization, high-quality transport permits the exploitation

of other major economies of scale. There are essentially technical econo-
mies associated with high levels of output and which include automation,
bulk handling, research and development activities, mass marketing, and
purpose-built equipment.
• Transport has always served a political and military role. Internally, a
country seeks good transport both to permit more effective defense of its
borders and to improve the political cohesion of the nation. The Romans
were certainly aware of this and most of their road building was to this end,
as was the construction of trans-continental railroads in the United States
and Canada. Externally, good transport permits a country to dominate
any colonial or subservient provinces, while more aggressive states require
transport to pursue their expansionist policies. Politically, the ownership of
expensive, modern transport infrastructure (especially aircraft or mercantile
marine) is also treated as a symbol of power and status. In most developed
countries the scale of transport required to meet economic needs normally
exceeds that required to meet political or military criteria, although indi-
vidual components of a transport system (for example, specific roads or
airports) may be provided explicitly for non-economic reasons.
• Without transport, social relationships and contacts can be very restricted.
Transport permits social intercourse, and with it may come a greater under-
standing of the problems and attitudes of various geographically distant
groups. In the developed world the enhancement of social understanding
brought about by increased international travel is well recognized, but in
many less developed countries the introduction of much more basic trans-
port technology can have profound effects upon the social relationships
between inhabitants of formerly isolated towns and villages which are, by
‘Western standards’, very close together.
• Modern transport has widened cultural opportunities, permitting people to
examine the artistic treasures of other countries and to explore their own
national heritage. It also allows for the staging of international exhibitions,
sports spectaculars, concerts, parades, and fairs that stimulate new trends
and innovations in the cultural and sporting spheres.
• Transport is desired to permit people to live and work apart; it permits
the geographical separation of employment from leisure. It increases the
life-style options open to people, giving them a choice among residential
locations away from cities but involving a heavy commitment to travel, or
ones much closer to the main employment center but involving short com-
muting journeys. Transport, quite simply, widens the location choices open
to households.

What becomes apparent from Thomson’s listing is the close link between
location decisions (of both individuals and firms) and the transport system.
Much of what he says is still valid but is becoming increasingly nuanced with new
technologies, personal life-style, and new approaches to logistics. Here we look

BUTTON 9781786435668 PRINT.indd 56 27/04/2022 11:07

 TRANSPORT AND LOCATION ­ 57

at both the more traditional economic models linking transport and economic
decisions regarding location and consider some of the more recent implications
for these models following the arrival of the information age. The latter topic is
extended in Chapter 10, which deals with the development in logistics and its
economic analysis.

3.2 Transport: the ‘Chicken’ or the ‘Egg’?

If one takes very broad statistics, then there is unquestionably a link between
mobility and affluence. This is, for example, clear from Table 3.1, which presents
some of the findings contained in a classic study by Wilfred Owen (1987), regard-
ing levels of per capita GDP across a range of countries and the transport mobil-
ity expressed as a combination of such things as passenger miles per annum of
automobile, rail, and domestic air travel, miles of railway, and number of trucks,
enjoyed by those living in these countries. While measures are not very sophis-
ticated, a general correlation is apparent. What is less clear is the direction of
causation. Does high income lead to or result from higher levels of mobility? Or
to put it in more technical terms, is mobility a function of the supply of transport
services? Or does the supply of transport services result from an inherent demand
for mobility? The answers are not immediately apparent.

Table 3.1 Mobility across countries

GNP per capita Travel mobility Freight mobility

Switzerland 139 104 81

Sweden 119 96 151
United States 106 160 260
Netherlands 101 83 42
France 100 100 100
Canada 95 114 374
Australia 91 107 335
Japan 87 96 94
United Kingdom 63 78 47
Italy 53 86 49
Spain 43 54 44
Venezuela 24 32 55
Yugoslavia 24 32 55
Brazil 18 18 23
Mexico 15 14 42
Colombia 11 6 47
Nigeria 6 5 5
Egypt 5 5 13
Pakistan 2 3 10
China 2 3 16
India 2 5 26
Bangladesh 1 2 3

BUTTON 9781786435668 PRINT.indd 57 27/04/2022 11:07

 58 TRANSPORT ECONOMICS, 4TH EDITION

At another level, in previous chapters it was suggested that one of the reasons
for the increased interest in transport economics from the late 1960s was the
recognition of the important link between transport and land-use patterns, and
especially those relating to urban form and intra-urban location. The effects of
changes in the transport system on land use tend to be long-term (they are often
called ‘activity shifts’) but, given the longevity of much transport infrastructure,
such interactions must to some extent concern current transport policy-makers.
The changes that occur in land use will also, in turn, by altering the nature and
size of the local residential population and industrial bases, exert an enormous
influence on future transport demand. A major new suburban underground
railway system, for example, will immediately attract some travelers away from
other modes of transport in addition to encouraging trips to be made by former
non-travelers. In the longer term, sites near the underground termini will become
desirable while those further away will appear relatively less attractive. There will
therefore be important implications for residential and employment location pat-
terns. Additionally, changing location and trip-making patterns will alter such
things as car ownership decisions.
While these interactions are now fully recognized, it is difficult to construct
a comprehensive theory that fully reflects all the linkages. The problem is further
compounded by the fact that transport and land-use changes are ongoing modifi-
cations to the spatial economy. There are continual cycles of cause and effect, and
it is impossible to decide upon a point where it is sensible to break into this contin-
uum of change. Consequently, from a pragmatic standpoint, there is need to make
careful judgments as to whether land use is influenced by transport or vice versa.
To some extent the final decision must rest with the questions being considered.
Urban and regional scientists tend to treat transport as the influential vari-
able because their focus of attention is on the spatial dimension. Questions are
posed, for example, in terms of why certain population densities occur or why
specific urban economies interact. In contrast, transport economics usually
accepts a given land-use pattern and looks at methods of providing efficient
transport services within this constraint. Questions center, for instance, on prob-
lems of aligning routes or controlling traffic flows.
An example of this latter approach that reveals both the methodology of
conventional transport economics, but also highlights some of the difficulties
in the modeling of urban decision-making, was developed by John Kain (1964).
His econometric study looking specifically at public transport subsidies and
calibrated using information from a 1953 survey of 40,000 households in Detroit,
adopts four steps in its argument:

• Workers initially select a residential density in which to live depending upon

their income, their preference for a specific plot size, and the price of resi-
dential land.
• Once a location has been selected, the decision to purchase a car is treated as
dependent upon the local residential density, family income, public transport
availability, and the composition of the worker’s family.

BUTTON 9781786435668 PRINT.indd 58 27/04/2022 11:07

 TRANSPORT AND LOCATION ­ 59

• The decision whether to use public transport for the journey to work, besides
depending upon the previous decisions regarding location and car owner-
ship, is thought to be influenced by the quality of local public transport, and
the demands of non-working members of the household to make use of the
car (if one is owned).
• The length of journeys is treated as dependent both on previous deci-
sions of the worker and on the price of residential land adjacent to
workplaces.

The implied chain of decision-making is, therefore, uni-directional (or

‘recursive’) and of the form seen in Figure 3.1.
While Kain was clearly aware of the feedbacks from transport to land use,
for a variety of statistical and theoretical reasons he could not adequately reflect
them in his model. Besides not allowing for the longer-term feedbacks from travel
behavior to land use, the sequence takes no account of the influence of public
transport quality on the car ownership decision, or the length of journeys on the
longer-term provision of public transport. What the sequence does do, however,
is to permit Kain to examine the case for subsidizing public transport within the
current urban land-use framework. The assumptions of the sequential type of
framework used by Kain are analogous to the standard ceteris paribus assump-
tions of conventional partial equilibrium microeconomics. They also suffer from
the same limitations but do provide boundaries within which useful analysis can
be conducted.
Michael Wegener (2004) has suggested another way of looking at this issue
involving a focus on feedback cycles (Figure 3.2) and the recognition that trip
and location decisions co-determine each other. It is also very much the approach
used in many engineering-based and human-geography-derived development
theories. Essentially:

• The distribution of land uses, such as residential, industrial, or commercial,

over the urban area determines the locations of human activities including
living, working, shopping, education, and leisure.
• The distribution of human activities in space requires spatial interactions or
trips in the transport system to overcome distance between the locations of
activities.
• The distribution of infrastructure in the transport system creates opportu-
nities for spatial interactions, and can be measured by accessibility.
• The distribution of accessibility in space co-determines location decisions
and so results in changes in the land-use system.

Residential Mode of
Car
space transport Trip lengths
ownership
consumption used

Figure 3.1 Recursive view of location and travel

BUTTON 9781786435668 PRINT.indd 59 27/04/2022 11:07

 60 TRANSPORT ECONOMICS, 4TH EDITION

Transport system

Accessibility Activities

Land use
Figure 3.2 The land-use transport feedback cycle

Much of this book is concerned exclusively with the transport sector and with
short-run travel decisions. In general, it therefore implicitly assumes that the
causal link runs from land use to transport, and that land use is pre-determined.
In a way this may be regarded as a short-term view, land use being less flex-
ible than transport. It seems inappropriate, however, not to give some brief
overview of the approach adopted by spatial scientists. Not to do so would in
effect ignore the part played by transport in both shaping land-use patterns and
determining the size of the market areas served by various industries. In the
remainder of this chapter, therefore, we present a brief introductory outline of
modern location theory, concentrating primarily on that aspect of theory that
gives a central role to transport. The theory is supplemented by some discussion
of the applied work that offers quantification of the important role played by
transport in this field. Later in the book (in Chapter 13) the subject of trans-
port and location interaction is touched upon again in the context of economic
development.

3.3 Industrial Location and Transport

From the beginning, economic theories of industrial location argued that transport
plays a key role in decisions concerning where industrial activities will be located.
As we have seen, provision of good transport permits producers to be separated
from both their sources of new materials and their eventual customers. Alfred
Weber (1929) developed an early model for mobile plant with transport costs
determining the location of manufacturing industry. In Figure 3.3 all potential
customers are located at A1 while the two raw materials required by the manufac-
turing industry are located respectively at A2 and A3. The ‘d’s represent distances
between the points of raw material supply and final demand. It is assumed that all
other factors of production are freely available at all potential production sites and
that, topographically, all activities are located on a uniform plain.
Transport costs are assumed to be proportional with respect to both dis-
tances covered and weight of goods carried. The location of a manufactur-
ing plant will therefore depend on the relative pulls of the various material
locations and the market. The problem is then one of finding the site, Z, for

BUTTON 9781786435668 PRINT.indd 60 27/04/2022 11:07

 61
TRANSPORT AND LOCATION ­

manufacture which minimizes total costs, in other words the location which
minimizes T where:

T = a1rA + a2rA + a3rA  (3.1)

1 2 3

where:

a1 is the physical amount of the final goods consumed at A ;

1
a2 is the physical amount of the raw material available at A required to
2
produce a1 of the final good;
a3 is the physical amount of the raw material available at A required to
3
produce a1 of the final good; and
rA , rA , and rA are the respective distances from sites A , A , and A .
1 2 3 1 2 3

It is easily seen that if any two of a1, a2, and a3 are exceeded by the third,
then the location of production is determined at the site associated with this
third variable. For example, if a2 > (a1 + a3), then production of the final com-
modity should be at site A2. If no location is dominant, then graphical methods
can be used to find the least-cost site. A second triangle is drawn with its sides
(‘d’s) proportional to a1, a2, and a3 and the three angles measured. We denote
the angle opposite the a1 as α1, that opposite a2 as α2, and finally that opposite
a3 as α3. These angles then form the basis for erecting similar triangles around
the original location triangle (see Figure 3.3). Circles are drawn which touch the
points of each triangle and the optimal production site Z is then found at the
location where all three circles intersect. (If Z is found to be outside of the origi-
nal location triangle, then it is simple to prove that one of the corner solutions,
A1, A2, or A3 is preferable.) This location minimizes transport costs, as defined
in equation (3.1).

A1
2
d2
3
d1 Z
A3

d3
A2

1

Figure 3.3 Weber’s model of industrial location

BUTTON 9781786435668 PRINT.indd 61 27/04/2022 11:07

 62 TRANSPORT ECONOMICS, 4TH EDITION

This simple analysis implicitly assumes that transport costs are linearly related
to distance, but there is ample evidence that there are often considerable
­diseconomies associated with short hauls and with partially full loads. While
Weber originally suggested that one could adjust the sides of the location trian-
gle to capture this, the situation requires rather more complicated modifications.
The difficulty is that in these circumstances location and transport costs are co-
determined; without knowledge of the final location (that is, Z), it is impossible
to assess the magnitude, if any, of economies of long-haul transport. There is a
suggestion, however, that, other things being equal, tapered transport rates (that
is, when the rate per mile declines with distance) tend, in some circumstances,
to draw industry to either the source of raw materials or the market for final
products.
Suppose that only one raw material is needed to produce a final product, or
that the range of raw materials required is located at a single point, and this is
to be found at location A in Figure 3.4. Further, there is no loss of weight in the
manufacturing process required to produce the final product that will eventually
be sold at location M. Either the transport system offers a through service from
A to M costing $10 per ton or alternatively there are services from A to the inter-
mediate site B costing $6 per ton and one from B to M also costing $6 per ton.
Since we assume no weight loss in production the cost-conscious manufacturer
is clearly indifferent between A and M but would not select site B because
(AB + BA) > AM in cost terms.
To relax the assumption of no weight loss in manufacturing does not remove
the disadvantage suffered by B relative to A which is now preferable to location
M. (It is possible in this situation for B to be preferred to M if a location at A
is impracticable for technical or planning reasons, but this would depend upon
the relative importance of the taper vis-à-vis the weight loss in manufacture.)
Weight loss will only influence the choice between A and M, and this may even
be true if different rates are charged not simply by haul length but also by type
of commodity.
Even if it is more expensive per ton to carry the final product it may still be
preferable to locate production facilities at the material source rather than the
final market. In the United States, for example, the meat-packing industry of the
nineteenth century was gradually drawn westwards to the main source of beef
(initially to Chicago and then to Omaha and Kansas City) even though rail rates
per ton-mile favored transport of live animals rather than carcasses. The simple

$6 $6

A B M

$10

Figure 3.4 The effects of tapered transport rates on industrial location

BUTTON 9781786435668 PRINT.indd 62 27/04/2022 11:07

 TRANSPORT AND LOCATION ­ 63

fact was that the dressed meat of a steer weighs only 54 percent of the live animal
weight; hence locations near the raw material were favored.
Of course, firms do not in general locate in isolation from one another; there
are often agglomeration economies in locating near similar firms or suppliers.
August Lösch demonstrated that with all firms facing identical production and
transport cost schedules, and confronted by a spatially uniform total market, the
market would be divided out so that each firm would serve a hexagonal market
area. The equilibrium number of firms, and the area served by each, would be
determined by transport costs. The existence of several different but interdepend-
ent product markets would, in the Löschian model, tend to encourage concentra-
tions of firms at specific locations.
Discussion of the details of these location and other models is beyond the
scope of this volume, but it should be becoming increasingly apparent that trans-
port costs have increasingly been recognized to be only one of many factors which
influence industrial location. The factors influencing location include (in addi-
tion to transport considerations) market structure, demand elasticities, external
economies of geographical concentration, expectations of future market changes,
and processing costs. Greenhutt (1963), for example, suggests that in practice
transport costs only become of major importance if freight costs form a large
proportion of total costs or differ significantly among potential locations.
While theoretical models of industrial location offer useful insights into the
role transport plays, its actual relevance in the real world requires detailed empiri-
cal study. If Greenhutt’s argument is followed, in many cases transport costs are
such a small component of overall production costs that it appears to be more
costly to acquire the information necessary to find the least-cost location than
to suffer the inefficiencies of a suboptimal situation. For example, Cook’s (1967)
classic study of industrial location in the Midlands of England found that many
firms were ignorant or only partially aware of their transport costs. One can
attempt to isolate such transport cost-insensitive industries by looking at the rela-
tive importance of transport costs in their overall costs of production.
Table 3.2 offers some estimates of the percentage of the value of net output
for a variety of United States and Japanese industries attributable to transport
costs in the mid-1980s. There are clear differences between the countries, and
between the various industries within them. It seems reasonable to conclude
from this table that, ceteris paribus, industries producing primary products such
as agriculture, petroleum, and wood products are going to be more influenced in
their location choices by transport considerations than others such as machinery,
furniture, and precision instruments. Production of raw materials represent nodal
solutions in the Weberian triangle, with production at the raw material source. (In
the United States it has often been suggested that Gary, Indiana, was originally
chosen specifically by US Steel to minimize transport costs.)
Changes in industrial structure since the 1950s, and especially the move-
ment away from basic industries to manufactures and services, suggests that
transport is experiencing a diminishing influence over location decisions at least
at the inter-regional level. Further, the existence of comprehensive transport and

BUTTON 9781786435668 PRINT.indd 63 27/04/2022 11:07

 64 TRANSPORT ECONOMICS, 4TH EDITION

Table 3.2 Transport costs as a percentage of final consumer costs

Producer Transport costs (%)

United States Japan

Agricultural products 7.85 2.43

Food products 1.59 2.16
Textiles 0.58 6.33
Apparel & leather products 0.22 0.69
Paper & wood products 1.91 3.37
Furniture 0.53 2.03
Publishing 1.46 1.16
Chemical products 1.38 0.59
Petroleum products 2.60 1.56
Rubber & plastic products 1.79 0.99
Glass, clay, & stone products 1.36 4.50
Primary metal products 2.54 1.34
Finished metal products 0.93 0.75
Machinery 0.47 0.64
Electrical equipment 1.08 0.71
Transport vehicles 2.56 0.88
Precision instruments 0.15 0.90
Other 0.69 1.13
Total 1.70 1.83

Source: Adapted from Pirog and Lancioni (1997).

communications networks in virtually all industrialized countries means that

proximity to good transport is much easier to achieve now than in the past.
These factors, however, may not be as new as sometimes supposed. In early
empirical work by Gudgin (1978) looking at the 1968 United Kingdom Census
of Production, it was found ‘that almost three-quarters of British industry
incurs total transport costs at levels of less than three percent of the value of gross
output. In 95 percent of industry, by value of production, the transport costs are
less than 5 percent of total costs’. This finding was reaffirmed in Diamond and
Spence (1989) in a study of 190 manufacturing and service industries, also in
the United Kingdom, which found that most firms reported transport costs of
between 3 and 6 percent of production costs. Interestingly, at 9.9 percent trans-
port costs at the time represented a considerably higher proportion of costs in the
service sector than for manufacturing firms (4.7 percent).
Cost statistics can, however, give a slightly distorted impression of the
influence of transport. While transport costs may form only a relatively small
proportion of output costs in many sectors they may, nevertheless, have signifi-
cant influence on levels of net profits. Again, drawing on United Kingdom data,
Chisholm (1971), for instance, suggests that transport costs may have represented
as much as 25 percent of profits in the manufacturing industry during the 1960s.
Additionally, while transport costs may, on average, be low for some industries
they may vary considerably among areas. Edwards (1975) suggested that there

BUTTON 9781786435668 PRINT.indd 64 27/04/2022 11:07

 TRANSPORT AND LOCATION ­ 65

existed a range of about 20 percent in transport costs of manufacturing industry

by region in 1963.
It should also be remembered that simple cost estimates may disguise varia-
tions in other attributes of transport (speed, regularity, etc.) which can influence
decision-makers. Reliable inter-urban transport, good international transport
links, and high-quality local transport (necessary to retain scarce skilled labor)
are, for instance, often very important for modern high-technology industry. In
an analysis of the largest United States commercial airports, for example, Button
et al. (1999) found they attract considerable amounts of high-technology employ-
ment, some 12,000 such jobs compared to cities without these facilities.
Survey evidence, questioning industrialists about the motivations underly-
ing their location or relocation decisions also provides some guidance to the
importance of transport considerations. There are obvious difficulties in using
such results – for example, the sample may be unrepresentative, respondents offer
answers which they hope may further their individual interests, while others offer
ex post rationalizations of their actions – but some information may be gleaned
from them.
The UK Trade and Industry Sub-committee of the House of Commons
Expenditure Committee in 1973, for example, when seeking information about
the effectiveness of regional economic policies, was told by five of the 17 major
industrial firms interviewed that high transport costs were a specific disadvantage
for locations in developed areas. While accepting this as an objective comment,
it is important to note not simply the relatively small number of firms concerned
but also their nature – for example, three were car manufacturers and one was a
large steel tube mill. These were large firms engaged in the production of bulky
products whose per unit transport costs were likely to be high.
The United Kingdom’s Armitage Committee in its 1980 Inquiry into Lorries,
People and the Environment supported this line and concluded that ‘[w]hen
industry and commerce make decisions about the location of factories or their
systems of distribution, it is often less important to reduce transport costs than
to reduce other costs such as those of stockholding or to take advantage of the
grants for setting up factories in assisted areas’. Such a view is consistent with
that of Cameron and Clark (1966), who in their study of 71 firms which located
in assisted areas found that accessibility to main markets was only ranked third
as a location factor (behind the availability of trained labor and local authority
cooperation), while local goods transport facilities were listed fifth and accessibil-
ity to main suppliers sixth.
One factor which emerges from these studies, and has limited support in
econometric work, is that during the post-war phase of full employment and
strengthened land-use controls, access to markets and raw material supplies was
often overshadowed in location decisions by the availability of scarce skilled labor
and factory space. Historically, international confirmation of this position was
seen in West Germany where a survey of newly located plants found that acces-
sibility to motorways ranked only fourth in the list of location criteria (Fischer,
1971).

BUTTON 9781786435668 PRINT.indd 65 27/04/2022 11:07

 66 TRANSPORT ECONOMICS, 4TH EDITION

More localized studies suggest that good passenger transport facilities may
influence industrial location rather more than the quality of long-distance or
freight transport. In retailing and some other activities this argument is further
extended to embrace firms’ desire to be accessible to customers. The importance
of local transport is, for instance, seen in a study of factors affecting location
choices of high-technology firms in Pennsylvania. Allen and Robertson (1983)
found that while proximity to market, proximity to family, and commuting dis-
tance came first, second, and fourth respectively in terms of factors influencing
firms, regional surface transport and proximity to an airport came only thirteenth
and sixteenth respectively. This confirms the emphasis that has been placed upon
the ready availability of trained workers.
It is certainly clear from a number of surveys – for example, Table 3.3 reports
the results of one United States overview focusing on 162 studies – that the availa-
bility of suitable labor is a key factor in location choices of both high-technology
and more traditional industries. There is also evidence that local transport can
be influential in attracting or retaining labor, although the relationships can be
complex and industry specific (Zandiatashbar and Hamidi, 2021).
Transport and transport-related considerations would seem, therefore, to
still be very relevant in the location decisions of modern industries although the
specific transport attributes of importance may have changed somewhat over the
years.
In addition to the appreciation that transport is often not the dominant
factor in location choice, there is now the increasing view amongst behavioral
economists that notions of cost minimization do not always motivate firms.
Hence, even if one can isolate the factors of interest to firms, it is not certain that
they should be treated within a cost-minimizing framework. In many cases, pro-
vided a site offers, ceteris paribus, a location where transport costs are below some
threshold, it is considered acceptable. In other cases, the first acceptable location
encountered is adopted rather than a protracted search pursued. The influence
of social setting and amenities on those who make the decisions about location,

Table 3.3 Factors influencing the location of branch plants in the United States

Rank High-technology Other plants

1 Labor Labor
2 Transport availability Market access
3 Quality of life Transport availability
4 Market access Materials access
5 Utilities Utilities
6 Site characteristics Regulatory practice
7 Communications characteristics Quality of life
8 Business climate Business climate
9 Taxes Site characteristics
10 Development organizations Taxes

Source: Adapted from Stafford (1985).

BUTTON 9781786435668 PRINT.indd 66 27/04/2022 11:07

 TRANSPORT AND LOCATION ­ 67

and on the staffs whose preferences they need to consider, were highlighted in
this context many years ago by Eversley (1965). Firms often, therefore, adopt
‘satisficing’ policies in site selection rather than attempting to profit- or revenue-
maximize or to cost-minimize. Under these conditions the exact role played by
transport costs becomes almost impossible to define. It seems likely, though, that
once a location has been chosen, a major rise in transport costs would be neces-
sary to overcome the basic inertia that seems to accompany such a managerial
objective.

3.4 Economic Gateways and Corridors

Economic activities clearly require the movement of resources: factors to points

of production and goods and services to markets, and even in the information age,
ideas and data from the generator to the user. The distances involved in this have
historically been increasing, and the making of any final produce may now require
many thousands of miles of transport input. Distance per se, however, is not
always (and has perhaps seldom) been the crucial element that influences the rela-
tive success of competitive production sites; other things being constant, factors
such as time, frequency, reliability, and security are often more important. In this
context appropriate transport is relevant. This, because of the attributes of trans-
port and the costs of delivering them, often leads to a channeling of movements
through gateways and along corridors. This is not in itself the issue. The challenge
concerns the nature and number of these gateways and corridors and their form.
As with any economic resource, transport services suffer from scarcity; they
are not ubiquitous, and, indeed, allocating resources to them inevitably involves
taking resources from elsewhere. In the past this scarcity, and with it an oppor-
tunity cost, was often neglected in trade theory and, by association, economic
development theory. Classical Ricardian economics of the nineteenth century,
for example, focused exclusively on the comparative advantages in production at
different locations, while in classical spatial economics, albeit at a more local level,
it was assumed there existed infinite radial transport links within a concentric
economic geographical space (a topic we implicitly look at in Section 3.6). But
transport supply is constrained by its characteristics.
Transport is a network industry and it is, therefore, natural to think in terms
of the role that transport may play in stimulating economic development along
links in transport networks and at various nodal points; normally towns and
cities. Historians have long viewed the trade and migration passages that existed
in prehistoric times as important for the spread of civilization as it emerged,
and subsequent trade routes as facilitating economic progress and disseminat-
ing knowledge. Those living in early commercial cities benefitted from the flows
passing through their areas when they could exercise control over them and
charge for passage.
The much more recent opening-up of North America has attracted a lot
of attention, in part one suspects because of the availability of written records.

BUTTON 9781786435668 PRINT.indd 67 27/04/2022 11:07

 68 TRANSPORT ECONOMICS, 4TH EDITION

Whebell (1969) explained the growth of southern Ontario, Canada, in terms of

a five-stage corridor development process occurring over an extended period and
embracing significant technical changes. In contrast Scholvin at al. (2019) focus
on gateway cities, distinguishing between them and hubs and examining the ways
in which they change over time. Gateways are explained in economic terms by
considering threshold values of distance and levels of productivity.
Discussions of gateways and corridors, together with their role in economic
development, are not new. They aroused interest in the 1980s and 1990s with the
onset of the ‘hi-tech boom’. Indeed, many high-technology concentrations were
designated either explicitly or implicitly by their corridor geography – Silicon
Valley, the Route 128 Corridor, and the Dulles Corridor in the US, the M4
Corridor in the UK, etc. – while others, such as the Research Triangle in the US,
were visualized in terms of having a gateway role. These concentrations highlight
the synergies between transport and certain types of capital investment, but are
more local in their orientation than the macro-corridors and gateways that are
seen as linking regions or countries to wider markets.
Gateways have tended gradually to move farther apart as it has become
easier for traffic and individuals to both pass through them and as transport
systems have evolved to traverse the distance between them. Figure 3.5 represents
a traditional view of gateways. At the national level in most countries there are
one or more major hub cities that are linked to their borders by corridors that end
at gateway cities offering links to the international market. In spatial economic
terms, the main distinction between the hub city and the gateway city is that,
while the former fulfils the classic role of serving a concentric hinterland, the
gateway city services a cone-shaped market extending away from the border and
along the corridor.
Corridors are essentially links between major nodes; in some ways they can
be treated as ‘super-links’ and this is seen as distinguishing them from spokes in
a hub-and-spoke network, a standard pattern of transport supply discussed in

Gateway City Hub City Gateway City

Figure 3.5 Notions of transport gateways, corridors, and hubs

BUTTON 9781786435668 PRINT.indd 68 27/04/2022 11:07

 TRANSPORT AND LOCATION ­ 69

Chapter 6. A major difficulty, however, involves the level of aggregation that one
is dealing with.
At the global-historic level, macro-macro level corridors are defined as routes
that mankind used to populated the world – for example, the ‘Bering land bridge’
allowing migration from Europe to the American continent. But in the modern,
high-technology, information age a corridor may be an electronic channel over
which a piece of information is sent. The term suffers from almost infinite vague-
ness. This is not very helpful when it comes to in-depth analysis or forecasting but
it does have its uses in general assessments of trends and can serve as a focal point
of terminology when policy-makers want to coordinate actions, as for example in
the creation of trade corridors.
It is also relatively easy to relate the picture seen in Figure 3.5 to specific con-
texts in more recent history. In the United States, for example, the two gateway
cities were often seen in the mid-1800s as New York on one coast and San
Francisco on the other. Once into the country, goods or migrants could move into
the hinterland, often dispersing more broadly through a hub such as Chicago.
Railroads largely facilitated this movement. The nature of international maritime
and domestic railroad transport at the time, as well as institutional controls, led to
this pattern of behavior. The gateways proved challenging barriers to cross and,
while trade and migration was extensive, it was not easy and it was costly, and
reverse migration and visits to family left behind proved almost impossible for the
many individuals even if they did succeed in their new land.
The pattern of Canada’s largest freight railroad, the Canadian National
route network (Figure 3.6), provides a classic representation of the form that in
theory a gateway/corridor structure looks like, and it is perhaps no accident that
much of the early analytical writings on the subject came from Canada. The mari-
time gateways on the two coasts, and the inland crossing gateways to the United
States funnel goods and, more in the past, individuals to and from the major hub
cities of the country – Toronto, Montreal, etc. Similar patterns emerge for the
more recent road network.
The world is changing, and transport has been both a cause of this change,
but, mainly because of the derived nature of the demand for its services, has
also had to react to it. These changes have implications for the demands that are
placed on mobility of both people and goods. They have produced significant
changes in the amount of geographical specialization and in both internal and
external trade. The traditional barriers to trade, most obviously the physical geo-
graphical ones associated with oceans, mountains, rivers, and distance, but also
the institutional ones tied to tariff and non-tariff barriers and fragmented finan-
cial markets, as well as the social ones of cultural and linguistic diversity, have had
their potency reduced.
Transport costs for both passengers and freight have fallen considerably
since the 1980s and 1990s. This is, in part, a function of technology improve-
ments, including those found in complementary sectors such as telecommunica-
tions and warehousing, but it also stems from institutional developments and
especially the liberalization of many transport markets. Economies of scale and

BUTTON 9781786435668 PRINT.indd 69 27/04/2022 11:07

 70 TRANSPORT ECONOMICS, 4TH EDITION

Figure 3.6 Canadian National’s railroad network

scope have come with the freeing of international trade more generally, and the
adoption of innovative methods of supplying logistics services of all types by the
private sector.
The increased volume of trade, both within and between countries, and
changes in its nature, have led to the emergence of new gateways and corridors
to handle it. The development of the inter-modal terminal port at Prince Rupert
in western Canada seen in Figure 3.6 is an example of this. The amount of com-
petition between various gateway/corridor combinations, and their importance
in economic development, even those involving existing combinations, have
also changed for a variety of reasons. The full picture of what has happened is
complex and is still not fully documented or agreed upon. It is a topic returned to
in a slightly different way when the links between transport and development are
discussed in Chapter 13.

3.5 Output, Market Areas, and Transport Costs

Transport costs are not only instrumental in influencing where firms locate, but
they also play an important role in determining the market area served by each
firm. Transport costs, given the place of industrial location, can determine the
total quantity of goods sold and the price and the spatial distribution of this
output. Lösch conducted much of the early work looking at market areas, but
here we focus on a specific transport-orientated model that was devised by van Es
and Ruijgrok (1974). The simple model treats transport demand as derived from
the demand for the final product and assumes all supply and demand curves to
be linear. For expositional ease, the relevant functions are treated in a manner
running counter to economic convention; specifically, price is treated as depend-
ent upon demand rather than vice versa. Initially our firm, which produces a

BUTTON 9781786435668 PRINT.indd 70 27/04/2022 11:07

 TRANSPORT AND LOCATION ­ 71

homogeneous product, supplies a single customer who is located some distance

from its pre-determined site. Hence, we have:

Ps = a0 + a1Qs + Pt (3.2a)

Pd = b0 – b1Qd (3.2b)

Qd = Qs (3.2c)

Ps = Pd (3.2d)

where:

Ps is the supply price of the commodity;

Pd is the demand price of the commodity;
Qs is the quantity of the commodity supplied;
Qd is the quantity of the commodity demanded; and
Pt is a constant transport cost per unit carried to the customer and treated as
a cost borne by the supplier.

Manipulation and combination of these equations yields the profit-­

maximizing supply, Qe, that is:
 b − a   Pt 
Qe =  0 o  −   (3.3)
 a1 + b1   a1 + b1 

It is immediately clear that transport costs exert a negative influence on the

quantity the profit-maximizing firm ought to supply, that is, if Pt = 0, then the
equilibrium output would rise by (Pt)/(b1 + a1). Further, we can derive the equi-
librium price (Pe) that should be charged to the customer:
 a b − a0 b1   b1P t 
Pc =  1 0 +  (3.4)
 a1 + b1   a1 + b1 

Here we see that the transport cost component increases the equilibrium
price by [(b1Pt)/(a1 + b1)]. The effects of this, together with the effects of transport
costs on Qe, are illustrated graphically in Figure 3.7. The vertical axis shows the
final price per unit paid by the customer and the horizontal axis the quantity of
goods sold. The introduction of the transport cost element to the diagram has the
effect of pushing the supply curve up from Ps(Pt = 0) to Ps(Pt > 0). It is evident
that transport cost rises will push up final prices and reduce the quantity sold.
The exact impact depends not only upon the magnitude of Pt but also the elastici-
ties of supply and demand – greater inelasticity increases the influence on price
exerted by transport cost considerations.
To estimate the market area served when potential customers are spread
evenly around the production site it is initially assumed that identical individuals

BUTTON 9781786435668 PRINT.indd 71 27/04/2022 11:07

 72 TRANSPORT ECONOMICS, 4TH EDITION

P
Ps(Pt > 0)

Ps(Pt = 0)

b1Pt/(a1 + b1)

Pt

Pd

0 Q
Pt/(a1 + b1)
Figure 3.7 The effect of transport costs on price and output

are located at equal distances along a straight road from the site of the s­ upplier.
The customers will be confronted by prices that are composed of a fixed factory
price reflecting production costs and a variable transport cost dependent
upon the distance they live from the production site. Since each customer – by
­assumption – exhibits a similar demand response it is, therefore, the transport
component that determines the amount each will buy. At the edge of the firm’s
market area, the amount supplied to the marginal customer vanishes to zero;
this will be when Pt = (b0 – a0). If j customers are served before this limit is
reached, then from equation (3.3) we can see that the sales of the firm (QT) will
amount to:

 b − a0   e  1 
QT = ∑Q ej = j  0  −  ∑Pj    (3.5)
j  a1 + b1   j   a1 + b1 

where Qe represents sales to customer j.

This approach can be extended to show the entire geographical area served
by the firm. In Figure 3.8 the vertical axis represents the quantity supplied to each
customer, on the assumption that the customers are evenly spread over the plane.
The amount sold to a customer falls from very high levels Qe = [(b0 – a0)/(b1 + a1)]
immediately adjacent to the site of supply – where transport costs are zero – and
falls to zero when transport costs become excessive.
The amount sold can be measured by calculating the volume of the cone,
that is:

D = bπ∫ 0 ( P + T )TdT
R
(3.6)

BUTTON 9781786435668 PRINT.indd 72 27/04/2022 11:07

 73
TRANSPORT AND LOCATION ­

Qe = (b0 – a0)/(b1 + a1)

Pt = 0 Qe = 0

Pt = b0 – a0
Figure 3.8 The influence of transport costs on market area

where:

D is demand as a function of the free on board (FOB) price net of mill price P;
b is twice the population density of a square in which it costs one money unit
to ship one unit of the commodity along one side;
d = f (P + T) is an individual demand as a function of the price of the com-
modity at the place of consumption;
P is the FOB net mill price of the commodity;
T is freight cost per unit from the factory to the consumer; and
R is the maximum possible transport cost.

While this type of approach (relating the sales, prices, and market area of
production to transport costs) is obvious theoretically, interestingly it does rely
upon many abstractions from reality for its internal consistency. As with many
theories, some of the assumptions associated with the model outline above
may be relaxed: it is possible, for example, to allow for variations in population
density, for heterogeneity in consumer tastes, and for non-perfectly elastic supply
­conditions – but this does tend to add complexity to the analysis. The general
impression conveyed, however, is always the same, namely that transport costs
are key in determining the size of the geographical market served by a firm and
the total volume of its sales. Further, one situation where the effects of transport
improvements may have a magnified effect on the market area is when the the
producing industry is capable of exploiting manufacturing scale economies. This
is not a novel idea and, indeed, it was recognized over two centuries ago by the
classical economist Adam Smith in his Wealth of Nations.

BUTTON 9781786435668 PRINT.indd 73 27/04/2022 11:07

 74 TRANSPORT ECONOMICS, 4TH EDITION

3.6 Urban Transport and Land Values

The previous sections have been concerned with the interaction between transport
and the physical, spatial economy, but little has been said regarding the way in
which transport quality can affect land values. Location, to date, has tended to be
viewed in terms of where a firm would locate and the area that would be served.
Little has been said about the distribution of land among alternative uses either
within one sector, such as manufacturing, or among sectors, notably between
industrial and residential use. Very early work on this problem can be traced back
to Johann von Thünen’s century-old land-rent model which attempted to explain
differences in agricultural land rent. He argued that concentric zones of crop spe-
cialization would develop around the central market, the key feature of the model
being that land-rent differentials over homogeneous space are determined entirely
by transport cost savings. While the nineteenth-century agrarian economy pro-
vided the inspiration for the ‘bid-rent curve’ (sometimes, ‘bid–rent curve’ is used)
analysis, it is in the context of twentieth-century urban development that it has
been most fully developed.

The Urban Bid-Rent Curve

Robert Haig in the 1920s was the first to apply von Thünen’s argument in the
urban context that:

[s]ite rents and transportation costs are vitally connected through their relationship
to the friction of space. Transportation is the means of reducing that friction, at the
cost of time and money. Site rentals are charges which can be made for sites where
accessibility may be had with comparatively low transportation costs.

People who are prepared to pay the highest price for improved transport provi-
sion (that is, to outbid rivals) will enjoy the most accessible locations.
This approach is clearly dependent upon some very stringent assumptions
that need to be spelt out before we proceed further. We focus initially upon the
residential location of households. The city under review is seen as a featureless
plain with all production, recreational, and retailing activities concentrated at a
single urban core (the central business district, or CBD). The population is homo-
geneous with respect to family size, income, housing demands, etc., but while
building costs are largely unaffected by location, transport costs rise with distance
from the CBD. With these assumptions, the sum of transport costs plus site rents
is constant across the entire city. That is, if we take a ray out from the CBD to the
perimeter of the city and concentrate exclusively on household decisions, we have
the situation depicted in Figure 3.9. Total site rent in the city may be estimated as
the volume of the inverted cone centered on the CBD.
An improvement in the transport system will result in a fall in land values at
each location and an outward expansion of the city, the extent of this outward
expansion being dependent upon the elasticity of demand for transport services.

BUTTON 9781786435668 PRINT.indd 74 27/04/2022 11:07

 75
TRANSPORT AND LOCATION ­

Site rent
Annual transport
expenditure per family
at city perimeter
Transport costs

0 Distance from CBD

Figure 3.9 The site rent/transport cost trade-off

If the demand for transport is perfectly inelastic then the boundary of the city
remains unchanged. If AB in Figure 3.10 is the initial rent gradient then, with
a perfectly inelastic demand for transport services, the result of, say, uniformly
lower public transport fares is to shift the gradient to A'B. The city’s perimeter
remains at B.
With a degree of elasticity, however, the reduced transport cost will encour-
age longer-distance travel to work and recreational activities and the eventual rent
gradient is likely to settle at a position such as A''B'' and the city’s boundary to
extend out to B''. It should be remembered that this simple model assumes that
transport costs vary linearly with distance from the CBD and that individuals

Rent ($)
A

A"

A'

0 B B"

Distance from CBD

Figure 3.10 Effects of transport changes on the bid-rent curve

BUTTON 9781786435668 PRINT.indd 75 27/04/2022 11:07

 76 TRANSPORT ECONOMICS, 4TH EDITION

are identical. If this is not the case, then, as Herb Mohring (1961) has pointed
out, the precise relation between transport cost and location patterns would be
obtained simultaneously rather than sequentially as above. Further, the change in
relative site rentals may also result in households wishing to own different sized
plots of land; this complication is incorporated by allowing for some elasticity in
the demand for quantities of land.
William Alonso greatly refined the model of urban location by extending
beyond a simple consideration of residential land rent. The priorities of house-
holds differ and there is clearly competition in a free market land economy among
the demands of industry, commerce, and various classes of households for differ-
ent sites. In general, there are so-called ‘agglomeration economies’ to be enjoyed
by industry and commerce from locating both close to each other and at the city
core – they present an identifiable geographical entity (for example, medicine in
Harley Street and bespoke tailoring in Savile Row in London, and Wall Street
in New York), can be easily served by specialized suppliers, and can provide
customers with a comprehensive range of services, etc. Consequently, given the
potentially higher revenue associated with a core location, business tends to bid
highly for central sites. Poorer people who cannot afford high transport fares
and place a relatively low priority on large sites are willing to bid higher rents for
inner area locations, while the wealthy will be more inclined to bid a higher rent
for suburban locations.
Figure 3.11 shows the bid-rent curves for three groups of urban land user:
business, poor households, and wealthy households. We see, in this very simple
example, that business will outbid both classes of household for sites near the
urban center, poorer households will locate adjacent to the CBD, while the
wealthy will outbid the other groups for sites at the edge of the city. Clearly this is
rather a stylized picture of urban land use and rents (Chapter 13 provides some
further refinements), but it does offer some insight into the influence transport

Low-income
bid curve

Middle-income
bid curve
High-income
bid curve

CBD
Distance
Low- Middle-income High-income
income
Figure 3.11 Bid-rent curves for various income households

BUTTON 9781786435668 PRINT.indd 76 27/04/2022 11:07

 TRANSPORT AND LOCATION ­77

can have on intra-urban location patterns. Quite simply, high transport cost
activities, ceteris paribus, will be located at a short distance from the CBD and low
transport cost activities will take locations further away.
General support for this pattern of land use in the United States is seen
in Table 3.4, which shows that the spatial concentration of poverty increased
steadily in the three decades to 1998, with over 25 percent of core metropolitan
areas’ populations being made up of the country’s poorest 20 percent by the end
of the period. But there has also been a continual overlapping suburbanization
process, with 61 percent of the United States living in suburbs by 1990 compared
to 36 percent in 1948. The advent of cheap motoring was a major cause of this.
Changes in transport costs may affect the amount of land taken up by
various activities, and some of the basic implications can be illustrated using the
bid-rent curve.

Transport Policies and Urban Bid-Rent Curves

Figure 3.12 replicates the curves seen in Figure 3.11 with 0 ⇒ P being the section
around the CBD occupied by poorer people, P ⇒ M that occupied by middle-
income people, and M ⇒ W that occupied by the wealthy. The introduction of
improved public transport with low fares, for example, may well increase the
willingness of poorer households to bid for locations further from the CBD
and consequently the central ring of land use, as we see in the top section of
Figure 3.12, may both widen and, more probably, move outwards to say P*. As
an example, the construction of the Washington Metro, with its relatively cheap,
good-quality service, increased the willingness of people to pay for land parcels

Table 3.4 Share of United States central-city population by income class and year

Year All metropolitan statistical areas/primary Central cities (%) Suburbs (%)
metropolitan statistical areas (%)

Low income (national lowest 20%)

1969 18.3 21.9 14.8
1979 18.5 23.7 14.5
1989 18.1 24.0 14.1
1998 19.0 25.5 14.9
Middle income (national middle 60%)
1969 59.4 59.8 59.1
1979 59.4 59.0 59.8
1989 59.4 58.8 59.8
1998 58.8 57.9 59.3
High income (national top 20%)
1969 22.3 18.3 26.2
1979 22.1 17.3 25.7
1989 22.5 7.2 26.1
1998 22.3 16.6 25.8

Source: US Department of Housing and Urban Development (2000).

BUTTON 9781786435668 PRINT.indd 77 27/04/2022 11:07

 78 TRANSPORT ECONOMICS, 4TH EDITION

Rent ($)

0 P P* M W
Distance from CBD
Rent ($)

0 P M W W*
Distance from CBD
Rent ($)
Pedestrian zone

0 P P* M W
Distance from CBD

Figure 3.12 The simple impacts of changing transport provision on urban land use

near metro stations. Light rail can have a similar effect within urban areas as seen
in Houston (Pan, 2019). Although not shown, the effect may well also produce a
growth in economic activity at the CBD. In the urban center, retailers were even
more willing to offer high rents for sites near metro services.

BUTTON 9781786435668 PRINT.indd 78 27/04/2022 11:07

 TRANSPORT AND LOCATION ­ 79

Alternatively, the construction of an extensive urban/suburban freeway

network is likely to improve the access offered to car-owning, wealthy house-
holds, shifting their bid-rent curve upwards and to the right (the center section of
Figure 3.12). The ring of land occupied by this group will thus be extended out
beyond the original city boundaries (that is, from M ⇒ W to M ⇒ W*). Work in
the United Kingdom on its motorway-building programs of the 1960s by Evans
(1973), for instance, specifically attributes the ‘flight to the suburbs’ to ‘the large-
scale construction of urban motorways which increases commuting speeds and
comfort for the commuter’ and argues that the only way to get higher-income
groups back into the central area is to ‘[m]ake the transport system slow, cheap
and uncomfortable, and not fast, comfortable and expensive’.
Another possibility is that land-use planning may instigate a pedestrian-only
zone around the CBD as seen in the bottom section of Figure 3.12. This will
make it cheaper and more convenient for poorer households to move around and
thus locations about the CBD more attractive places for them to live. Although
the exact effect of such a policy depends upon its detailed design, in the bottom
section of Figure 3.12 it is assumed that the result is that the belt of lower-income
households expands out to P*.
Of course, reality is a little more complicated than these examples suggest; for
example, we have treated the city in isolation and as an abstraction, ignored vari-
ations in the geography of the area, assumed away any land-use planning agency,
and treated location change as instantaneous. We have also ignored important
feedback effects including the ability to change times of travel and routes, and to
use modes of transport differently. The aim however, is to make it apparent that
any attempt at formulating an urban transport policy, or any advance in transport
engineering, will have important implications for urban form and on the people
that live in the city even physically distant from the actual policy initiative.

Exhibit   The impact of parking policy on house prices in the Netherlands

Paid parking was initiated in 1935 in the United States in Oklahoma and is now widespread
where there is limited off-street parking. Excess demand for parking leads to cruising as
residents and visitors need to search for an available parking space. Appropriate parking
prices reduce the number of parking places being demanded, and have the effect of reducing
cruising for parking and associated traffic congestion costs.
The 1970s and 1980s saw car ownership increase in Amsterdam and Utrecht in the
Netherlands with parking fees being used to ration spaces; see the table. The Mulder Act of
1989, allowing municipalities to extract revenues from traffic violations, including parking,
witnessed stricter parking policy with higher parking tariffs and tighter enforcement. With
parking discouraged in Amsterdam’s center, drivers parked in surrounding areas leading
to the introduction of paid parking in these areas. A similar development of paid parking
occurred in Utrecht, which witnessed a large increase of its paid-parking area in the
mid-1990s.

BUTTON 9781786435668 PRINT.indd 79 27/04/2022 11:07

 80 TRANSPORT ECONOMICS, 4TH EDITION

In Dutch cities, residents receive parking permits when paid parking is introduced to
increase their political support. The implications of this policy on residents was assessed by
examining the effect of its introduction on house prices for Amsterdam and Utrecht over
a period of 30 years. A hedonic price index was used to allow for the effects of parking
policies on house values making allowances for closeness to the city, size of property,
whether the property was insulated, the attribute of a garden, etc.
As seen in the model results presented in abbreviated form, the introduction of paid parking
(the first variable) had no significant effect on house prices in either Amsterdam or Utrecht.
This finding is consistent with the idea that residents only vote in favor of a local policy
when it has no negative impact on their house prices.

Full sample Implementation Amsterdam Utrecht

areas

Paid parking 0.039*** –0.011 0.009 0.003

Paid parking × private parking 0.026** –0.016 –0.026 –0.004
Outdoor parking 0.029*** 0.064*** 0.073*** 0.086***
Garage parking 0.060*** 0.093*** 0.103*** 0.087***
Log size 0.770*** 0.774*** 0.853*** 0.839***
Log size distance to city center –0.005 –0.004 –0.024 –0.185***
Log size × distance to city center² –0.003 0.002 0.001 –0.003

Notes: Variables statistically significant as * 10 percent, ** 5 percent, and *** 1 percent levels. Additional control
variables affecting house values, such as garden, central heating, insulation, year of construction, distance from city
center, zip code, were included.

See also: J. de Groote, J. van Ommeren, and H.R.A. Koster (2018) The impact of parking
policy on house prices, Journal of Transport Economics and Policy, 52(3), 267–82.

3.7 Urban Wages

Transport costs not only influence urban land-use patterns, but they are also
instrumental in determining spatial variations in urban wage rates. As Leon
Moses (1961) pointed out, ‘the wage differential, positive or negative, a worker is
willing to accept is completely determined by the structure of money transport
costs’. For simplicity, we assume that all employment is either concentrated at the
city center or else spread evenly over the surrounding, mainly residential, area. All
households are assumed initially to be in equilibrium, all enjoying the same level
of welfare. Moreover, all workers are paid identical wages, work the same hours,
and undertake the same number of commuting trips. Initially, then, net monies
after work-trip outlays will vary among workers according to the nature of their
intra-urban transport costs and the distance of their homes from the CBD.
Variations in land values with distance from CBD, as we saw in the previous
section, act as an adjustment mechanism to ensure uniformity of welfare. People
living away from the urban core will pay more in transport costs, but their land
rentals will be correspondingly lower.

BUTTON 9781786435668 PRINT.indd 80 27/04/2022 11:07

 TRANSPORT AND LOCATION ­ 81

In this simple world, however, a worker could improve their well-being by

giving up their job at the CBD (and thus saving commuting costs) and working
at one of the jobs which are spread evenly over the urban area and which are
near their home. They would be willing to accept a lower income in this situation;
indeed, they would be willing to sacrifice their wage rate down to the point where
it is cut by as much as the commuting costs that are saved. Thus, there will be an
equilibrium wage at locations nearer home than the CBD that will be lower than
the core wage.
The result of this type of approach is the development of an urban wage gra-
dient, the shape of which is determined by commuting cost factors. In Figure 3.13
the downward-sloping curve then represents the fall-off in wages from W at the
CBD needed at each location to be paid to a worker as employment moves from
the urban center. This decline represents the decreasing amounts of compensa-
tion required to be paid by an employer to attract workers who have easier com-
mutes because they do not have to go to the CBD. The cost of enticing a local
worker to a job at, say, location A is only W* and well below that at the CBD.
The worker living at A is indifferent between commuting to the CBD and earning
0W and working at home and earning 0W*. The commuter wage gradient curve
traces out the wage rates at which a worker living at A would have to be paid to
make them indifferent between working at home and at any intermediate site
between A and the CBD. (It also shows the wage rates at which people living
between A and the CBD will be indifferent between working at home and at the
core.) The exact shape of these curves varies according to the types of jobs avail-
able and the quality and qulaifications of workers as well as the nature of the
transport system (Autor, 2019).
If we introduce the notion of some secondary employment concentra-
tion, say at L, then reverse commuting may develop, especially if public poli-
cies support it (Cervero and Tsai, 2003). If employers at this subcenter require

Wages

Reverse commuting
wage gradient

Commuting wage
W* gradient

CBD B L A
Distance from the CBD

Figure 3.13 Transport and wage costs

BUTTON 9781786435668 PRINT.indd 81 27/04/2022 11:07

 82 TRANSPORT ECONOMICS, 4TH EDITION

amounts of labor that they cannot attract from households to the right of L they
will need to compensate workers who travel out from areas between L and the
CBD. Since the transport system tends to be less costly, in overall terms of money,
time, and comfort factors (see Chapter 5 for more detail) for reverse commuting,
the reverse commuting wage gradient is likely to be less steep than the commuting
wage gradient. In this case, they can obtain the workers that they need by taking
them from B => L as well as to the right of L, with wages being determined by the
reverse commuting wage curve. The main influence on reverse commuting costs
vis-à-vis those for CBD commuters is the generally lower levels of traffic conges-
tion away from the city center.
Just how does the wage gradient theory stand up to empirical investigation?
There is remarkably little work on this question, but we can pull together some
insights. Quite clearly the spread of national wage agreements, combined with
imperfections in the land and transport markets (for example, public transport
subsidies, the existence of company cars, and incomplete road networks), makes
any exact testing of the theory difficult. American evidence from New York
in the 1960s found wages to be higher in suburban counties than at the CBD
and, even after allowing for variations in industrial structure, no wage gradient
emerged (Segal, 1960). However, a little later, Rees and Shultz (1970) in their
study of Chicago found a ‘strong positive association of wages with distance
traveled to work’, but the wage gradient for blue-collar workers had its peak in
the area of heavy industrial concentration (the south-west of the city) and sloped
downwards towards the north-west. Similarly, analysis of the Panel Survey
of Income Dynamics (basically looking at the activities of the same group of
individuals over time) showed that ‘[u]rban wage gradients exist in America’
(Madden , 1985).
More recently, Darren Timothy and William Wheaton (2001) have provided
further confirmation of the trade-off between commuting times and wages in the
United States, taking account of the larger shifts that are taking place in urban
form, notably suburbanization. They used a very simple model:

Wi – W 0 = F2(C1),  F'2 > 0 (3.7a)

Ci = F1(Ei),   F'1 > 0  (3.7b)

where:

Ci is the average commuting cost of workers in employment zone i;

Ei is the employment in zone i;
Wi is the wage rate in zone i; and
W0 is the base metropolitan wage rate.

Using data for two large metropolitan areas, after adjusting for types of employ-
ment and such, they found that much of the 15 percent variation in urban wages
could be related to commuting time differences between locations.

BUTTON 9781786435668 PRINT.indd 82 27/04/2022 11:07

 TRANSPORT AND LOCATION ­ 83

In the United Kingdom, Evans (1973) suggested that while a wage gradient
could be discerned for Greater London there was little conclusive evidence of one
elsewhere. Indeed, in provincial cities, clerical wages were often almost uniform
across metropolitan areas. Clerical wages were, however, found to be higher in the
City and the West End of London than in the suburbs.
Although these findings are not completely conclusive, the evidence sup-
porting the wage gradient theory is at least balanced by that rejecting it. The
problems are that the studies to date fail to allow for the multiplicity of factors
other than transport that affect wage levels. We have already mentioned some of
the ­problems – notably imperfections in certain markets – but to these we must
add the tendency for employers to compensate workers for high transport costs
by unrecorded payments (free meals, shorter working hours, more flexible time-
keeping, etc.). Additionally, in some cities there are unrecorded advantages of
working in the city center (better out-of-work and shopping facilities, increased
career potential, etc.). The studies do not, therefore, refute the idea that transport
costs influence urban wage patterns, but rather that the situation is more complex
than the simple wage gradient theory suggests.

References

Allen, D.N. and Robertson, G.E. (1983) Silicon, Sensors and Software: Listening to
Advanced Technology Enterprises in Pennsylvania, Institute of Public Administration,
Pennsylvania State University.
Autor, D.H. (2019) Work of the past, work of the future, American economic, Review,
Papers and Proceedings, 109, 1–32.
Button, K.J., Lall, S., Stough, R., and Trice, M. (1999) High-technology employment and
hub airports, Journal of Air Transport Management, 5, 53–9.
Cairncross, F. (1997) The Death of Distance: How the Communications Revolution will
Change our Lives and our Work, Orion Business.
Cameron, G.C. and Clark, B.D. (1966) Industrial movement and the regional problem,
University of Glasgow Occasional Paper No 6.
Cervero, R. and Tsai, Y-.H. (2003) Job access and reverse commuting initiatives in
California: review and assessment, Transportation Research Record, 1859, 78–86.
Chisholm, M. (1971) Freight transport costs, industrial location and regional develop-
ment, in M. Chisholm and G. Manners (eds), Spatial Policy Problems of the British
Economy, Cambridge University Press.
Cook, W.R. (1967) Transport decisions of certain firms in the Black Country, Journal of
Transport Economics and Policy, 1, 344–56.
Diamond, D. and Spence, N. (1989) Infrastructure and Industrial Costs in British Industry,
HMSO.
Edwards, S.L. (1975) Regional variations in freight costs, Journal of Transport Economics
and Policy, 9, 115–26.
Evans, A.W. (1973) The Economics of Residential Location, Macmillan.
Eversley, D.E.S. (1965) Social and psychological factors in the determination of industrial
location, in T. Wilson (ed.), Papers on Regional Development, Blackwell.
Fischer, L. (1971) Die Berucksichtignung Raumordnungs Politiseler Zeilsetviengen in de
Verkehrsplanung, Strassenbahn und Strassenverkesstectnick.
Greenhutt, M.L. (1963) Microeconomics and the Space Economy, ­Scott-Foresman.

BUTTON 9781786435668 PRINT.indd 83 27/04/2022 11:07

 84 TRANSPORT ECONOMICS, 4TH EDITION

Gudgin, G. (1978) lndustrial Location Processes and Regional Employment Growth, Saxon
House.
Kain, J.F. (1964) A contribution to the urban transportation debate: an economic model
of urban residential and travel behavior, Review of Economics and Statistics, 47, 55–64.
Madden, J.F. (1985) Urban wage gradients: empirical evidence, Journal of Urban
Economics, 18, 291–301.
Mohring, H. (1961) Land values and the measurement of highway benefits, Journal of
Political Economy, 69, 236–49.
Moses, L.M. (1961) Towards a theory of intra urban wage differentials and their influence
on travel behavior, Papers of the Regional Science Association, 9, 53–63.
Owen, W. (1987) Transport and World Development, Hutchinson.
Pan, Q. (2019) The impacts of light rail on residential property values in a non-zoning city:
a new test on the Houston METRORail transit line, Journal of Transport and Land Use,
12, 1310.
Pirog, S.R. III and Lancioni, R. (1997) US–Japan distribution channel cost structures:
is there a significant difference? International Journal of Physical Distribution and
Logistics Management, 27, 53–66.
Rees, A. and Shultz, G.P. (1970) Workers and Wages in an Urban Labour Market,
University of Chicago Press.
Scholvin, S., Breul, M., and Revilla Diez, J. (2019) Revisiting gateway cities: connecting
hubs in global networks to their hinterlands, Urban Geography, 40, 1291–309.
Segal, M. (1960) Wages in the Metropolis, Harvard University Press.
Stafford, H.A. (1985) Environmental protection industrial location, Annals of the
Association of American Geographers, 75, 227–40.
Stevenson, R.L. (1879 [2004]) Travels with a Donkey in the Cévennes, Penguin.
Thomson, J.M. (1974) Modern Transport Economics, Penguin.
Timothy, T. and Wheaton, W.C. (2001) Intra-urban wage variations, employment location,
and commuting times, Journal of Urban Economics, 50, 338–66.
US Department of Housing and Urban Development (2000) State of the Cities, HUD.
Van Es, J. and Ruijgrok, C.J. (1974) Modal choice in freight transport, in E.J. Visser (ed.),
Transport Decisions in an Age of Uncertainty, Martinus Nijhoff.
Weber, A. (1929) Theory of the Location of Industry, University of Chicago
Press.
Wegener, M. (2004) Overview of land use transport models, in D.A. Hensher, K.J. Button,
K.E. Haynes, and P.R. Stopher (eds), Handbook of Transport Geography and Spatial
Systems, Elsevier.
Whebell, C.E.L. (1969) Corridors: a theory of urban systems, Annals of the Association of
American Geographers, 59, 1–26.
Zandiatashbar, A. and Hamidi, S. (2021) Transportation amenities and high-tech firm
location: an empirical study of high-tech clusters, Transportation Research Record,
2675, 820–31.

BUTTON 9781786435668 PRINT.indd 84 27/04/2022 11:07

 4 The Demand for Transport

4.1 Demands for Transport

Chapter 3 considered the inter-relationship between land-use patterns and trans-

port. This offered insights into some of the factors influencing the demand for
transport services. For example, one of the most pronounced characteristics of
the demand for transport is its regular fluctuation over time. In urban areas, the
demand for road space and public transport services is markedly higher in the
early morning and late afternoon than during the rest of the day. In the inter-
urban and international context, the demand for passenger transport fluctuates
regularly over a year with high seasonal peaks and peaks around public holidays.
With international freight transport, and especially shipping, there are long-term
cycles in demand.
This tendency for peaks and troughs in demand is often a reflection of fluctu-
ations in the demand for the final products made accessible by transport services.
In general, people wish to go on holiday to some destinations in the summer to
enjoy the warm weather, or the winter for skiing. Hence the seasonal peak in the
demand for highways, and rail and air services. Many businesses have tradition-
ally found it efficient to operate standard hours (that is, from ‘nine to five’) with
the consequential concentration of commuter traffic. Whether the post-Covid
pandemic era will support this in future is another matter. Longer-term fluctua-
tions in the demand for shipping services reflect the state of business cycles in the
trading nations – at the nadir of such cycles demand slumps; at the zenith it is
extremely buoyant.
There are also pronounced variations in the demands for different modes of
transport according to the distance of a trip – a feature found for freight trans-
port as well as passenger (de Bok et al., 2021). This is largely a function of the
technical features of the various modes that affect their overall attractiveness for
shorter or longer trips. Figure 4.1 offers a very generalized picture of the way that
conventional forms of transport dominate different lengths of personal trips. This
domination varies from walking and cycling being important for short journeys,
to air transport dominating inter-continental travel. Evidence supporting this is
seen from English data for short- and medium-length trips (Table 4.1), although
to be exact, the curves shown are not only related to demand, but also the avail-
ability of the various options.
What is slightly different in the twenty-first century is that in many circum-
stances the traditional transport modes have been confronted with alternatives
to travel. This is particularly true when the purpose of trips is to meet friends
or to undertake tasks that involve transference of information – for example,

BUTTON 9781786435668 PRINT.indd 85 27/04/2022 11:07

 86 TRANSPORT ECONOMICS, 4TH EDITION

Cost per mile

Inter-city rail
Walk Cycle Transit Car
Air

Telecommunication
0
Transit Car Inter-city rail Air
Walk Cycle
Distance

Figure 4.1 Dominant modes of transport

Table 4.1 Average trip length by main mode (England)

2005 2010 2015 2019

Private:
Walk 0.7 0.7 0.7 0.7
(Walks of over a mile) 1.4 1.5 1.4 1.4
Bicycle 2.4 2.8 3.0 3.3
Car/van driver 8.4 8.4 8.6 8.4
Car/van passenger 8.8 8.6 9.3 9.0
Motorcycle 9.8 9.9 11.1 8.0
Other private transport 18.4 17.1 15.0 15.3
Public:
Bus in London 3.8 3.5 3.9 3.7
Other local bus 4.4 4.7 5.0 5.1
Taxi-/mini-cab 4.8 4.8 5.2 5.0
London Underground 8.5 8.5 8.8 8.6
Surface rail 32.7 30.6 32.7 32.3
Other public transport 22.9 25.4 19.5 16.2

Source: UK Department for Transport.

conferences and business meetings – or to engage in various forms of transaction

involving, for example, shopping, finance, or marketing. In this case there has
been a pronounced switching in demand away from transport in its conventional
form to interactions involving some form of communication. In 2020, the long-
term potential of telecommunications for transference of information became
clear during the Covid-19 pandemic. This does not exclusively apply to person
movements, but also to the uptake of electronic correspondence in place of paper
communications by mail and courier.

BUTTON 9781786435668 PRINT.indd 86 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 87

The extent of competition between transport modes – the issue of cross-

elasticity of demand that we return to later – only really relates to cases where
there are genuine alternatives. If this is not so then one mode will dominate a
market. Competition is thus normally at the margin. There are few real alterna-
tives for tourists, for example, other than flying when it comes to long-distance
travel across oceans.
In terms of aggregate demand, despite regular fluctuations, it has been sug-
gested that over long periods of time there has been a remarkable stability in the
demand for travel. Households, for example, on average make roughly the same
number of trips during a day albeit for different purposes, over different routes,
or by different modes. There may be more leisure travel, but there are fewer work
trips and greater use is now made of air transport and the automobile at the
expense of walking and cycling. It is also suggested that this situation reflects that
there is a limit to the time people have available for travel, especially if they are
to enjoy the fruits of the activities at the final destinations. They, therefore, seek
to use some notion of a ‘travel time budget’ more effectively as travel technology
and travel choices change.
As analysis of travel time budgets has advanced, it has become clear that
the situation is more complicated than this, and that the constancy suggested
above should be subjected to much closer inspection. In the United Kingdom,
for example, there is evidence that average travel times have increased steadily
throughout the twentieth century. Explanations are difficult to find, but one
suggestion is that this is the result of rising incomes and that the constant time
budget implied by Yacov Zahavi (1977) and others only holds for each income
group. Zahavi, for example, found that people who made at least one motorized
trip per day in an urban area, spent a constant 1.1 hours on their daily travels.
Thus, people are moved from low-income groups with low travel time budgets to
higher-income groups with associated higher travel time budgets. One example
of a situation supporting this result is the Netherlands. In this case, ‘travel time
budgets’ increased over the last three decades of the twentieth century, with time
expenditure on travel per head increasing in relation to changes in travel costs and
income levels. Such findings emphasize the importance of time as well as conven-
tional variables in travel demand analysis and, as we see in the following chapter,
considerable emphasis is placed upon the role that time costs play in transport
decision-making.
Given this rather general, aggregate background we move to look in more
detail at the influences and motivations that affect travel and transport-related
demand. It seems appropriate to begin by considering the simple economic
demand function as an aide mémoire.

4.2 Influences on Travel Demand

We begin by recapping some basic demand theory but adding some discussion
of the peculiarities associated with transport. Demand is an abstract concept

BUTTON 9781786435668 PRINT.indd 87 27/04/2022 11:07

 88 TRANSPORT ECONOMICS, 4TH EDITION

reflecting what individuals would like to consume under different scenarios. It is

generally considered that the quantity demanded of a commodity, Da, is influ-
enced by its price, Pa, the prices of other goods (P1, P2, … , Pn), tastes, T, and the
level of income, Y. The neoclassical demand curve entails focusing on the impli-
cations of price changes on demand holding the other factor constant, that is:

Da = f(Pa, P1, P2, … , Pn, T, Y) (4.1)

A fall in price has two implications for demand – the price of the activity under
consideration falls relative to other prices (substitution effect) and more can be
bought for a given income (income effect) – both of which stimulate the quantity
of the good or service demanded. This means that normally there is a negative
relationship between price and the quantity demanded, provided other elements
in equation (4.1) remain constant (Figure 4.2). It gives the ‘downward-sloping
demand curve’. The elasticity of demand reflects the relative sensitivity of a
change in the quantity demanded to a change in price.
The curve shifts, however, if there are changes in other elements of equation
(4.1). An increase in income by allowing more to be bought at any price level
moves it out, whereas a fall in the price of a substitute, but making the good rela-
tively more expensive, pulls the demand at any price down. In some cases, as with
fashion goods or where public attitudes alter, the demand curve may shift because
of changes in ‘taste’, essentially a catch-all concept used to mop up effects not
captured in income or price effects.
For calculation purposes, as with many areas of demand analysis, a log-
linear specification of the form seen in equation (4.2) is often used to estimate
the effects of the various factors on demand. This allows a simple interpretation
of parameters such as elasticities, but does assume that, for example, the price
elasticity of demand is the same at all price levels:

$
 Price

 Income
 Price of other goods
 Tastes

Demand

0 Quantity
Figure 4.2 The simple demand curve

BUTTON 9781786435668 PRINT.indd 88 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 89

lnQM = α + β1lnPM + β2lnY + β3lnPN (4.2)

where:

QM is the quantity of mode M demanded;

PM is the price of mode M;
Y is income; and
PN is the price of an alternative, N.

All variables are expressed as logs (ln) and the parameters would normally be
estimated using multiple regression analysis. The price elasticity (defined as
{∆Q/Q}/{∆P/P}) is the parameter β1 in a double logarithmic equation as in seen
above; the income elasticity of demand, reflecting the sensitivity of the quantity
demanded to income changes, is β2, and β3 is the cross-elasticity of demand
reflecting the sensitivity of demand for mode M as the prices of alternatives vary.
While this simple framework holds for transport, as for all other goods and
services, there are refinements and detail that need to be highlighted if one is to
gain an understanding of the way the transport markets operate. The individual
terms in the equation are, in fact, often not simple variables but, rather, represent
complex compounds of several interacting factors (Litman, 2021). Price, for
instance, is not simply the fare paid but must embrace all the other costs involved
in obtaining the transport service (of which ‘time costs’, as we noted above, are
generally held to be the most important), while it may not be total income that
influences travel demand by individuals but rather income exceeding some thresh-
old subsistence level. Further, there is the need to be very clear on what exactly
it is that is being demanded: is it a trip per se or is it something more specific
than this, for example, a bus trip or a journey over a route? Quandt and Baumol
(1966) have gone so far as to suggest that it is not transport at all which is being
demanded but, rather, a bundle of transport services. We look at this idea more
fully in the context of forecasting in Chapter 12.
The emergence of behavioral economics has added to interest in these sorts
of issues. Behavioral economists have, for example, produced numerous examples
of when people do not make decisions, as assumed in standard economics, in
isolation from the opinions of others. A Rolls-Royce may offer very similar trans-
port services to a VW Beetle but it also offers prestige and status. On the other
hand, some may see the fuel advantage of the Beetle as helping to enhance the
ecological image of the user. In technical terms, there can be missing variables in
any transport demand function simply because many attributes of transport are
not only difficult to quantify, but are also not directly related to travel.
These types of problems and issues are clearly difficulties that cannot be
entirely circumvented in general discussions of the influences affecting transport
demand. They should be borne in mind as we move on to look in more detail at
some of the items contained in the demand function as set out in equation (4.1).

BUTTON 9781786435668 PRINT.indd 89 27/04/2022 11:07

 90 TRANSPORT ECONOMICS, 4TH EDITION

4.3 Pricing a Transport Service

As has been suggested, the price of transport embraces considerably more than
the simple money costs paid out in fuel, fares, or haulage fees. In transport mod-
eling and quantitative work these other components of price (that is, time costs,
waiting, insecurity, etc.) may be combined to form a generalized cost index of the
type we discuss in Chapter 5 where we focus in more detail on cost, but here we
concentrate on money prices and on the sensitivity of transport users to the price
of transport services.
Generalizations are obviously difficult, especially across all modes of trans-
port, but in many instances, it seems clear that price changes within certain
limits have relatively little effect on the quantity of travel or transport services
demanded. The demand for cargo shipping is, for example, very inelastic, in part
because of the lack of close substitutes for shipping services, in part because of
the short-term inelastic nature of the demand for the raw materials frequently
carried, and in part because of the relatively low importance of freight rates in the
final selling price of cargoes.
While, in broad terms, demand elasticities exhibit a degree of stability over
time, they do not remain entirely constant. In part this is due to shifts in the
demand function resulting from such things as rising incomes and changes in con-
sumer tastes. Studies of urban public transport in the 1970s, for example, covering
a variety of countries, indicate relatively low price elasticities with a direct fare
elasticity of around –0.3 being considered normal. Smith and McIntosh (1974),
looking at British urban bus undertakings, for instance, produce figures ranging
from –0.21 to –0.61, but the majority fall at the lower end of the spectrum. Charles
Lave’s (1970) American work on direct fare elasticities, however, found an elastic-
ity of –0.11 for transit trips in Chicago, implying the fare elasticity in the United
States in this period was slightly lower than in the United Kingdom at that time.
Studies in the 1980s surveyed by Phil Goodwin (1992) tend to produce similar
figures to Smith and McIntosh, although highlighting that the short-term elastici-
ties found in ‘before and after’ studies of fare change are in the order of a third of
the size of longer-term elasticities covering a reaction time in excess of five years.
The effect of price change on private car transport must be divided between
the effect on vehicle ownership and that specifically on vehicle use. Most early
United Kingdom studies of car ownership indicate an elasticity of about –0.3
with respect to vehicle price and –0.1 with respect to gasoline price (Mogridge,
1978). American empirical work suggests a rather higher sensitivity (the Chase
Econometrics Associates model, for example, implies a –0.88 purchase price elas-
ticity and a –0.82 fuel price elasticity), but responsiveness is still very low.
For car use, all the evidence suggests an extremely low fuel price elasticity
in the short term, which may be attributed to changing patterns of household
expenditure between vehicle ownership and use and people’s perception of
motoring costs. (We return to patterns of short- and long-run elasticities below.)
Terence Bendtsen (1980) brought early findings together in an international com-
parison that found the petrol price elasticity of demand for car use to be –0.08

BUTTON 9781786435668 PRINT.indd 90 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 91

in Australia for the period from 1955 to 1976; –0.07 in the United Kingdom for
1973–74; –0.08 in Denmark for 1973–74 and –0.12 for 1979–80; and –0.05 in the
United States for the period from 1968 to 1975. Oum et al. (1992) found a slightly
greater degree of sensitivity when looking at seven studies covering the United
Kingdom, the United States, and Australia: they yielded car usage elasticities in
the range –0.09 to –0.52.

Exhibit   Demand shocks on airline fares produced by high-speed rail transport

China rapidly built the world’s largest high-speed rail (HSR) network – about a third of
the world’s total. At the same time, it has seen considerable growth in its domestic airline
industries which were freed from fare, if not capacity, regulation in the early 2000s. Over
intermediate distances of 400 km to 1,000 km these modes compete for traffic. While much
of the competition involves incremental changes in terms of fares and frequency, ‘shocks’
or jolts in the market have impacted on their respective demands. These shocks may be
predictable, such as a large, discrete increase in capacity, or they may be unexpected, such
as a major technical failure. While often unique in their nature, they can reveal the extent of
cross-elasticity between these modes.
The launch of the Beijing–Shanghai high-speed railways (the Jing-hu HSRs) and the Wenzhou
train accident can be looked at as two examples. The Jing-hu HSRs were opened on June
30, 2011, and carry the largest passenger volume in China. The line is 1,348 km with
22 stops, with a design speed of 350 kph. Air services also connect seven of the stops. On
July 24, 2011, there was a major accident on the HSR resulting in 40 fatalities and nearly
200 injuries. The overall impact of the national HSR network was a significant reduction of
ridership, and resulted in the government mandating a travel speed of 300 kph and a
5 percent reduction in fares.

Beijing South
Langfang
Tianjn West

Cangzhou West

Dezhou East

Jinan West
Taian

Tengzhou East
Zaozhuang
Xuzhou East
Suzhou East Zhenjiang South
Bangbu South Danyang North
Dingyuan Changzhou North
Chuzhou Kunshan North
Nanjing South Shanghai Hongqiao
Wund East
Suzhou North

BUTTON 9781786435668 PRINT.indd 91 27/04/2022 11:07

 92 TRANSPORT ECONOMICS, 4TH EDITION

A difference-in-differences analysis was used. This methodology uses longitudinal data from
treatment and control groups to obtain an appropriate counterfactual for estimating a causal
effect of some event, by comparing the changes in outcomes over time between the two
groups. In this case the control group consists of airlines outside the influence of the new
HSR route and the control group of those airlines that compete with the new HSR services.
For the analysis, airline fare data from June 20 to July 24 – that is, 10 days before the
opening to 24 days after – were used to assess the impact of the HSR entering the market.
For the implications of the Wenzhou accident, the period 23 days before the incident to 12
days after was used.
Wei et al. (2017) find evidence of HSR substitution for airlines. The latter’s average fares
for services along the Jing-hu route declined by 30.6 percent upon the launch of the rail
service, but the mean fares rose for these same routes following the Wenzhou rail accident,
rebounding by 27.6 percent. In more detail, analysis shows a larger impact of the two HSR
events on low-cost and regional airlines, on shorter (less than 1,000 km) and/or tourism
routes, and on flights departing during the evening. This suggests that the HSRs serve as a
low-end substitution in the Chinese airline market.
See also: F. Wei, J. Chen, and L. Zhang (2017) Demand shocks, airline pricing, and high-speed
rail substitution: evidence from the Chinese market, Journal of Transport Economics and Policy,
51(4), 266–89.

Table 4.2 provides a survey of estimates of automobile and public transport

elasticities from 1990. We immediately note the higher long-term elasticities in all
cases, a topic we return to below, but also the relatively high-income elasticities
reflecting the possibilities for modal switching or other changes in travel behavior
as incomes change – an important consideration for policy-makers in developing
countries.
If we move to the other extreme of the transport market and look at airline
operations the evidence is that demand is slightly less elastic with long-term
demand (estimated from cross-sectional data) emerging as elastic. What these
coefficients, and indeed those cited above relating to other modes, often disguise
are quite significant differences in the elasticities for different groups of travel-
ers and between individual services. Mutti and Mural (1977), in an investigation
of air travel across the North Atlantic air market, generally found inelasticity,
but with variability between routes. Examinations of internal air traffic within
the United States, however, produce much more varied results. Samuel Brown
and Wayne Watkins (1968) and Gronau (1970) showed a remarkable degree
of ­consistency by producing price elasticities of –0.85 and –0.75 respectively,
but Jung and Fujii (1976) came to a somewhat different conclusion: air travel
demand for distances under 500 miles in the south-east and south-central por-
tions of the United States is price-elastic. To some extent variations in these
findings are influenced by a common problem found in estimations of elastici-
ties, they can be sensitive to the level of aggregation in the data deployed (Levin,
et al., 2017).

BUTTON 9781786435668 PRINT.indd 92 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 93

Table 4.2 Summary of demand elasticities provided by Graham and Glaister

Short/long run Elasticity

Fuel demand with respect to fuel price SR −0.25

LR −0.77
Fuel demand with respect to income SR 0.47
LR 0.93
Traffic (car-km) with respect to fuel price SR −0.15
LR −0.31
Traffic (car-trips) with respect to fuel price SR −0.16
LR −0.19
Traffic (car-km) with respect to car time SR −0.20
LR −0.74
Traffic (car-trips) with respect to car time SR −0.60
LR −0.29
Traffic (car-km) with respect to income SR 0.30
LR 0.73
Freight traffic with respect to price n.a. −1.07
Car ownership with respect to cost SR −0.20
LR −0.90
Car ownership with respect to income SR 0.28
LR 0.74

Note: n.a. = not available.

Source: Graham and Glaister (2004).

The relevance of this earlier work, other than in its technical innovations, is
clouded by the considerable institutional changes that took place from the late
1970s when air transport, as we discuss in Chapter 14, began to be deregulated.
Table 4.3 provides a larger listing of some of the main studies that have examined
air transport demand elasticities, including more recent ones. While no single
figure emerges from these studies that can be isolated as the ‘representative’ fare
elasticity, some general conclusions can be drawn.
There is a distinction between the fare elasticities for business and
­non-business air travel, with the latter being generally higher. This is a pattern
also seen, as one might expect, in terms of the type of fare being paid. This
conforms to the intuition that vacation travelers have more flexibility in their
actions (destinations, times of flights, etc.) whereas business trips often fly at
short notice.

Exhibit   Fuel efficiency of United States cars following the 1973 and 1979 oil crises

The oil embargo of 1973 resulting from the Yom Kippur War increased the world price of oil
from $3 per barrel to nearly $12. The United States retail price for a gallon of regular gasoline
rose from 38.5¢ in May 1973 to 55.1¢ in June 1974. In the wake of the Iranian Revolution, the
global price increased to $39.50 per barrel in 1979. As can be seen from the table, the fuel
efficiency of the United States’ car fleet was poor in the early 1970s with the oligopoly of

BUTTON 9781786435668 PRINT.indd 93 27/04/2022 11:07

 94 TRANSPORT ECONOMICS, 4TH EDITION

automobile manufacturers competing on design and size rather than miles per gallon. It has
been estimated that the results of the two disruptions in the fuel market resulted in the annual
cost of owning and operating a car by 1981 having risen by $2,400 compared with the pre-
1973 period. About a third of this was due to new environmental regulations, including, in
1975, the establishment of corporate average fuel economy (CAFE) standards.

Miles per US gallon Real price of gasoline (1967 = 100)

City Highway Harmonic mean

1968 12.59 18.42 14.69 97.3

1969 12.60 18.62 14.74 95.4
1970 12.59 19.00 14.85 98.0
1971 12.27 18.18 14.37 87.6
1972 12.15 18.90 14.48 85.9
1973 12.01 18.07 14.15 88.7
1974 12.03 18.23 14.21 108.3
1975 13.68 19.45 15.79 106.0
1976 15.23 21.27 17.46 105.3
1977 15.99 22.26 18.31 103.7
1978 17.24 24.48 19.89 100.5
1979 17.70 24.60 20.25 122.2
1980 20.35 29.02 23.51 149.6
1981 21.75 31.12 25.16 150.8
1982 22.32 32.76 26.06 134.7
1983 22.21 32.90 26.01 126.1
1984 22.67 33.69 26.59 119.2

The market reaction to this has involved changes in the number of trips made, driving
behavior, mode choices, and a market shift to smaller, more fuel-efficient automobiles. As
the table shows, the two crises saw a downsizing of cars and a clear increase, with a lag,
in fuel efficiency after 1974 and again after 1979. Overall, the fuel economy of the United
States’ fleet improved by over 50 percent between 1974 and 1984.
There was a significant increase in imports of smaller cars, most notably from Japan and
Germany, which, besides fuel economy, were of a better quality. By 1980, domestic luxury
cars with a gross weight averaging 4,500 pounds were no longer made and there began a
phasing-out of rear-wheel-drive layouts in compact cars in favor of lighter front-wheel-drive
designs. More cars offered fuel-efficient four-cylinder engines.
In terms of mileage traveled, however, while there were very short-term dips (from 1.306
billion in January 1974 to 1.284 billion in August 1967 and from 1.562 billion in May 1979
to 1.518 billion in September 1980) as the vehicle fleet adjusted, the steady growth in the
American automobile soon returned to its upward trends. The impacts of the large and
sudden fuel price increases were seen in the types and sizes of cars used rather than having
any long-term effect on their use.
See also: R.W. Crandall, H.K. Gruenspecht, T.T. Keeler, and L.B. Lave (1986) Regulating the
Automobile, Brookings Institution.

BUTTON 9781786435668 PRINT.indd 94 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 95

Table 4.3 Examples of findings of air fare elasticities studies

Study (publication year) Focus of study Values

Oum et al. (1992)c Trip purpose (business/non-business) −1.15 to −1.52

Mixed or unknown −0.76 to −4.51
Royal Commission Business travel −1.57 to −3.51
 on National Passenger Non-business travel −4.38 to −4.50
Transportation (1992)b Short trip (under 500 miles) −1.16 to −2.70
Long trip (over 500 miles) −1.34 to −2.56
Straszheim (1978)d First class −0.65
Economy, peak period −1.92
Economy, average −1.48
Economy, standard −1.12
Economy, promotional −2.74
Economy, high discount −1.82
De Vany (1983)a 28-mile trip (one way) −0.78
400-mile trip (one way) −1.02
650-mile trip (one way) −1.07
1,500-mile trip (one way) −1.14
2,500-mile trip (one way) −1.17
Gillen et al. (2002)c,e Long-haul international business −0.26
Long-haul international leisure −0.99
Long-haul domestic business −1.15
Long-haul domestic leisure −1.52
Short/medium-haul business −1.39
Bhadra (2003)a Less than 250 miles −0.67
250–499 miles −0.56
500–749 miles −0.74
750–999 miles −1.45
1,000–1,249 miles −1.82
1,250–1,499 miles −0.85
1,500–1,749 miles −1.08
1,750–1,999 miles −0.84
2,000–2,249 miles −1.06
2,250–2,499 miles −1.38
2,500–3,000 miles −0.86

Notes: a. United States data. b. Canadian data. c. Synthesis of previous studies. d. North Atlantic
data. e. From a range of international studies.

The lower sensitivities associated with first-/business-class fares also reflect the
service requirements of users who often seek room to work on planes and in
lounges. The estimates of the elasticities are also sensitive to the length of service
with shorter routes generally exhibiting higher fare elasticities, in part because
other modes of transport become viable options.
The difficulty with many statistics on demand sensitivity is, therefore, that
they are elasticities averaged over several groups. In fact, the price elasticity of
transport, as with the price elasticity of other goods, should ideally be set in a
specific context. In the case of transport, four broad types of classification are
important and we consider each in turn. Table 4.4 sets aviation elasticities in the

BUTTON 9781786435668 PRINT.indd 95 27/04/2022 11:07

 96 TRANSPORT ECONOMICS, 4TH EDITION

Table 4.4 
Elasticities of demand for passenger transport (expressed in absolute
values)

Mode Range surveyed Number of studies

Market demand Mode choice
elasticities elasticities

Aira
Vacation 0.40–4.60 0.38 8
Non-vacation 0.08–4.18 0.18 6
Mixedb 0.44–4.51 0.26–5.26 14
Rail: Inter-city
Leisure 1.40 1.20 2
Business 0.70 0.57 2
Mixedb 0.11–1.54 0.86–1.14 8
Rail: Intra-city
Peak 0.15 0.22–0.25 2
Off-peak 1.00 n.a. 1
All dayb 0.12–1.80 0.08–0.75 4
Automobile:
Peak 0.12–0.49 0.02–2.69 9
Off-peak 0.06–0.88 0.16–0.96 6
All dayb 0.00–0.52 0.01–1.26 7
Bus:
Peak 0.05–0.40 0.04–0.58 7
Off-peak 1.08–1.54 0.01–0.69 3
All dayb 0.10–1.62 0.03–0.70 11
Transit system:
Peak 0.00–0.32 0.1 5
Off-peak 0.32–1.00 n.a. 3
All dayb 0.01–0.96 n.a. 15

Notes: a. The distinction between vacation and non-vacation routes is rather arbitrary and may
partly account for the very wide range of elasticity estimates. b. This includes studies which do not
distinguish between categories (peak, off-peak, or all day).
n.a. = not available.

Source: Adapted from Oum et al. (2008).

context of other transport demand elasticities. With the exception of HSR, the
demand for airline services is more sensitive to price variations than modes, and
there is also a wider dispersion of results across studies.

4.4 Trip Purpose

Trips are made for a variety of purposes and by a diversity of individuals or com-
panies. Some are made at short notice and the price of the trip may be a relatively
small consideration compared with scheduling, reliability, and speed. Decisions
may also depend upon who is paying for the trip, whether it is the individual trave-
ling or an employer. Another consideration is whether there is a group purpose

BUTTON 9781786435668 PRINT.indd 96 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 97

in the trips, say a family holiday, or whether it is an individual traveling alone.

Studies of travel demand, therefore, control for the purpose of the trip.
There is an abundance of evidence that the fare elasticity for certain types
of trips is much higher than others. Business travel demand emerges from most
empirical studies as relatively less sensitive to changes in transport price than
other forms of trip. This is because business travel is often difficult to book
in advance and in many cases there is a principal–agent situation. The prin-
cipal (the owner of a business who pays for the trip) may be concerned about
keeping costs down but the agent (the employee booking the trip) has less incen-
tive to do so. Non-work trips are generally paid for by the traveler who is more
interested in economy, and especially when leisure trips involving a family group
are involved.
Over the years, Kraft and Domenich (1970) found that, in Boston, public
transport work trips exhibited a fare elasticity of –0.17 compared with –0.32 for
shopping trips, figures closely conforming to those found for public transport in
the United Kingdom. Considering air traffic, Mutti and Mural (1977) attribute
part of the variation they found in fare responsiveness on the North Atlantic
routes to the fact that ‘we expect personal travel to be more price elastic than
business travel’. Straszheim (1978) subsequently provides confirmation of this
view and isolates elasticities for different types of service. He concludes, ‘First
class fares can be raised and will increase revenue … . The demand for stand-
ard economy service is about unity, and highest for peak period travel … . The
demand for discount and promotional fares is highly price elastic …’. Tae Oum
et al. (1986) came to identical conclusions when looking at North American data.
Quite clearly, therefore, it is dangerous to attempt to analyse transport demand
without considering the specific type of trip being undertaken.

4.5 Levels and Methods of Charging

Users of different forms of transport (or, sometimes, different services of the

same mode) are often confronted with entirely different methods of payment.
Consequently, their perception of the price of a journey may differ from the actual
monies expended. Early work by Sherman (1967), for example, has suggested that
motorists perceive very little of the true overall price of these trips because they
base decisions on a limited concept of short-term marginal cost. Harrison and
Quarmby (1969) found that drivers perceived only between 50 and 60 percent of
their own costs (fuel, tire wear, vehicle maintenance, etc.) and this did not include
environmental costs and the costs of congestion imposed on others.
Users of public transport, on the other hand, are usually made much more
aware of the costs of their trip-making by the requirement to purchase a ticket,
usually prior to beginning their journeys. Nevertheless, given the range of season
tickets (which permit bulk buying of journeys over a specific route) and travel
card facilities (which permit bulk buying of journeys over a specified network),
the distinction is not a firm one.

BUTTON 9781786435668 PRINT.indd 97 27/04/2022 11:07

 98 TRANSPORT ECONOMICS, 4TH EDITION

The empirical findings regarding season tickets and travel cards are very
limited and are also not very helpful in providing general parameters. One
reason for this is that season tickets are only purchased periodically, and allow-
ing for other things that may occur between a fare change and a season ticket
purchase decision is challenging. Season tickets may also come in a variety of
forms – monthly, quarterly, and annual – with varying discounts. Peter White’s
(1981) review of the empirical information available on travel cards in the United
Kingdom did, however, point to a much lower price elasticity for travel card
systems than for conventional single ticket cash payment systems, findings subse-
quently supported by Mark Wardman (2014).

4.6 The Time Period

As with other purchasing decisions, people confronted with a change in transport

price may act rather differently in the ultra-short run, the ‘market period’, the
short run, and the long run. Immediate reactions, in the ultra-short term, to a
public transport fare rise may, for example, be dramatic, with people, almost on
principle, making far less use of services, even boycotting them, but knee-jerk
reactions are extremely short-lived and are seldom considered by economists,
although they are often of interest to politicians.
The ultra-short-term elasticity may, therefore, be extremely high but short-
lived. This type of situation may be less common than is sometimes thought
and, indeed, the reverse response may result in the slightly longer period. In the
market, for example, people may appear relatively unresponsive to a price change
either because they do not consider it a permanent change or because technical
constraints limit their immediate actions. The demand for gasoline, a major input
into modern transport, offers a good example of how its demand varies with
significant price rises, a situation found during the ‘oil shocks’ of the 1970s, and
more recently in 2007–2008.
In Figure 4.3 the price of oil rises from P1 to P2, and, ignoring any knee-jerk
reactions in the ultra-short term, the gasoline purchased will change little. The
transport users are locked into a pattern of behavior that makes reducing their
fuel consumption difficult; this is the ‘market period’. In the short run, however,
individuals can change their travel patterns by switching modes, combining trips,
and cutting out some travel, and businesses can reschedule the use of their vehicle
fleets and modify their collection and delivery patterns. The demand, therefore,
becomes more elastic in relation to the new price, and the amount bought falls to
Q2. In the long term, people can change the type of car they use, their employ-
ment, and their residential location, and industry can modify their entire supply
chain. Hence the quantity of gasoline demand will fall to Q3 in the figure, and the
demand curve will flatten out further.
Table 4.5 offers more direct indications of how price and income elasticities
can vary over time after a price shock in the gasoline market. Figure 4.4 makes

BUTTON 9781786435668 PRINT.indd 98 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 99

Price of oil
Short-run
demand Market demand

P2

P1

Long-run demand

0 Q3 Q2 Q1 Gasoline purchased

Figure 4.3 Abstraction showing changes in demand elasticities for oil over time

Table 4.5 
Price and income elasticities of gasoline demand by automobile users

Price elasticity Income elasticity

Short run –0.2 to –0.3 0.35 to 0.55

Long run –0.6 to –0.8 1.1 to 1.3

Source: Basso and Oum (2007).

0.9
0.8
0.7
Absolute elasticity

0.6
0.5
0.4
0.3
0.2
0.1
0
0 2 3 4 5 10
Years following fare change
Figure 4.4 E
 stimated United Kingdom public transport demand elasticity over the period
after a fare change

BUTTON 9781786435668 PRINT.indd 99 27/04/2022 11:07

 100 TRANSPORT ECONOMICS, 4TH EDITION

use of Goodwin (1992) to indicate the dynamic absolute fare elasticity of demand
for United Kingdom transit fares.
In a more direct transport context, when considering the effect of general
rises on commuter travel costs, the necessity of having to make journeys to work
is likely to result in minimal changes in travel patterns in the short term but over
a longer period, relocations of either residence or employment may produce a
more dramatic effect. Teleworking has added to the options in recent years. This
implies that one must take care when assessing elasticity coefficients, and it is
useful to remember that cross-sectional studies tend to offer estimates of long-run
­elasticity while time-series studies reflect short-term responses.

4.7 The Absolute Level of the Price Change

Price elasticities are generally found to increase for both passenger and freight
movements (de Bok et al., 2021) the longer the trip under consideration. This
should not be seen simply as a function of distance but rather a reflection of the
absolute magnitude of, say, a 10 percent rise on a $5 fare compared with that on
a $500 fare. It is also true that longer person journeys are made less frequently,
and thus people gather information about prices in a different way. Additionally,
they often tend to involve leisure rather than business travel; this suggests that
distance may be picking up variations in trip purpose. In the air transport market,
for example, Arthur De Vany (1974) found in a classic study that price elasticity
rose from –0.97 for a 440-mile trip in the United States to –1.13 for an 830-mile
trip. For similar journeys, Richard Ippolito (1981) found the respective elasticities
to be –0.525 and –1.0.
While it is important to treat elasticities with care because of these types of
aggregation issues, there is a further reason for counseling caution when consider-
ing such parameters: there are several statistical methods employed to calculate
elasticities and these can influence the values obtained. In some instances, these
related to the time span of the elasticity being studied; some techniques are
mainly used for short-term elasticity estimation while others are more relevant
for cross-sections of data, and thus long-term elasticity calculations. The inten-
tion here is not to go into the technicalities of the various modeling frameworks,
although some discussion of this is to be found in Chapter 12 when explicitly
discussing demand modeling and forecasting, but rather to highlight what seem
to be the two main trends:

• First, if aggregate data are used, for instance looking at demand for an entire
railroad network, then, although it is far from universally true, elasticities
tend to be higher than from ‘discrete choice’-type models that use data at the
individual service or customer level.
• Second, even within a modeling framework (be it involving the use of aggre-
gate data or disaggregate data) the exact mathematical form of the equa-
tion used influences the elasticity calculated. Equation (4.2) highlighted the

BUTTON 9781786435668 PRINT.indd 100 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 101

Table 4.6 
Demand elasticities for North American rail freight transport (expressed in
absolute values)

Commodities Log-linear Aggregate Translog Discrete

logit choice

Aggregate commodities 1.52 0.25–0.35, 0.09–0.29, n.a.

0.83 0.60
Chemicals n.a. 0.66 0.69 2.25
Fabricated metal products n.a. 1.27 2.16 n.a.
Food products 0.02, 1.18 1.36 2.58, 1.04 n.a.
Iron & steel products n.a. n.a. 2.54, 1.20 0.02
Machinery n.a. 0.16–1.73 2.27–3.50 0.61
Paper, plastic, & rubber 0.67 0.87 1.85 0.07­–0.09
products
Petroleum products n.a. n.a. 0.99 0.53
Stone, clay, & glass products n.a. 2.03 n.a. 0.56
Transport equipment n.a. n.a. 0.92–1.08 2.68
Wood & wood products 0.05 0.76 1.97, 0.58 0.08

Note: n.a. = not available.

Source: Oum et al. (1992).

log-linear, or ‘Cobb–Douglas’, specification, the generic features of which

we return to in Chapter 5. Table 4.6 provides some general guidance to these
with respect to estimated elasticities of demand for rail freight transport
elasticities for rail freight transport.

4.8 Income Levels

After housing and food, transport represents, at about 10 to 15 percent for a

single-car-owning United Kingdom household, the largest element of expendi-
ture out of income in industrialized countries. While there is ample evidence that
transport is a normal good in the sense that more is demanded at higher levels
of income, this generalization does not apply to all modes of transport or to all
situations. As was seen in Chapter 2, at the national level, income exerts a posi-
tive influence over car ownership decisions, and we consider this relationship at
a more disaggregate level below. But the situation is not so clear-cut with public
transport use, and in some cases this mode clearly becomes an ‘inferior good’ with
its use falling after some level of income has been attained. As incomes have risen
and, with them, car ownership has become more widespread, public transport has
in many situations proved to be an inferior good. Gwilliam and Mackie (1975)
suggest that the long-run elasticity of demand with respect to income was of the
order –0.4 to –1.0 for urban public transport trip-making in the United Kingdom
during the 1970s. They argue that, although car ownership rises with income,
and hence some trips are diverted from public transport, there is still a limited
off-setting effect inasmuch as wealthier households make more trips in total. This

BUTTON 9781786435668 PRINT.indd 101 27/04/2022 11:07

 102 TRANSPORT ECONOMICS, 4TH EDITION

effect would seem to be less relevant today with much higher levels of automobile
ownership in developed countries.
The income elasticity of demand for many other modes of transport is seen
to be relatively high, and especially so for modes such as air transport. John
Taplin (1980), for example, suggests a figure of the order of 2.1 for vacation air
trips overseas from Australia. By its nature air travel is a high-cost activity (the
absolute costs involved are high even where mileage rates are low), so that income
elasticities of this level are to be expected. There is also some evidence that wealth
influences the demand for air travel, and a study of the Israeli market found a
wealth elasticity of 2.06 (Alperovich and Machnes, 1994).
As with price, income changes exert somewhat different pressures on
transport demand in the long run compared with the short. In general, it may
be argued, a fall in income will produce a relatively dramatic fall in the level of
demand, but as people readjust their expenditure patterns in the long term the
elasticity is likely to be much lower. Looking at the responsiveness of car owner-
ship levels to income changes, British and United States studies suggest a short-
term income elasticity of between 2.0 and 4.5, while in the long run it appears
to fall to around 1.5 (Button et al., 1982). However, as with price elasticities, the
relationships between long- and short-term effects are not completely clear-cut.
Reza and Spiro (1979), for example, produce an estimate of 0.6 for the short-run
income elasticity of demand for petrol, rising to 1.44 in the long run, findings
generally in line with what we saw in Table 4.5. If one assumes that gasoline
consumption is a proxy for trip-making, then one could attempt to justify this in
terms of a slow reaction to changing financial circumstances – a reluctance, for
example, to accept immediately the consequences of a fall in income. In fact, the
situation is likely to be more complex because the long run may embrace changes
in technology, and possibly locations, that alter the fuel-consumption–trip-
making relationship. Thus, these figures may still be consistent with the initial
hypothesis regarding the relative size of short- and long-run income elasticities
of demand for travel.
There is a literature on the possibility of a constant travel income budget
akin to the travel time budget mentioned in Section 4.1 with households tending
to spend a fixed proportion of their income on transport. Zahavi (1977), for
example, in his examination of data from a large sample of urban transport users,
noticed that the proportion of disposable income spent on cars by car-owning
households at any income level appears to be approximately constant at a given
moment in time. (United Kingdom data suggest a proportion of around 15.5
percent – slightly larger for low incomes – for 1971 to 1975.) The evidence for bus
transport is less clear, but Mogridge suggests that while the proportion of dispos-
able household income spent on bus travel seems to rise with income, a constant
proportion still emerges if adjustments are made for the number of people in each
household.
In the longer term there is evidence at the aggregate level that since the early
1980s there has been a steady increase in the overall proportion of income or
disposable income allocated to travel in the United Kingdom. (This contrasts

BUTTON 9781786435668 PRINT.indd 102 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 103

with a roughly constant proportion in Canada and the United States.) This may,
however, be explained in terms of rising income levels but constant proportional
travel budgets within each income group. But the conclusion concerning the idea
of some overall budget mechanism governing individual travel decisions must be
that the evidence available still leaves many questions unanswered and the theory
still largely unproved. In an examination of 30 or so travel surveys from across the
world, Andreas Schafer (2000) finds some consistency in travel money budgets
for countries with about 0.85 cars per household, but the link is weak and only
applies at a very aggregate level.

Exhibit   Meta-analytical synthesis of demand elasticity results

Meta-analysis is the statistical analysis of a collection of results from individual studies. It

is a quantitative alternative to narrative discussions of research studies. The analysis uses
probability-based statistics to summarize primary findings or to elicit moderator variables
that influence variations in the results of these studies.
Most of these studies make use of meta-regression analysis and can be generalized as:

bj = β + ∑ kk = 1 akZjk + ej; j = 1.2, … , L

where:
bj is the reported estimate of β in the jth study in the L studies being examined;
β is the ‘true’ value of the parameter of interest;
Zjk are independent variables that measure the relevant characteristics explaining
variations between studies;
αj is a coefficient that reflects the biasing effect of study characteristics; and
ej is the error term.
Zjk includes, specification variables (for example, the form of the empirical model, type of
regression, and data sources), sample size, characteristics of the primary literature, and
variables reflecting whether possibly relevant effects have been omitted from a primary study.
Wardman used meta-analysis to examine 1,633 price elasticities of demand found in 167
United Kingdom studies of car, rail, bus, and underground travel. Generally plausible
relationships with journey purpose, mode, ticket type, area, and distance were found with
anticipated differences between the short and the long run. The study makes use of the
parameters obtained to imply general elasticities taking account of the variations in the
primary studies’ contexts, their technical specifications, and control variables they used. The
results for United Kingdom passenger rail are seen in the table and compared to the fare
elasticities laid down in the official Passenger Demand Forecasting Handbook. (Similar tables
were calculated for other modes.)
As anticipated, the short-term impacts of rail fare changes are less than the longer-term
implications. This applies to all fare categories. In general, the demand associated with
changes in first-class and full fares is lower than that for discount fares. The former

BUTTON 9781786435668 PRINT.indd 103 27/04/2022 11:07

 104 TRANSPORT ECONOMICS, 4TH EDITION

categories are more generally associated with business travel and commuter travel, whilst
the discount fares are more often used for discretionary travel. By distance, work shows
that, across the studies included, the long-run demand curve for rail travel is more sensitive
to fare changes than is the short-run curve. Finally, the parameters suggested by the United
Kingdom’s Passenger Demand Forecasting Handbook would seem to be over-estimating the
sensitivity of rail passengers’ demand regarding fare changes.

Journey type Short PTE Short Short Inter-urban Inter-urban

London other London non-London

Season SR −0.18 −0.19 −0.31 −0.31 −0.28

Static −0.31 −0.33 −0.52 −0.52 −0.48
LR −0.36 −0.38 −0.61 −0.61 −0.57
PDFH LR −0.60 −0.50 −0.70 −0.75 −0.90
Non-season SR −0.37 −0.39 −0.62 −0.62 −0.57
(first & LR −0.73 −0.77 −1.23 −1.23 −1.14
standard) Static −0.63 −0.66 −1.05 −1.05 −0.98
PDFH LR −0.85 −0.80 −1.00 −1.05 −1.00
First LR −0.65 −0.69 −1.10 −1.10 −1.02
Full LR −0.70 −0.74 −1.19 −1.19 −1.10
Reduced LR −0.94 −0.99 −1.58 −1.58 −1.47

Notes: SR = short-run; LR = long run; static = intermediate between SR and LR; PTE = public transport executive;
PDFH = Passenger Demand Forecasting Handbook.

See also: M. Wardman (2014) Price elasticities of surface travel demand: a meta-analysis of
UK evidence, Journal of Transport Economics and Policy, 48, 367–84; and K.J. Button (2019)
The value and challenges of using meta-analysis in transportation economics, Transport
Reviews, 39, 292–308.

4.9 The Price of Other Transport Services

The demand for any transport service is likely to be influenced by the actions of
competitive and complementary suppliers. (Strictly speaking, it is also influenced
by prices in all other markets operating in the economy, but, with the possible
exceptions of the land market, which was discussed in Chapter 3, and electronic
communications, which we look at in Chapter 5, the importance of these is less
great.) We have only touched upon the importance of motoring costs vis-à-vis
the demand for public transport services and more is said on this topic later in
the chapter. Moreover, there are the cross-price effects between modes of public
transport. Table 4.7 presents the results from several different studies looking at
elasticities of demand (both own fare and cross-fare) for transport in Greater
London for 1970 to 1975.
The variation in results generally reflects the adoption of alternative estima-
tion procedures, the inclusion of difference background, control variables, and
time-lag allowances. One of the more interesting points is the almost total insen-
sitivity of the demand for urban car use to the fare levels of both bus and rail

BUTTON 9781786435668 PRINT.indd 104 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 105

Table 4.7 Greater London estimated Monday to Friday fare elasticities (1970 to 1975)

Study Elasticity of With respect to

Bus Local rail

Fairhurst & Morris (1975) Bus –0.60 0.25

Rail 0.25 –0.40
Glaister (1976) Bus –0.56 0.30
Rail 1.11 –1.00
Collings et al. (1977) Bus –0.41 n.a.
Lewis (1978) Peak road traffic 0.03 0.06

Note: n.a. = not available.

Source: Glaister and Lewis (1978). This paper contains the full references to the studies cited.

public transport modes. This fact, which has been observed in virtually all studies
of urban public transport, is the main reason that attempts by city transport
authorities to reduce or contain car travel by subsiding public transport fares have
proved unsuccessful.
The table suggests that there is likely to be more switching of demand
between public transport modes as a result of one changing its fare structure than
between that mode and private transport. Gilbert and Jalilian (1991) also pro-
vides cross-elasticities for London between public transport modes but throws up
somewhat different results with the indication that bus travel in London is more
sensitive to underground fares with a cross-fare elasticity of 0.90. Analysis of the
cross-elasticity of demand between car and rail costs for inter-city rail travel in
the United Kingdom, shows much less sensitivity: an average cross-elasticity of
0.09, falling to 0.027 and to 0.028 on some freeways (Acutt and Dodgson, 1995).
What emerges is that is that despite generally low levels of substitution
between public transport modes passengers are particularly sensitive to in-vehicle
access/egress and waiting time in choosing a mode and less so for fare varia-
tions. In general, rail demand is less sensitive to changes in bus demand than
bus demand is to changes in rail demand. We also find that peak-hour demand
more markedly switches between public transport modes than off-peak demand
(Fearnley et al., 2018).
In other transport markets the cross-elasticity of demand may be higher,
both between operators of the same mode of transport and between modes them-
selves. Rate reduction in non-conference shipping lines, for example, has periodi-
cally attracted considerable traffic away from the cartel, conference, and alliance
carriers. Similarly, scheduled airlines have experienced a relative contraction of
demand over time as low-cost carriers have entered the market.
Evidence on the cross-price elasticity of complementary transport services,
such as feeder links to longer-distance trunk hauls, is scant. The expansion of
the motorway network in the United Kingdom, the autobahns in Germany, and
the freeways in the United States reduced motorway travel costs and increased the
demand for certain feeder roads, while at the same time reducing it on competing

BUTTON 9781786435668 PRINT.indd 105 27/04/2022 11:07

 106 TRANSPORT ECONOMICS, 4TH EDITION

routes. The exact implications of such network effects are much more difficult to
trace out than changes in modal split but, in practical terms, are important fea-
tures of the transport system.
The topic of cross-elasticities of demand between modes is returned to in
Chapters 8 and 9 when discussing their importance when trying to divert traffic to
less environmentally damaging and congestive forms of transport.

4.10 Tastes, Human Desires, and Motives

One of the items which influences equation (4.1) but not mentioned to date, that
is often included in elementary discussion of demand, is the ‘catch-all’ variable,
‘taste’. Traditional microeconomic theory assumes that consumers’ taste stays
constant when prices or income change. In practice this is a very strong assump-
tion to make. For example, a regular commuter tries a new bus service, then this
provides additional knowledge of the transport alternatives available. This may
affect future commuting decisions.
While there may be circumstances in which such a term could and, indeed,
should be included in the demand function, in general, tastes are more likely to
influence the form of the demand equation – linear, log-linear, or whatever – and
its location in price/quantity space. Consequently, a change in taste may be seen
to affect the relationships between demand and the explanatory variables rather
than result in some movement along a demand curve following the pattern of an
established relationship. But situations do vary.
The economic meaning of ‘taste’ is seldom made clear, but in practice it
seems to embrace all influences on demand not covered by the previous headings.
Over time, tastes in transport certainly have changed. Globally, there has been
an increased car orientation of society as people have moved towards private
transport irrespective of cost or income considerations, while in freight transport
the changing structure of the national economy (especially the switch from basic
heavy industry to light industry producing high-value, low-weight products)
has shifted the emphasis from price to other aspects of transport service. Both
changes must to some extent be related to rising standards of living. With more
wealth and greater free time there is likely to be an enhanced desire to benefit from
the greater freedom and flexibility offered by private transport. A change in loca-
tion patterns is also possible with larger residential plots away from urban centers
now becoming attractive for many people.
Another aspect of ‘taste’ concerns inertia and asymmetry in d ­ ecision-making
(Goodwin, 1977). This has two implications:

• First, there may be discontinuities in the demand curve for transport,

or at least parts of the demand curve reflect almost total insensitivity to
price changes, as a result of habit and inertia on the part of individuals
and firms. It may be explained in some cases quite simply by the fact that
there are costs involved in seeking out information about alternatives, and

BUTTON 9781786435668 PRINT.indd 106 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 107

continuing as before is thus the rational response until more major price
changes occur.
• Second, there may be cases where responses are not symmetrical; a ratchet
effect exists whereby the reaction to a price fall is not the same as the reac-
tion to an identical price rise. Limited empirical work has been done on such
‘path dependencies’, although Blasé (1980) did find evidence of asymmetries
in travel behavior in the context of fuel price variation and Dargay (1993)
provided further support for this across a number of national studies.

Rather more effort has been put into the question of service quality. It is
noticeable, for example, from empirical studies, that public transport demand is
sensitive to changes in service quality, especially to any reduction in the speed or
frequency of services. Again, this fact reflects the decreased importance attached
to the purely monetary dimension. Market research in the West Midlands of the
United Kingdom, for example, revealed that only 27.1 percent of people felt that
keeping fares down would be the greatest improvement to local public transport;
the remainder looked for service quality improvements, for example, 14.6 percent
for greater reliability, 10.4 percent for higher frequency, 10.4 percent for more bus
shelters, 10.0 percent for cleaner vehicles, etc. (Isaac, 1979).
An extensive survey by Armando Lago et al. (1981) examined a range of
international studies concerned with urban public transport service elastici-
ties. Their general conclusion, that services will generate less than proportional
increases in passenger and revenue (that is, εS < 1), seems to contradict the above
findings. But this may be misleading. To begin with, the survey looks at several
service-quality attributes in isolation rather than at a package of service features.
It also admits that many of the services sought by potential public transport
users are qualitative rather than quantitative and hence are not amenable to the
types of analysis reviewed. The survey also highlights the fact that service quality
is far more important when the initial level of service is poor; the general elas-
ticities found for peak-period ridership, for instance, are much lower than those
for off-peak. The evidence presented suggests that service headway is one of the
more important service variables; the studies examined indicates an elasticity of
the order of –0.42 compared with, for example, –0.29 for in-vehicle bus travel
time.
The evidence suggests that, as the modern world of inventory management
and tracking has evolved, low price is also often no longer the dominant determi-
nant of freight modal choice. In a survey conducted in the United Kingdom by
the Price Commission as long ago as 1977, for instance, it was found that only in
52 percent of cases did consignors elect to use the cheapest road haulage operator
available for local trips, 77 percent for intra-regional trips and 64 percent for trunk
hauls. Subsequently, McGinnis’ (1990) study of freight transport in the United
States, for example, found that rates only became an important consideration
after required service criteria were met.
The UK’s Price Commission found that many companies were so uncon-
cerned about finding the lowest price that competitive quotations were not

BUTTON 9781786435668 PRINT.indd 107 27/04/2022 11:07

 108 TRANSPORT ECONOMICS, 4TH EDITION

sought. An emphasis placed upon vehicle suitability is seen to reflect customer

concern about such factors as weather protection, systems for securing loads,
and compatibility of vehicle with product. These are concerns unlikely to have
been of paramount importance when heavy industry dominated the economy
but are of much more a concern for more modern, high-technology firms and for
large retailers. These firms were increasingly turning to ‘just-in-time’ production
methods whereby inventories are kept to a minimum. To optimize such processes,
reliability of supply is vital and companies were willing to pay the additional
financial costs that this may entail. This is a subject considered in more detail in
Chapter 10.

Exhibit   The role of ancillary revenues in airline finances

The fare an airline passenger paid prior to the trend towards economic liberalization from
1978 provided a package of services; for example, a seat with a given pitch in each class,
meals, ability to change flights, and a relatively large checked baggage allowance. The 2000s
saw this change as initially low-cost airlines offered a basic service at a low fare with other
services being available for a supplement. Basically, a passenger bought a package of the
services that they were willing to pay for; technically there was unbundling. Then additional
services were added, including such things as insurance, hotel rooms, and car rental to the
bundle for further fees.
The airlines did this to increase their load factors, bringing in travelers who did not want
all the elements of the standard ‘package’, and to relate fares more closely to the costs of
carrying individuals. The idea was that by not imposing the cost of services that a potential
passenger did not value highly, they could price down the demand curve to fill seats by
attracting those who would not fly. Additional items were added to allow pricing up for
those willing to pay for them.
The table indicates, for United States and non-United States regions in 2012, a rough
breakdown of the forms of ancillary revenues that were generated. The figure shows the
results of an online survey conducted in 2011 by John O’Connell and David Warnock-Smith
asking how likely people were to by buy accommodation, car rental, travel insurance, etc. in
the future.

À la carte Excess Onboard Sales of Hotels, car rental,

services baggage fees retail miles insurance

Non-US carriers 15% 15% 10% 30% 30%

United States carriers 25% 20% – 50% 5%

These results, combined with expert opinion and along with secondary data, indicate that
airport car parking and checked baggage charges are the most accepted commission-based
and unbundled products for airlines to sell.

BUTTON 9781786435668 PRINT.indd 108 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 109

35%

30%

Absolute elasticity 25%

20%

15%

10%

15%

5%

0
Most Very Quite Some what Not very Not at all
likely likely likely likely likely likely

See also: J.F.O. Connell and D. Warnock-Smith (2013) An investigation into traveler
preferences and acceptance levels of airline ancillary revenues, Journal of Air Transport
Management, 33, 12–21.

4.11 The Notion of a ‘Need’ for Transport

The demand function indicates what people would buy given a budget constraint,
but it is often argued that allocation of resources on this basis results in inequali-
ties and unfairness because of differences in household income or other circum-
stances. There are, thus, some advocates of the idea that transport services, or at
least some of them, should be allocated according to ‘need’ rather than effective
demand. The concept of need is seldom defined, or at best rather imprecisely
(Banister et al., 1984), but seems to be closely concerned with the notion of merit
goods – that is, needs ‘considered so meritorious that their satisfaction is provided
for through the public budget over and above what is provided for through the
market and paid for by private buyers’. The idea is that, just as everyone in a civi-
lized society is entitled to expect a certain standard of education, medical cover,
security, etc., so they are entitled to enjoy a certain minimum standard of trans-
port provision. Looked at another way, the government, or other donor, provides
such a good based on ‘merit’ because it can better provide for individual welfare
than allowing consumer sovereignty.
One can point to several transport policy initiatives over time which are
based upon this idea. The 1930 Road Traffic Act in the United Kingdom, for
example, introduced, besides other things, road service licenses into the bus
industry which embraced the notion of public need. The Traffic Commissioners
interpreted this to mean the provision of a comprehensive network of services
for an area, irrespective of the effective demand for specific routes. Licenses were
granted on this basis and operators cross-subsidized the unremunerative services
with revenue from the more profitable ones. More explicit were the social service
grants given to the railways under the 1968 Transport Act whereby 222 services

BUTTON 9781786435668 PRINT.indd 109 27/04/2022 11:07

 110 TRANSPORT ECONOMICS, 4TH EDITION

were subsidized for social reasons, once again despite deficit effective demand
for their services. Additionally, the United Kingdom government has provided
capital and operating cost subsidies to assist the shipping and air services to the
remoter islands of Scotland.
In a different context, the 1978 Airline Deregulation Act in the United
States has provided subsidies for services to small communities (the Essential
Air Service Program) and the 1987 National Transportation Act in Canada
provided explicitly for subsidies to ‘essential’ air services in the northern part of
the country.
Allocation by some notion of need implies that the market is not used to
determine the output or price of transport services. Figure 4.5 takes the simple
case of a transit service that, under market conditions, would produce a fare of
P1 and offer Q1 seats. If public policy requires that a higher level of services is
warranted, say Q2 seats, then this distorts the market. For a commercially driven
market to offer Q2 seats, a fare of P2* would be required, but only Q2* seats
would be willing to pay this. For the Q1 transit seats to be efficiently used, a fare
P2 is needed. The result is the necessity to subsidize the difference between this
fare and the commercially necessary fare for Q2 seats. We deal with some of the
issues involved in transport subsidies, and alternative methods of finance, later in
the book.
This notion of need rather than effective demand raises two important
issues. First, exactly what is the nature of ‘need’? And second, if one accepts that
the concept has some operational meaning, how can it be incorporated into eco-
nomic analysis? We look at these two questions in turn.
The need for adequate transport provision stems from the idea that people
should have access to an acceptable range of facilities. It is, therefore, essentially a
‘normative’ concept. Transport is seen as exerting a major influence on the quality

$
Supply2

Supply1

P2*

P1

P2

Demand

0 Q2* Q1 Q2 Transit seats

Figure 4.5 Allocation by ‘need’

BUTTON 9781786435668 PRINT.indd 110 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 111

of the lives of people and a certain minimum quality should be ensured. The
United Kingdom has traditionally emphasized this view of mobility; for example:
‘The social needs for transport also rank high – the needs of people to have access
to their work, shops, recreation and the range of activities on which civilized
society depends’ (UK Department of Transport, 1977). Defining the exact level
of mobility in this context is difficult, but it is helpful to look at the groups who,
for one reason or another, seem in need of transport services in addition to those
who would be forthcoming in the market.
The most obvious group is the poor, who cannot afford transport. Transport
expenditure forms a substantial part of a household budget and, consequently,
those on the lowest income must make fewer trips, shorter trips, or trips on infe-
rior modes of transport. A major problem is that, as we see later, as income levels
rise, in general there is a tendency towards higher car ownership, leaving only
depleted and expensive public transport facilities for those at the lower end of the
income distribution. A household with a car in the United Kingdom tends, for
instance, to make on average about 300 fewer bus journeys per year than compa-
rable households without a car. But there are also wider issues, in that this change
in the transport sector has implications for population distribution.
Higher car ownership in rural areas, and the resultant reduction in the
demand for local public transport, has put pressure on rural bus and rail services.
Between 1970 and 1974, for example, the National Bus Company in England,
which was responsible for most rural services in England and Wales, reduced its
bus kilometers by 7 percent. This, in turn, has been one of the causes of rural
depopulation. Similar trends were seen in Chapter 2 with respect to the United
States. There is a long tradition of rural public and special transit service operating
and capital assistance from federal and state government. The US Federal Transit
Administration has over 5,300 rural and inter-city transit programs, and there is
special federal and state transport support for special mobility services transport.
In addition, social services funding is available. The question then arises as to
whether society in general needs a balance between urban and rural society.
While inadequate income poses one problem, there are other groups in
society that are often felt to need assistance. The old, the infirm, and children are
obvious examples where, irrespective of income, effective demand may be felt to
be an inadequate basis upon which to allocate transport resources. The available
evidence suggests that only about 10 percent of households in the aged or disa-
bled category have private transport at hand. Even when a household does own
a car (or has one made available through employment agreements), there are still
members of it who may be deemed in need of additional transport.
Women enjoy less mobility than men, although the gap is closing, at least in
wealthier countries. An early study of mobility by Hillman (1973), for instance,
found that 70 percent of young married women in the outer metropolitan area
of London had no car available for their everyday use – even 30 percent of those
qualified to drive were in this position. Even in the mid-1990s, women in the
United Kingdom aged between 26 and 59 years, the prime years of domestic
caring responsibilities, only traveled about half the miles of men, making fewer

BUTTON 9781786435668 PRINT.indd 111 27/04/2022 11:07

 112 TRANSPORT ECONOMICS, 4TH EDITION

trips and shorter trips. While there are national differences, the British pattern
was far from atypical.
Although there have been changes since Hillman’s study, the evidence
is that women still enjoy both less mobility and accessibility than their male
counterparts.
There are political-economy arguments, therefore, that these various groups
are in need of adequate and inexpensive public transport services (or special
transport provision in the case of the disabled) and that the normal market
mechanism is inadequate in this respect.
If one accepts the notion that need is, in certain contexts, the relevant concept
rather than effective demand, then, for practical purposes, this idea requires
integration into more standard positive economic theory. (It should perhaps be
noted that many people do not accept the idea of ‘need’ as an allocative device
but advocate tackling problems of low income or disadvantage at their source
through measures such as direct income transfers, but this is an issue outside our
present discussion.) Perhaps the simplest method of reconciling the difficulty is to
treat the monies paid out by government and other agencies in subsidies to social
transport services, as the effective demand of society for the services. One can
then perceive the situation as analogous to that of conventional consumer theory.
Just as effective demand reflects the desire of an individual to purchase a service,
so governmental response to need reflects society’s desire to purchase transport
services for certain of its members.

4.12 The Valuation of Travel Time Savings

The importance of travel time in transport economics should by now be appar-

ent. While the action of travel involves some time costs, it is perhaps more useful
to consider travel time in this chapter that focuses on demand and benefits rather
than costs. This is because travel or transport time savings are normally con-
sidered to be a major component of any scheme designed to improve transport
efficiency. As we see in Chapter 11, travel time savings form the major component
of inter-urban road investment benefits – a situation also found in most fields
of transport. For reasons of comparability with other forms of benefit, a vast
amount of intellectual energy has gone into devising methods of placing money
values on such benefits.
Two quite distinct methodologies have been developed for time evaluation,
the distinction being made between time saved in the course of employment and
time saved during non-work travel (including commuter trips when fixed working
hours are involved). The distinction is drawn because work time involves lorry
drivers, seamen, pilots, etc., not simply in giving up leisure but also in incurring
some actual disutility from the work undertaken. Hence, if they could do the
same amount of work in less time, these people would be able to enjoy more
leisure and suffer less disutility (or the employer must pay them more to encour-
age a continuation of the same work hours with a higher output). Savings in

BUTTON 9781786435668 PRINT.indd 112 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 113

non-work time do not, by definition, reduce the disutility associated with work
and, consequently, although more leisure may be enjoyed, they are likely to be
valued below work travel time savings.
The valuation of work travel time (which embraces all journeys made when
travelers are earning their living) is made simpler if we accept the traditional
economic idea that workers are paid according to the value of their marginal
revenue product. On this basis, the amount employers pay workers must be
enough to compensate them for the marginal time and disutility associated with
doing the job. Thus, it becomes possible to equate the value of a marginal saving
in work travel time with the marginal wage rate (plus related social payments and
overheads). An alternative way of arriving at this cost-savings approach is by
reflecting upon the opportunity costs involved – as Benjamin Franklin once said,
‘remember that time is money’. Time savings at work permit a greater output to
be produced within a given time period, which, again drawing on the marginal
productivity theory of wage determination, will be reflected in the marginal
wages paid. Official policy in many countries is to value work travel time savings
as the national average wage for the class of the transport user concerned plus the
associated costs of social insurance paid by the employer and a premium added
to reflect overheads.
A major problem with the wage equivalence approach is that it assumes
employees consider the disutility of travel during work to be the same as the
disutility of other aspects of their work that they may be required to undertake
if travel time is reduced. In many cases workers may consider the travel much less
arduous than these alternative tasks. This implies that savings in work time travel
should, in such cases, be valued at less than the wage rate plus additions. Also,
some people may view travel time as highly productive – many rail and air travel-
ers, for instance, certainly work on their journeys – suggesting that reduced travel
time would not significantly alter output. Even time spent in car travel can be used
to complete mental tasks – for example, Fowkes et al. (1986) found that about 3
percent of business travel time by car was spent working.
More recently, analysis of San Francisco Bay Area travelers found that,
overall, commute time is not unequivocally a source of disutility to be minimized,
but rather offers some benefits (such as a transition between home and work).
Most people were found to have a non-zero optimum commute time, which can
be violated in either direction – that is, it is possible (although comparatively rare,
occurring for only 7 percent of the sample) to commute too little (Redmond and
Mokhtarian, 2001). Examining the wage rate alone ceases to be a useful measure
of work travel time savings in such cases.
While labor economics provides a useful foothold to obtain values of work
travel time, rather more empiricism is required in the evaluation of non-work
travel time. The behavioral approach involves using revealed preferences to con-
sider trade-off situations that reflect the willingness of travelers to pay in order to
save time. In other words, if a person chooses to pay $X to save Y minutes, then
he/she is revealing an implicit value of time equal to at least $(X/Y) per minute.
More formally:

BUTTON 9781786435668 PRINT.indd 113 27/04/2022 11:07

 114 TRANSPORT ECONOMICS, 4TH EDITION

∆A = – α∆T – β1∆C + β2∆E (4.3)

where the relative attractiveness of the alternatives, ∆A, is the difference between
the travel time cost, ∆C, and travel time, ∆T. The coefficients α and β1 are both
negative because a decrease in either the time or cost of a trip is seen to generate
positive utility.
Empirical studies attempting to value non-work travel time in this way have
looked at several different trade-off situations (a survey is offered by Waters,
1992), notably when travelers have a choice among:

• route;
• mode of travel;
• speed of travel (by a given mode over a given route);
• location of home and work; and
• destination of travel.

The standard approach in these trade-off studies is to employ a simple equa-

tion of the general form:

ey
P1 =
where y = α0+ α1 (t1t2) + (c1t2) (4.4)
1+ ey ( )
where:

P1 is the probability of choosing mode (route, etc.) 1;

y is choice of mode (route, etc.); takes value of 1 for mode (route, etc.) 1 and
0 for mode (route, etc.) 2;
e is an exponential constant;
ti is the door-to-door travel time by the j' mode (route, etc.); i = 1,2;
ci is the door-to-door travel cost by the jth mode (route, etc.); i = 1,2; and
α0, α1, and α2 are constants to be estimated.

A value of time can then be inferred by looking at changes in the dependent vari-
able that result from a unit change in either the time or the cost difference. Strictly,
it may be found as the ratio α1/α2 in equation (4.4).
Many of the early studies of non-work travel time concentrated on urban
commuter trips because there was pressure at the time to provide information for
cost–benefit analysis (CBA) of urban transport investment plans. In consequence,
mode and route choice evaluation techniques were developed to a high level of
mathematical sophistication. Early work by Michael Beesley (1965) specifically
employed discriminant analysis to examine the journey to work-mode choices of
employees at the UK Ministry of Transport during 1965–66. This technique essen-
tially finds the trade-off value of time that minimizes the number of misallocations
of commuters to alternative modes with different time and cost characteristics.
The upper part of Figure 4.6 considers the possible options available to a
traveler when choosing between two modes. If the figure relates to Mode A, and

BUTTON 9781786435668 PRINT.indd 114 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 115

only time savings and costs are important then Option 3 dominates. Conversely,
Mode B will dominate Option 2 (where Mode A is slower and more expensive). But
in some cases, travelers may not have such a clear-cut choice and will have to elect
between Options 1 and 4 where there are trade-offs between time and money costs.
The lower portion of the figure gives some actual decisions by individuals
regarding the choices that are made between the two modes (filled and unfilled
circles). The line AB (where ∆C = θ∆T) provides the partitioning that leads to
the minimum number of ‘misclassifications’ and reflects the trade-off between
changes in travel time and travel costs being made. The slope of this line, θ, identi-
fies the value of travel time savings.
The discriminate function used to determine the minimum number of mis-
specifications takes the form:
n m
Zij = ∑αk [f (Xki, Xkj)] + ∑ β1U1 (4.5)
K=1 1

where:

Xki, Xkj are the values of the kth attributes of the ith and the jth trip packages;
Ul are user attributes;
αk are parameters associated with the alternative systems;
βl are parameters associated with user characteristics; and

C

Option 1 Option 2
More expensive/faster More expensive/slower

T

Option 3 Option 4
Less expensive/slower Less expensive/faster

C
A

T

Figure 4.6 The ‘Beesley graph’ of travel time–cost trade-off

BUTTON 9781786435668 PRINT.indd 115 27/04/2022 11:07

 116 TRANSPORT ECONOMICS, 4TH EDITION

f (Xki,Xkj) is a function that may take either of the forms (Xki – Xkj) or
(Xki/Xkj).

Beesley estimated that commuter trip time savings were valued at between 30
and 50 percent of the gross personal income of the commuters. One of the main
problems with this pioneering study, though, was that it failed to isolate in-vehicle
travel time from the other components of journey time (for example, waiting and
walking time). This defect was subsequently remedied in a larger study of mode
choice in Leeds undertaken by David Quarmby (1967), which embraced seven
variables including walking and waiting time as well as in-vehicle time. The find-
ings indicated that savings in walking and waiting times are valued at between
two and three times savings in in-vehicle time – parameters that have proved to be
remarkably robust over the years.
Table 4.8 provides details of the non-work time values that have been revealed
in these and subsequent studies. Many of these moved on from the discriminate
approach to adopt logit and probit specifications (discussed in the context of car
ownership modeling in Chapter 11) that offer more flexibility and more statistical
measures of the quality of a model.
Model specifications can influence the travel time values obtained and care is
needed in their selection. There is also a tendency to lump all non-work travel time
together because it is not directly related to theories of labor productivity, but this
may produce aggregation bias. Mark Wardman (2001), for example, in reviewing
British value of travel time savings work found that the valuation of commuting
time savings are about 35 percent higher than for leisure travel in London and
south-east England, although only about 14 percent higher for work pertaining
to other parts of the country.
Stated preference, or experimental, techniques, whereby hypothetical ques-
tions are posed to travelers to gain information about trade-offs they would be
willing to make have become more common in recent years. A pioneering example
was that of Lee and Dalvi (1969), who used questionnaires, rather than looking
at actual choices, to discover the level of fare increase required before passengers
switched from one mode of public transport to an alternative. Interestingly, in
Manchester it was found that in-vehicle time, walking time, and waiting time were
not separately important and travelers did not distinguish between them. Overall,
it was estimated that non-work travel time was being valued at 15 to 45 percent
of hourly income.
While most urban studies have tended to focus on mode choice decisions, the
evaluation of non-work inter-urban travel time has tended to concentrate rather
more on route and speed choice situations – although imperfections in travelers’
knowledge of the latter make speed choice trade-offs suspect. Pioneering work on
route choice by Paul Claffey (1961) looked at choices made between tolled and
free roads in the United States and attempted to allow for differing accident rates
and levels of driver discomfort when assessing the time/money-cost trade-offs.
Mathematical weakness limits the value of this specific model but subsequent
reworking suggests time differences are unimportant in route choices of this type.

BUTTON 9781786435668 PRINT.indd 116 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 117

Table 4.8 Computation examples of estimated values of travel time savings

Study Country Value of time as % Trip purpose Mode

of wage rate

Beesley (1965) United 33–50 Commuting Auto

Kingdom
Quarmby (1967) United 20–25 Commuting Auto, transit
Kingdom
Thomas & United States 86 Inter-urban Auto
Thompson (1970)
Lee & Dalvi (1971) United 30 Commuting Bus
Kingdom
40 Commuting Auto
Talvitte (1972) United States 12–14 Commuting Auto, transit
Hensher & Hotchkiss Australia 2.70 Commuting Hydrofoil,
(1974) ferry
Kraft & Kraft (1974) United States 38 Inter-urban Bus
McDonald (1975) United States 45–78 Commuting Auto, transit
Guttman (1975) United States 63 Leisure Auto
145 Commuting Auto
Hensher (1977) Australia 39 Commuting Auto
35 Leisure Auto
Nelson (1977) United States 33 Commuting Auto
Edmonds (1983) Japan 42–49 Commuting Auto, bus,
rail
Deacon & Sonstelie United States 52–254 Leisure Auto
(1985)
Hensher & Truong Australia 105 Commuting Auto, transit
(1985)
Guttman & Menashe Israel 59 Commuting Auto, bus
(1986)
Fowkes (1986) United 27–59 Commuting Rail, bus
Kingdom
Chui & McFarland United States 82 Inter-urban Auto
(1987)
Mohring et al. (1987) Singapore 60–129 Commuting Bus
Cole Sherman (1990) Canada 93–170 Commuting Auto
116–165 Leisure Auto

Source: Waters (1992). This paper contains the full references to the studies cited.

Dawson and Everall (1972), using a modification and looking at route choices of
motorists traveling between Rome and Caserta and between Milan and Modena
where the autostrada offered alternatives to ordinary trunk roads, found that
observed trade-offs indicate that commuting and other non-work travel time was
valued at about 75 percent of the average wage rate.
It is clear from the selection of studies cited above that non-work time
savings are, indeed, valued below the wage rate, but it is equally clear that the
values obtained from economic studies are extremely sensitive to the assumptions
made and the estimation technique employed.

BUTTON 9781786435668 PRINT.indd 117 27/04/2022 11:07

 118 TRANSPORT ECONOMICS, 4TH EDITION

David Hensher (1979) goes further and points to the rather strong assump-
tions that are implicit in the not-uncommon practice of taking time values
obtained from, say, a mode choice study and employing them in route or speed
choice situations. He also questions whether enough consideration is given to the
composition of time savings beyond the in-vehicle/waiting-time split and to pref-
erences between constant journey speed (with a lower average) and faster, variable
speeds (with a higher average). There is also the common practical problem that
it is difficult to separate the influence of comfort and convenience factors from
travel time savings.
In Britain the UK Department of Transport and its successors have since
the 1960s recommended standard values of time for transport analysis purposes.
The use of standard figures is to encourage uniformity in investment appraisal.
While the work–travel time figures are open to only minor criticisms (and even
quite major errors here would seem unlikely to distort decisions), the use of
standard non-work travel time values has met with more serious criticism.
Empirically, non-work travel time values have generally been shown cor-
related with income levels (Ian Heggie’s 1976 work being one of the few excep-
tions), but on occasions an average value across all income levels has been
used for policy formulation purposes. The argument supporting this ‘equity’
value is that if time values were directly varied with income this would tend
to bias project selection towards projects favoring the higher-income groups.
In evaluation, the travel time savings of such groups would automatically be
weighted more heavily than those of the less well-off. The Leitch Committee
(UK Department of Transport, 1978) in the United Kingdom, however, rejected
this line of argument because it is not consistent with the way other aspects of
transport investment are evaluated. Since the overall distributional effects of
transport investment may be treated more directly in the appraisal process (see
Chapter 11), the notion of ‘equity’ values was rejected in favor of income-based
time evaluations.
Even if generally acceptable values of travel time could be obtained, there
are still difficulties associated with using them. One of the major problems is
that some projects can result in a small number of large time savings while others
produce a magnitude of extremely small savings. The problem becomes one of
deciding whether 60 one-minute savings are as valuable as (or more valuable than)
one saving of an hour’s duration. It could be argued that travelers, especially over
longer routes, tend not to perceive small time savings or cannot utilize such times.
If this is so, it would tend to make urban transport schemes appear less attractive
vis-à-vis inter-urban ones because the main benefits of urban improvements have
been small time savings spread over thousands of commuters.
One suggestion is that a zero value should be adopted in these contexts:
small travel time savings with a positive value only being employed once a hold
level of saving has been reached (say ten minutes). This ignores the fact that small
time savings may, in some circumstances, be combined with existing periods of
free or idle time to permit substantial increases in output or in leisure enjoyment.
Further, if there are non-linearities in the value of travel time this would imply

BUTTON 9781786435668 PRINT.indd 118 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 119

that widely used trade-off methods of time evaluation based upon average time
savings must be giving biased estimates of the value of travel time. The debate
over the handling of small travel time savings is unlikely to be resolved easily.
Finally, there is the question of whether revealed or stated preference meas-
ures of the value of travel time savings are more ‘accurate’. There is some indica-
tion from Table 4.9, which provides details of the travel time savings values used
in the United Kingdom, that, in fact, they are relatively consistent – a finding
consistent with Wardman’s (2001) review. The table relates to the values used in
COBA, the program used to assess road investments. The method of determining
the values changed between 1988 and 1994 from a revealed to a stated preference
approach. While there are some clear differences, the broad pattern is very similar
considering the very different methods of estimation used.
Transport studies in less developed countries tend to adopt the convention
that while work travel time savings should be given a monetary value based upon
the cost-savings approach (although the wage rate is generally modified to allow
for imperfections in the local labor market), savings in non-work travel time –
especially in rural areas – are given a zero value. The justification for this is that
the prime objective of improving transport infrastructure in the Third World
is to assist in economic growth and thus the emphasis should be exclusively
concentrated on economically productive schemes – leisure time is not seen as
‘productive’. Thomas (1979) pointed to an anomaly, however, when this argu-
ment is carried into practice. While non-work travel time savings in rural areas
are ignored, savings in vehicle operating costs for such travel is not. Not only is
this inconsistent but it also has important distributional implications because the
main beneficiaries of low operating costs are almost invariably high-income car
owners.

4.13 The Demand for Car Ownership

While the demand for cars is not strictly a direct transport matter, it is in some
ways more to do with industrial economics and the demand for consumer dura-
bles. The importance of the automobile in affective travel behavior, land-use pat-
terns, and the environment makes it a matter of considerable interest to transport
economists (Dargay and Giuliano, 2006).
Car ownership, as we saw in Chapter 2, has risen considerably since the
First World War with only brief halts during periods of major military conflict
and occasional decelerations in the trend during periods of macroeconomic
depression. This upward trend is not unique to the United States or the United
Kingdom but is also to be found in all other countries irrespective of their state of
economic development, or the nature of their political institutions. The upward
trend in car ownership is the result of both the considerable benefits that accom-
pany car availability (notably improved access and greater flexibility of travel)
and the long-term increases in income enjoyed by virtually all countries since
the Second World War. The ‘demonstration effect’ has tended to accelerate the

BUTTON 9781786435668 PRINT.indd 119 27/04/2022 11:07

 120 TRANSPORT ECONOMICS, 4TH EDITION

Table 4.9 Overall values of time used in United Kingdom road investment analysis

1994 study COBA COBA (1988) COBA (1988) COBA (1988)

(1988) full income income income growth
growth elasticities adj.10

Driver – – – – –
Commuting 5.4 – – – –
Other 4.4 – – – –
Non-work total 4.9 5.6 6.2 5.9 –
Employee
  (Business) 6.7 – – – –
Employer
  (Business) 14.7 – – – –
Working total 21.4 19.1 20.9 20.6 –
Passenger – – – – –
Commuting 6.0 – – – –
Other 3.1 – – – –
Non-work total 4.03 5.6 6.2 5.9 –
Employee
  (Business) 6.7 – – – –
Employer
  (Business) 14.7 – – – –
Working total 21.4 15.9 17.4 17.1 –
LGV – – – – –
Hire & reward 45.0 – – – –
Own account 35.0 – – – –
Total 40.06 – – – –
COBA 9 – 19.35 – – 20.3
OGV/HGV – – – – –
Hire & reward 45.0 – – – –
Own account 35.0 – – – –
Total 40.06 – – – –
COBA 9 – 14.05 – – 14.7
PSV – – – – –
Scheduled coach 50.0–60.0 – – – –
Motorway charter 23.0–33.0 – – – –
Scheduled bus 17.0 – – – –
Trunk road
  charter 0.0–25.0 – – – –
Total PSV – 84.17 – – 88.4

Notes: All results are in 1994 £0.01/minute.

Source: Gunn (2008).

process in less developed countries as attempts are made to emulate the consump-
tion patterns of more affluent nations.
Considerable effort has been focused on exploring both the rate of increase
in vehicle ownership and reasons why this should differ between countries and
between areas within a single country. Information on the underlying demand
functions is sought for a variety of reasons. Car manufacturers need to know the

BUTTON 9781786435668 PRINT.indd 120 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 121

nature of changing demands for new vehicles, both within the country and within
their export markets, and to be able to forecast likely changes in the type of vehi-
cles which are wanted. While work in this area often sheds some useful light on
the workings of the car market, it is only of limited use to transport economists.
By contrast, central government is generally more interested in the aggregate
number of vehicles in the country, mainly for road planning purposes, but also,
to a lesser extent, to assist the finance authority in its fiscal duties. Regional vari-
ations, which can be quite pronounced within countries (Table 4.10), are also of
interest for strategic planning purposes, particularly for developing national and
international road networks.
The theory underlying much of the early forecasting work looking at car
ownership levels is closely akin to the management theory of a ‘product life’
cycle, where a product has a pre-determined sales pattern almost independ-
ent of traditional economic forces, although taste and costs are not altogether
absent from the model. The logistic curve fitting model developed by the
United Kingdom’s Transport Research Laboratory (then the Road Research
Laboratory and subsequently the Transport and Road Research Laboratory
over the period), in its basic form, treats per capita vehicle ownership as a
function of time (Figure 4.7), with the ownership level following a symmet-
ric, sigmoid growth path through time until an eventual saturation level is
approached (Tanner, 1978). Broadly, it is argued that long-term growth in
ownership follows a predictable diffusion process. Initially, high production
costs and unfamiliarity will keep sales low, but after a period, if the product
is successful, economies of scale on the supply side coupled with bandwagon
effects on the demand side would result in the take-off of a comparatively rapid
diffusion process. Finally, there is a tailing-off as the market becomes saturated
and everyone wishing to own a car does so.

Table 4.10 Regional variations in Great Britain’s car ownership patterns

Region Automobiles per 1,000 population Average vehicle age

North-east 403 6.0

North-west 469 6.1
Yorkshire & the Humber 432 6.2
East Midlands 487 6.7
West Midlands 510 6.4
East of England 513 7.1
London 345 7.4
South-east 542 6.9
South-west 522 7.5
Total England 471 6.8
Wales 483 6.9
Scotland 433 5.8
Great Britain 480 6.7

Source: UK Department for Transport (2007).

BUTTON 9781786435668 PRINT.indd 121 27/04/2022 11:07

 122 TRANSPORT ECONOMICS, 4TH EDITION

Sales
Introduction Take-off Maturity Decline

0 Time
Figure 4.7 The life cycle of a product

The initial form of model was a logistic curve that simply traced out per capita car
ownership over time using national car ownership data (Figure 4.8). The logistic
equation took the form:
α
C=
(4.6)
1 + βexp ( −αct )
where:

C is per capita car ownership;

t is a time trend; and
α, β, and c are parameters to be estimated.

For calculation purposes it was necessary to feed an exogenous value of ‘satura-

tion level’ (the parameter α) into the equation. This was done either by judgment
based upon such things as the proportion of the population who would be able
to drive if there were no constraints on car ownership (for example, taking out
the young and the infirm from the population) or by calculation ‘mean reversion’
parameters. The latter involved taking two periods of regional cross-sectional
data on car ownership levels and regressing the levels of ownership in the base
year against the changes in those levels. The constant term in the resultant equa-
tion (that is, when ΔC = 0) is, de facto, the saturation level. Once the overall equa-
tion is fitted using up to point t in the figure then it is possible to extrapolate to
produce forecasts.
Historically, the Road Research Laboratory’s (RRL) extrapolative
approach provided relatively good forecasts in the United Kingdom in the
l960s (Table 4.11), but tended to be less reliable in later years and to suffer
from ­over-prediction. Some of the difficulties were associated with problems

BUTTON 9781786435668 PRINT.indd 122 27/04/2022 11:07

 123
THE DEMAND FOR TRANSPORT ­

Cars
per capita
Saturation level

Forecast
C

Observed growth
path

0 t Time

Figure 4.8 The logistic growth curve approach to car ownership forecasting

of estimating key parameters such as the ultimate saturation level, or with the
correct configuration of the growth curve – in later work, a power function,
which does not imply a symmetric sigmoid growth path, replaced the logistic
(equation (4.7)). In addition, the time variable was supplemented by other vari-
ables to reflect changing costs of motoring and income. The resultant extrapola-
tive model is still widely used to gain general insights into car ownership and to
gauge the implications of policies regarding motoring costs, such as fuel taxa-
tion, on the future carpool:

S (4.7)
Ct = − k1S − k2 S− k3St
 S − C0   Yt   Pt 
1+      e
 C0   Y0   P0 

where:

C0 is car ownership per person at time t = zero;

S is the saturation level to which C is asymptotic as t increases;
Yt is income per head;
Pt is ‘cost of motoring’; and
k1, k2, and k3 are constants.

Despite these limitations, this type of general framework can be useful when
there are serious data constraints. Button et al. (1993) examined factors affecting
car ownership levels across a range of very-low-income countries where there
is little information on many of the economic variables that one would like to
consider – in this case just data on income and time were used in a quasi-logistic
equation.

BUTTON 9781786435668 PRINT.indd 123 27/04/2022 11:07

 124 TRANSPORT ECONOMICS, 4TH EDITION

Table 4.11 Comparison of actual car ownership and RRL forecasts

Year of Base year for (Forecast annual growth in cars per Forecast car
publication calculation capita)/actual annual growth in pool 1975 pool 1975

1962 1960 1.14 1.13

1965 1964 1.57 1.57
1967 1966 1.67 1.68
1969 1968 1.84 1.84
1970 1969 1.66 1.66
1972 1971 1.62 1.58

Source: Button et al. (1980).

At least one school of thought rejects the underlying extrapolative philoso-

phy as inadequate. It has been argued that car ownership forecasting should
be based only upon explicit economic variables such as income and vehicle
prices rather than ‘proxy’ variables such as time (the counterargument being
that ‘time’ somehow encapsulates changing tastes). The final RRL forecast-
ing framework attempted to meet this criticism by incorporating economic
variables, but a time trend is still retained and the income and vehicle operating
cost elasticities are not estimated internally within the model but derived from
‘external’ sources. The demand model developed by Bates et al. (1978), as part
of a larger regional highway traffic model (RHTM) in the United Kingdom, in
contrast is based entirely upon ‘causal’ variables, and all the relevant elasticities
are estimated directly within the forecasting model. This type of framework is
now widely used in economic forecasting, especially at the local level, but also for
national studies.
The basis for the underlying model is in economic utility theory. It is
assumed that the utility, U, that a household, n, gets from owning a thing, j, is
represented by:

Uj.n = Vj.n + εj.n (4.8)

where Vj,n is a deterministic component and εj,n is a random factor.

For estimation purposes, a multi-nomial logit model is generally used in the
car ownership context, although in some cases alternatives are adopted, especially
for local analysis where the interest may be in forecasting the number of cars per
household rather than the national carpool. The logit model takes the general
form:
p
Vn = β1n + ∑βnk X nk
(4.9)
k =1

where:

p is the number explanatory variables;

β1n is the ‘alternative-specific’ constant associate with n vehicles;

BUTTON 9781786435668 PRINT.indd 124 27/04/2022 11:07

 125
THE DEMAND FOR TRANSPORT ­

βnk is the ‘weight’ for variable k in the utility of having n vehicles; and
Xnk is the value of variable k for vehicle ownership level n.

To show what this means in practical terms, a simple form of the actual type
of equation calculated is provided by Fairhurst (1975) in the context of car own-
ership, not in the United Kingdom as a whole, but in London:

P
log 0 = c + b logY + d logH + f log ( B + 1) + g log ( R + 1)
(4.10)
1 − P0

where:

P0 is the proportion of households without a car;

H is persons per household;
B is a bus public transport index; and
R is a rail public transport index.

The data used in these types of models are not the time-series registration
statistics employed by the RRL, but rather a series of cross-sectional sets of
statistics generally obtained from household surveys. For forecasting it is neces-
sary to be able to predict reliably the level and distribution of future income and
other variables. The approach also concentrates on the probability of households
having a certain level of vehicle ownership rather than, as with the time-trend
model, on forecasting the national average or regional ownership level; this con-
forms more closely to other recent trends in transport demand forecasting (see
Chapter 12). In the case of the United Kingdom, for example, where the geo-
graphical size of the country makes national forecasting meaningful, this type of
economic model makes use of data from the Family Expenditure Survey and the
National Travel Survey. It relates household car ownership to household income,
number of adults in a household, number of children in a household, number of
employed household members, number of retired people, and the geographical
nature of the location of the house.
This type of economic framework is now regularly used in aggregate analy-
sis. Dargay (2002), for example, examines trends in the growth of car stocks over
past decades and make projections of its development over the next 25 years for
82 countries at different levels of economic development, from the lowest (China,
India, and Pakistan) to the highest (the United States, Japan, and European
nations). The countries account for 85 percent of the world population and 93
percent of the vehicle stock. The projections are based on estimates of a dynamic
time-series model and an S-shaped function to relate the vehicle stock to GDP.
The estimates are used in conjunction with assumptions concerning income and
population growth, to produce projections of the growth in the vehicle stock on
a year-by-year basis.
Differences in the geographical demand for car ownership interest transport
planners both because they need to be able to forecast future demand for links in

BUTTON 9781786435668 PRINT.indd 125 27/04/2022 11:07

 126 TRANSPORT ECONOMICS, 4TH EDITION

the local road network and because, where ownership is low, social commitments
may require that alternative public transport is provided. These models often
need to be a little more detailed in terms of the variables included. For example, it
seems likely that spatial variations at the local level may, once allowance has been
made for differing income and demographic factors, be explained in terms of the
quality of local transport services, as in the case of Fairhurst’s work on London.
Good, uncongested roads combined with poor public transport increases
the demand, ceteris paribus, for private car ownership. Regional econometric
studies of car ownership have attempted to reflect this cost-of-transport effect by
incorporating variables such as residential density in their models (it being argued
that a densely populated area is normally well served by public transport while
motoring is adversely affected by the higher levels of traffic congestion). More
sophisticated local models have shown the frequency of public transport services
to influence car ownership rates, at least slowing the rise in automobile ownership
but not reversing it. If these studies are correct then there is some evidence that
the long-term growth in car ownership may be contained by improving public
transport services, although from a policy point of view the overall cost of such
actions needs to be fully assessed.

4.14 What Does ‘Behavioral Economics’ Tell Us?

Most of our analysis so far of demand, excepting the discussion of habit and
hysteresis, and to some extent need, is founded on the idea that individuals act as
if they were homo economicus. It assumes that consumers have full information,
make cognitive, non-emotional, optimal decisions, and that these decisions are
marginal and independent of the actions of others. Psychologists however, are
increasingly finding that people often exhibit inertia in their actions, they can be
philanthropic, there is asymmetric information in their transactions, and they
search within boundaries. While in many cases it is quite reasonable to use the
conventional concept of homo economicus when discussing and quantifying many
aspects of transport demand – people will buy less fuel if oil prices rise – in some
circumstances this is not the case.
Because of this, there has been an increased interest shown by transport
economists in behavioral economics (Garcia-Sierra et al., 2015). Broadly, this
involves embedding study of the effects of psychological, cognitive, emotional,
cultural, and social factors in the decisions of individuals and institutions, and
how those decisions vary from those assumed in neoclassical economics. Put
another way, these factors can mean that individuals appear to act irrationally
when contrasted with the conventional neoclassical economics we are familiar
with.
The transport decision considered may be of a long-term nature: choice of
residential location or employment, whether to acquire a driving license, own
a car or not, etc., or shorter-term choices of destinations, modes to use, routes
to take, and the timing of trips. In practice, any of these decisions are complex

BUTTON 9781786435668 PRINT.indd 126 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 127

and made in the face of imperfect information. This often means resorting to
heuristics and the adoption of mental shortcomings. Human behavior as a result
deviates from that postulated in standard neoclassical models. Behavioral eco-
nomics, in recognition of this, has begun to play a role in policy-making, includ-
ing in transport policy. In 2010 a Danish Network was created and a year later
the United Kingdom Department for Transport published a Behavioral Insights
Toolkit in 2011.
The mounting intellectual interest in behavioral economics is confirmed in
the awarding of the Nobel Prize for Economics to such as Herbert Simon, George
Akerlof, Daniel Kahneman, and Richard Thaler. Thaler’s work on nudging, basi-
cally any aspect of choice architecture that alters people’s behavior in a predict-
able way without forbidding any options or significantly changing their economic
incentives, and his best-selling book, Nudge: Improving Decisions on Health,
Wealth, and Happiness (Thaler and Sunstein, 2008), have brought behavioral eco-
nomics to the public eye.
The approach is essentially inter-disciplinary; Kahneman, for example, was
trained in psychology, and unlike many neoclassical economists does not have
roots in engineering. While, as we see later, behavioral economics has impacted
on some aspects of the supply theory of economics, including Leibenstein’s work
on X-efficiency, Robert Metcalfe and Paul Dolan (2012) and Filippini et al. (2021)
highlight the ways in which it is increasingly influencing the analysis of travel
demand and pushing through policy initiatives.
Examples of this, in terms of cartography and nudges, are when words like
freeway, route, highway, parkway, etc., are used as designators to convey different
impressions of the quality and nature of a road. In some cases, the categoriza-
tion is used to nudge drivers to use specific routes. Differing colors for roads on
an old-fashioned map, or the outlining of some, serve the same purpose. Google
Maps with brighter colors highlighting more congested routes has updated this
behavioral tool further. As Zhan Guo (2011) has demonstrated, the schematic
transit maps of the kind used by the London Underground, which distort infor-
mation about distances between stations to give a clearer perspective of possible
linkages between lines, influence routes, and modes chosen. Results show that
the elasticity of the map distance is twice that of the travel time, suggesting that
passengers often trust the Tube map more than their own travel experience when
deciding the best travel path. Intelligent transport systems, involving indicators of
parking availability, act as ‘boosts’ to those whose apparently irrational behavior
of cruising around to seek a parking place had, in fact, been the result of inad-
equate information.
Kahneman’s ‘prospect theory’, when individuals assess their loss and gain
perspectives in an asymmetric manner – for example, when losing $1,000 could
only be compensated by the pleasure of earning $2,000 – has been applied to
travel demand models and other markets (Li and Hensher, 2011). Figure 4.9
depicts a situation where a driver could take a ‘short-cut’ that could reduce a trip
time by X minutes (+X) but, if there is congestion, may alternatively increase it
by X minutes (–X). If the driver were risk-neutral then they would be indifferent

BUTTON 9781786435668 PRINT.indd 127 27/04/2022 11:07

 128 TRANSPORT ECONOMICS, 4TH EDITION

Exhibit   When London Underground workers go on strike

On February 5 and 6, 2014 a labor dispute caused a partial shutdown of London’s

underground. While some stations were closed, some stayed open. This allowed Shaun
Larcom et al. (2017) to look the subsequent use of the ‘Tube’ by those travelers who were
seriously affected by the strike and those who were not.
London’s Underground has the eighth-largest number of stations in the world (270) and
carries the 11th-largest number of passengers. In many cases it offers alternative routing
options between an origin and a destination. The network is described by a schematic map
showing only relative positions of Tube stations and of interacting surface rail stations.
The distances between stations are thus distorted. A labor dispute led to a 48-hour partial
closure of the systems strike from the evening of the February 5, 2014. Individual workers
were left to decide their actions but enough went on strike to close 171 stations. The
Circle, Bakerloo, and Waterloo & City lines were also closed. This was the first major
disruption to the system in three years.
Larcom et al. gathered data from January 19 to February 15 for all modes of public transport
excluding buses. Because Oyster cards – a rechargeable ‘pay-as-you-go’-type plastic card –
were used for about 80 percent of trips and also had individual numbers, these were used
for route tracking of trips during the rush hour each day. Some idea of the scale of the
data used can be seen since there were 19,301,730 data entries on January 24. Two panels
of users were isolated: one (the unconditional) whereby no information of post-strike
travel was used and the other (the balanced) where both pre- and post-travel details were
included. This allowed comparisons between before and after behavior using difference-in-
differences techniques.
It is found that the benefits enjoyed by travelers forced to experiment with alternative
routing because of station closures, who remained with the new travel pattern (about 5
percent of the total), conservatively gained about 20 seconds per journey. Put another way,
they would gain about four hours of commute time over 3.3 years. At an aggregate level,
the time saved because of long-term modal adjustments was larger than the time lost due to
the strike.
These findings suggest that many commuters had failed to find their optimal travel pattern
before the strike and, by exploration, found a better, if not always optimal, pattern that
continued to be used once the system again became fully operational. Explanations lie in the
noisy information available, the stylized nature of the Tube maps, and the fact that different
train lines travel at different speeds. The strike essentially provided a ‘boost’ to look at
alternatives.
See also: S. Larcom, F. Rauch, and T. Willems (2017) The benefits of forced experimentation:
striking evidence from the London underground network, Quarterly Journal of Economics,
132(4), 2019–55.

BUTTON 9781786435668 PRINT.indd 128 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­129

Value

VG
Loss Gain

–X +X

VL

Figure 4.9 The asymmetric nature of valuing a gain and loss

between the options. But in most cases, because of the lower valuations drivers
put on a minute gained as opposed to a minute lost, the value of the potential
time savings from diverting (VG) are less than the potential loss (VL); the time-
savings/value trade-off curve is stepped.
A common example of this, when it comes to trip-making, is that arriving
early is not seen as being as important as arriving the same number of minutes
late. A similar situation arises in terms of travel insurance, relative to the actual
probabilities of a fatal accident: individuals over-insure themselves against an
aviation accident and under-insure against an automobile accident.
The idea that individuals are only concerned about the effects of their actions
on themselves is also coming under question in some contexts; the very notion of
‘need’ as an allocative mechanism is one indication of this. There is also plenty of
evidence from drivers’ behavior that many give way to other drivers when there is
no self-benefit from doing so.
When it comes to speeding and traffic safety, Peer (2011) asked drivers to
estimate how much time can be saved by increasing speed. The drivers gener-
ally under-estimate the time saved when increasing from a relatively low speed
and over-estimate the time saved when increasing from a relatively high speed.
This time-saving bias affects speed predictions in conventional demand models.
Indeed, in predicting drivers’ personal speed choices, the time-saving bias was
second only to the frequency of committing ordinary violations and outper-
formed more conventional variables such as drivers’ age, gender, education, the
number of years a driving license has been held, monthly driving distances, and
drivers’ prior speeding violations and crashes.
Behavioral economic approaches are gradually finding that travel demand
is particularly affected by novel, accessible, and simple measures. Simplicity is
important because people respond more to things they understand. The increasing

BUTTON 9781786435668 PRINT.indd 129 27/04/2022 11:07

 130 TRANSPORT ECONOMICS, 4TH EDITION

amount of information that providers now accumulate through big data suggests
that fares and services can be more fine-tuned to customers’ demands.

References

Acutt, M.Z. and Dodgson, J.S. (1995) Cross-elasticities of demand for travel, Transport
Policy, 2, 271–7.
Alperovich, G. and Machnes, Y. (1994) The role of wealth in demand for international air
travel, Journal of Transport Economics and Policy, 28, 163–73.
Banister, D., Bould, M., and Warren, G. (1984) Towards needs-based transport planning,
Traffic Engineering and Control, 25, 372–5.
Basso, L.J. and Oum, T.H. (2007) Automobile fuel demand: a critical assessment of
empirical methodologies, Transport Reviews, 27, 449–84.
Bates, J.J, Gunn, H.F., and Roberts, M. (1978) A disaggregate model of household car
ownership, Department of the Environment and Transport Research Report 20,
HMSO.
Beesley, M. (1965) The value of time spent in travelling: some new evidence, Economica,
32, 174–85.
Bendtsen, P.R. (1980) The influence of price of petrol and of cars on the amount of auto-
mobile traffic, International Journal of Transport Economics, 7, 207–13.
Blasé, J.R. (1980) Hysteresis in travel demand, Transportation Planning and Technology,
6, 109–16.
Brown, S. and Watkins, W. (1968) The demand for air travel: a regression study of time-
series and cross-sectional data in the US domestic market, Highway Research Record,
213, 21–34.
Button, K.J., Fowkes, A.S., and Pearman, A.D. (1980) Disaggregate and aggregate car
ownership forecasting methods in Great Britain, Transportation Research Part A, 14(4),
263–73.
Button, K.J., Pearman, A.D., and Fowkes, A.S. (1982) Car Ownership Modelling and
Forecasting, Gower.
Button K.J., Ngoe, N., and Hine, J. (1993) Modelling vehicle ownership and use in low-
income countries, Journal of Transport Economics and Policy, 27, 51–67.
Claffey, P.T. (1961) Characteristics of passenger car travel on toll roads and comparable
free roads, Highway Research Board Bulletin, 306, Highway Research Board.
Dargay, J.D. (1993) Demand elasticities: a comment, Journal of Transport Economics and
Policy, 27, 87–90.
Dargay, J.D. (2002) Road vehicles: future growth in developed and developing countries,
Proceedings of the Institution of Civil Engineers, 151, 3–11.
Dargay, J.D. and Giuliano, G. (2006) Car ownership, travel and land use: a comparison of
the US and Great Britain, Transportation Research Part A, 40, 106–24.
Dawson, R.F.F. and Everall, P.P. (1972) The value of motorists’ time: a study in Italy,
Transport and Road Research Laboratory Report, LR.
de Bok, M., de Jong, G., Wesseling, B., Meurs, H., van Bekkum, P., Mijjer, P., Bakker,
D., and Veger, T. (2021) An ex-ante analysis of transport impacts of a distance-based
heavy goods vehicle charge in the Netherlands, Research in Transportation Economics,
101091.
De Vany, A.S. (1974) The revealed value of time in air travel, Review of Economics and
Statistics, 56, 77–82.
Fairhurst, H.M. (1975) The influence of public transport on car ownership in London,
Journal of Transport Economics and Policy, 9, 193–208.

BUTTON 9781786435668 PRINT.indd 130 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 131

Fearnley, N., Currie, G., Flügel, S., Gregersen, F.A., Killi, M., Toner, J., and Wardmand,
M. (2018) Competition and substitution between public transport modes, Research in
Transportation Economics, 69, 51–58.
Filippini, M., Kumar, N., and Srinivasan, S. (2021) Nudging adoption of electric vehicles:
evidence from an information-based intervention in Nepal, Transportation Research
Part D, 97, 102951.
Fowkes, A., Marks, P., and Nash, C.A. (1986) The value of business travel time savings,
Working Paper 214, Institute for Transport Studies, University of Leeds.
Garcia-Sierra, M., van den Bergh, J.C.J.M., and Miralles-Guasch, C. (2015) Behavioural
economics, travel behaviour and environmental-transport policy, Transportation
Research Part D, 41, 288–305.
Gilbert, C.L. and Jalilian, H. (1991) The demand for travel cards on London Regional
Transport, Journal of Transport Economics and Policy, 25, 3–30.
Glaister, S. and Lewis, D. (1978) An integrated fares policy for transport in London,
Journal of Public Economics, 9, 341–55.
Goodwin, P.B. (1977) Habit and hysteresis in mode choice, Urban Studies, 14, 95–8.
Goodwin, P.B. (1992) A review of new demand elasticities with special reference to short
and long run effects of price changes, Journal of Transport Economics and Policy, 26,
155–70.
Graham, J.D. and Glaister, S. (2004) Road traffic demand elasticity estimates: a review,
Transport Reviews, 24, 261–74.
Gronau, R. (1970) The Value of Time in Passenger Transportation: The Demand for Air
Travel, Columbia University Press.
Gunn, H.J. (2008) An introduction to the valuation of travel time-savings and losses,
in D.A. Hensher and K.J. Button (eds), Handbook of Transport Modeling, 2nd edn,
Elsevier.
Guo, Z. (2011) Mind the map! The impact of transit maps on path choice in public transit,
Transportation Research Part A: Policy and Practice, 45, 625–39.
Gwilliam, K.M. and Mackie, P.J. (1975) Economics and Transport Policy, Allen & Unwin.
Harrison, A.J. and Quarmby, D.A. (1969) The value of time in transport planning:
a review, in Report of the Sixth Round Table on Transport Economics, European
Conference of Ministers of Transport.
Heggie, I. (1976) A diagnostic survey of urban journey to work behaviour, in I. Heggie,
(ed.), Modal Choice and the Value of Travel Time, Clarendon Press.
Hensher, D.H. (1979) Formulating an urban passenger transport policy: a re-appraisal of
some elements, Australian Economic Papers, 18, 119–30.
Hillman, M. (1973) Personal mobility and transport policy, PEP Broadsheet
542.
Ippolito, R.A. (1981) Estimating airline demand with quality of service variables, Journal
of Transport Economics and Policy, 15, 7–15.
Isaac, J.K. (1979) Price and quality in road passenger transport, Journal of the Chartered
Institute of Transport, 38, 359–61.
Jung, J.M. and Fujii, E.T. (1976) The price elasticity of demand for air travel: some new
evidence, Journal of Transport Economics and Policy, 10, 257–62.
Kraft, G. and Domenich, T.A. (1970) Free Transit, Heath.
Lago A.M., Mayworm, P.D., and McEnroe, J.M. (1981) Transit service elasticities: evi-
dence from demonstration and demand models, Journal of Transport Economics and
Policy, 15, 99–119.
Lave, C.A. (1970) The demand for urban mass transportation, Review of Economics and
Statistics, 52, 320–23.
Lee, N. and Dalvi, M.Q. (1969) Variations in the value of travel time, The Manchester
School of Economic and Social Studies, 37, 213–36.

BUTTON 9781786435668 PRINT.indd 131 27/04/2022 11:07

 132 TRANSPORT ECONOMICS, 4TH EDITION

Levin, L., Lewis. M.S., and Wolak, F.A. (2017) High Frequency evidence on the demand
for gasoline, American Economic Journal: Economic Policy, 9, 314–47.
Li, Z. and Hensher, D.H. (2011) Prospect theoretical contributions in understanding trave-
ler behavior: a review and some comments, Transport Review, 31, 97–115.
Litman, T. (2021) Transport demands and elasticities: how prices and other factors affect
travel behavior, Report, Victoria Transport Policy Institute.
McGinnis, M.A. (1990) The relative importance of cost and service in freight transporta-
tion choice: before and after deregulation, Transportation Journal, 30, 12–19.
Metcalfe, R. and Dolan, P. (2012) Behavioural economics and its implications for trans-
port, Journal of Transport Geography, 24, 503–11.
Mogridge, M.J.H. (1978) The effect of the oil crisis on the growth in the ownership and use
of cars, Transportation, 7, 45–65.
Mutti, J. and Mural, Y. (1977) Airline travel on the North Atlantic, Journal of Transport
Economics and Policy, 11, 45–53.
Oum, T., Gillen, D.W., and Noble, S.E. (1986) Demand for fare class and pricing in airline
markets, Logistics and Transportation Review, 22, 195–222.
Oum, T.H., Waters, W.G., and Yong, J.-S. (1992) Concepts of price elasticities of transport
demand and recent empirical evidence, Journal of Transport Economics and Policy, 26,
139–54.
Oum, T.H., Waters, W.G., and Fu, X. (2008) Transport demand elasticities, in D.A.
Hensher and K.J. Button (eds), Handbook of Transport Modelling, 2nd ed, Elsevier.
Peer, E. (2011) The time-saving bias, speed choices and driving behavior, Transportation
Research Part F, 14, 543–54.
Quandt, R.E. and Baumol, W.J. (1966) The demand for abstract transport modes, theory
and measurement, Journal of Regional Science, 6, 13–26.
Quarmby, D.A. (1967) Choice of travel mode for the journey to work: some findings,
Journal of Transport Economics and Policy, 1, 273–314.
Redmond, S. and Mokhtarian, P.L. (2001) The positive utility of the commute: modeling
ideal commute time and relative desired commute amount, Transportation, 28, 179–205.
Reza, A.M. and Spiro, H.M. (1979) The demand for passenger car transport s­ ervices and
for gasoline, Journal of Transport Economics and Policy, 13, 304–19.
Schafer, A. (2000) Regularities in travel demand: an international perspective, Journal of
Transportation and Statistics, 3, 1–31.
Sherman, R. (1967) A private ownership bias in transmit choice, American Economic
Review, 77, 1211–17.
Smith, M.G. and McIntosh, P.T. (1974) Fares elasticity; interpretation and estimation,
in Symposium on Public Transport Fare Structure, Transport and Road Research
Laboratory Report, SR37UC.
Straszheim, M. (1978) Airline demand functions on the North Atlantic and their pricing
implications, Journal of Transport Economics and Policy, 12, 179–95.
Tanner, J.C. (1978) Long-term forecasting of vehicle ownership and road traffic, Journal of
the Royal Statistical Society, 141A, 14–63.
Taplin, J.H.E. (1980) A coherence approach to estimates of price elasticities in the vacation
travel market, Journal of Transport Economics and Policy, 14, 19–35.
Thaler, R.H. and Sunstein, C.R. (2008) Nudge: Improving Decisions on Health, Wealth, and
Happiness, Yale University Press.
Thomas, S. (1979) Non-working time-savings in developing countries, Journal of Transport
Economics and Policy, 13, 335–7.
UK Department for Transport (2007) Transport statistics Great Britain 2007, DfT.
UK Department of Transport (1977) Transport policy, Cmnd 6836, HMSO.
UK Department of Transport (1978) Report of the advisory committee on trunk road
assessment, HMSO.

BUTTON 9781786435668 PRINT.indd 132 27/04/2022 11:07

 THE DEMAND FOR TRANSPORT ­ 133

Wardman, M. (2001) A review of British evidence on time and service quality valuations,
Transportation Research Part A, 34, 171–205.
Wardman, M. (2014) Price elasticities of surface travel demand: a meta-analysis of UK
evidence, Journal of Transport Economics and Policy, 48, 367–84.
Waters, W.G. (1992) Values of travel time savings used in road project evaluation: a cross-
country/jurisdiction comparison, Australian Transport Research Forum.
Wei, F., Chen, J., and Zhang, L. (2017) Demand shocks, airline pricing, and high-speed
rail substitution: evidence from the Chinese market, Journal of Transport Economics
and Policy, 51, 266–89.
White, P.R. (1981) ‘Travelcard’ tickets in urban public transport, Journal of Transport
Economics and Policy, 15, 17–34.
Zahavi, Y. (1977) Equilibrium between travel, demand, system supply and urban structure,
in E.L. Visser (ed.), Transport Decisions in an Age of Uncertainty, Martinus Nijhoft.

BUTTON 9781786435668 PRINT.indd 133 27/04/2022 11:07

 5 Direct Costs of Transport

5.1 Factors Influencing the Supply of Transport

The supply of any good or service is generally a positive function of its price.
Suppliers, and stakeholders, are more willing to put the effort and resources into
providing more output as prices rise. The detailed nature of the relationship is
heavily influenced by the costs involved, although factors such as the motivation
of suppliers are relevant. Private companies, for example, generally have differ-
ent objectives from state-owned suppliers. This chapter, and the one that follows,
look at the various costs associated with supplying transport services and at the
relationships between the resources required to provide these services and the
types and amounts of output finally ‘consumed’ by travelers and freight consign-
ors. This chapter is specifically concerned with the production functions perceived
by the direct providers of transport services, which relate the various factor inputs
to the final services offered, and with the financial costs of these factor inputs.
There are two broad overlapping groups that have an interest in the supply
of transport services. First, there are the stock-holders who shoulder the direct
burden of financing the transport system and the services that are being supplied.
These may be thought of as entrepreneurs in the case of small-scale businesses
often found in sectors such as trucking and repair garage services or share-holders
in the case of larger undertakings such as airline or shipping companies. But they
may also be large corporations or government agencies in the case of large-scale
operations. In the case of the public sector, the tax-payers are the stock-holders.
The second group, the stake-holders, include not only the stock-holders but
also those who do not pay the direct costs of providing transport but nevertheless
incur costs from its provision. These latter parties may include users in terms of
fares and rates paid, but there are also those who have external costs imposed on
them. This latter group may include those whose health is damaged by air pollu-
tion associated with transport and those suffering third-party injuries caused by
accidents. The exact natures and orders of magnitude of these two broad groups
vary considerably on a case-by-case basis.
The chapter differs from the following one in that it deals with direct costs
as borne by the supplying agency: the stake-holders. These are normally, but not
always, financial costs that are incurred as the result of purchasing factor services
in the market (that is, the wages of labor, the interest on capital, the price of fuel,
etc.). There is one very important complication. The actual cost of the travelers’
or consignors’ own inputs is the direct cost of making a trip. Transport is special
(but not unique) in that the person being transported contributes their own time
inputs and, when private motor transport is involved, their personal energies,

BUTTON 9781786435668 PRINT.indd 134 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 135

skills, and expertise. Equally, with freight transport, the consignor of cargo bears
the direct costs of having inventory tied up in transit.
The opportunity costs of this time and the utilization of acquired skills will,
therefore, directly enter the production function for trip-making. We consider the
external costs in the following chapter, although representing genuine resource
costs does not directly influence the decisions of transport suppliers in their provi-
sion of transport services. In brief, therefore, this chapter is exclusively concerned
with the perceived or reaction costs that influence the supply of transport ser-
vices. It should be pointed out, however, that the separation of direct and external
costs is something of an expositional device and that in practice this distinction is
becoming increasingly blurred as official policies attempt to make transport agen-
cies fully cognizant of the full resource implications of their actions.
All these factors influence the shape of the transport supply function which,
in its simplest form, is seen in Figure 5.1. There is nothing special in the general
sense about the supply curve of transport services. It is normally positively
sloped, reflecting the incentive that higher prices have on suppliers to put more
of their services on the market. The importance of the need for consumers, at
least in terms of passenger movements, to put their own time into the delivery of
transport services, however, often means that one should treat price in a broader
sense to embrace time costs. This gives us a notion of ‘generalized costs’ that we
discuss in some detail later.
The supply curve’s position in price/quantity space is also influenced by
where it is positioned vis-à-vis factors such as levels of income, the price of alter-
natives to transport (which may be teleworking or teleshopping in the short term,
but can embrace relocating home or work in the longer term), and to the costs of
the inputs required to provide the transport services (including the costs of the
time of the travelers or of inventory holdings in the case of freight transport).

 Price

 Input costs
 Price of other goods
 Technology
Supply

0 Quantity

Figure 5.1 The simple transport supply curve

BUTTON 9781786435668 PRINT.indd 135 27/04/2022 11:07

 136 TRANSPORT ECONOMICS, 4TH EDITION

The financial costs of transport to immediate users have, as a broad generaliza-

tion, been falling over time. The money costs of traveling from, say, Washington
DC to San Francisco, for example, has fallen considerably over the past century
relative to incomes in the United States with the improvements in air and road
transportation, and it is certainly a lot quicker. This has been due to technical
progress, but also more recently to changes in the regulatory regimes that have
controlled the supply of transport – a subject we shall return to in detail in
Chapter 14.
Not all transport users have enjoyed lower fares, especially in the short run. As
technology and demand patterns have changed, so some modes or journeys have
become more expensive, often because a superior service has replaced them. In
other cases, the subsidies given to the various modes have changed, again affecting
the costs to the final user. Table 5.1 provides, in very broad terms, details of how
changes in passenger fares, measured in current prices, have varied since 1970 in
the United States. What it does not allow for is the general rise in price levels over
this period, which has been about 530 percent. Nor does it allow for the improved
quality of transport services, the speed, or the length of journeys. There are wide
variations by route or region and in part this chapter offers some reasons for these
micro- and meso-differences. Other modes exhibit different patterns.

5.2 Fixed and Variable Costs

Direct costs can be divided in several ways but two are particularly relevant to
transport, namely distinctions according to variability over time and distinctions
among the parties responsible for incurring the various elements of cost. The first
of these distinctions is discussed in this section.
In the long run, or so the introductory texts tell us, all costs are variable,
but the long run is itself an imprecise concept (even a tautology) and the ability
to vary costs over time differs among modes of transport. The long run in the
context of a seaport is, for instance, very different from that in road haulage or in
the bus industry. Port infrastructure is extremely long-lived, specific, indivisible,
and expensive. It is impossible to consider the standard question associated with

Table 5.1 Average public passenger transport fares in the United States (current prices)

1970 1980 1990 2000 2010 2018

Air carrier, domestic, scheduled 40.65 84.60 107.96 339.00 336.09 349.63
Class I bus, inter-city 3.81 10.57 20.22 29.46 n.a n.a
Transit, all modes 0.22 0.30 0.67 0.93 1.21 1.60
Commuter rail 0.84 1.41 2.90 3.33 4.81 6.43
Inter-city rail/Amtrak 3.19 17.72 39.59 52.15 60.70 69.59

Note: n.a = not available.

Source: US Bureau of Transport Statistics.

BUTTON 9781786435668 PRINT.indd 136 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 137

long-run costs – namely, ‘what is the cheapest way of providing a given capacity
in the long run’? – when talking of time horizons 20, 30, or even 50 years hence.
Trucking is different, capital costs are lower, physical durability less, and there is
always the prospect of varying the use, within limits, of the vehicle fleet. Trucks
are, unlike ports, mobile both among a range of potential employers and among
a range of locations.
The nature of many costs, therefore, means that they may be considered fixed
in the short term: there are temporal indivisibilities. The period under considera-
tion will, as we have already seen, differ among transport sectors, but it will also
differ within a single transport undertaking. Railway operators owning track and
rolling stock offer a useful illustration of this.
A railway service involves using several factor services, many of these being
highly specific and each with its own physical life-span. When it comes to con-
sidering line closures the essential questions revolve around deciding exactly
which costs are fixed. Figure 5.2 offers a general illustration of the main cost
items associated with a rail service together with an appropriate, although not
exact, indication of the physical life of existing equipment. In the very-short
term – the ‘market period’ since all other items have already been purchased
(that is, they are fixed) – the only reaction a railway can make when confronted
with a sudden change in demand is by adjusting extremely variable costs. These
are costs notably associated with such factors as labor, fuel, and maintenance.
However, if the railways are earning income to cover these costs then there is no
justification for closing the line. Wagons last about ten years, and applying the
same logic, these should be replaced after that time if returns exceed the costs
of doing so.
After a period of about 18 years, however, locomotives become due for
replacement. Consequently, locomotive costs become variable over a 15-year

Earthworks (infinite)

Track (40 years)

Locomotives (18 years)

Wagons (10 years)

Fuel costs

0 Time
Figure 5.2 Railway costs over time

BUTTON 9781786435668 PRINT.indd 137 27/04/2022 11:07

 138 TRANSPORT ECONOMICS, 4TH EDITION

horizon and the managers must decide whether revenues justify replacement.
Is 15 years, therefore, the long run? The answer is probably ‘no’ because other
factors are still longer-lived, such as rolling stock for 25 years and track and
signaling for 45 years. (Earthworks have an infinite life and once constructed do
not enter further decision-making processes.) Further decisions regarding closure
must, therefore, be made after 25 and 45 years even if the line earns enough
revenue to cover its motive replacement needs. Consequently, the long run for
railways, in this context, is anything up to 40 years with, in the interim, a series of
successive short-run cost calculations having to be made.
The reverse calculations are made regarding the expansion of services. If the
wagons are covering their costs to an extent that they would also cover the costs
of additional locomotives then capacity expansion is commercially justified.
It should be quite apparent that the distinction between fixed and variable
costs is a pragmatic device requiring a degree of judgment and common sense on
the part of the decision-maker. In the short term – whatever that may be – some
costs are clearly fixed, resulting in a falling short-run average cost of use until
capacity is reached. Just to show the generality of the principles, we can switch to
a different mode. Figure 5.3, for instance, may be seen as illustrating the average
cost of the increased use of a ship (SRAC1), which falls steeply until capacity is
fully utilized. A second ship may then, if demand is strong enough, be brought
into operation, exhibiting a short-run average cost curve of SRAC2.
The fixed capacity constraint for each ship typifies that found in most modes
of transport and tends to differ from the smooth, stereotypical, symmetrical
‘U-shaped’ SRAC curves often assumed to be associated with manufacturing
industry described in economics texts. While the upturn in the curve may not be
quite as dramatic as shown, and not entirely vertical if, for example in our ship-
ping case, a vessel steams more slowly or takes a shorter route, then there may be

SRAC1 SRAC2 SRAC3

LRMC =
SRMC1 SRMC2 SRMC3 LRAC

0 Tonnes
Figure 5.3 The long- and short-run costs of shipping

BUTTON 9781786435668 PRINT.indd 138 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 139

some curvature to the upward part of the curve. The long-run curve is, following
elementary economics, formed from the envelope of the short-run average cost
curves.
In many sectors of the economy the long-run average cost curve is not
horizontal, as in our illustration, at least not to any high level of minimum cost,
but is often found to be downward-sloping as a result of economies of scale.
These economies may potentially take a variety of forms in transport and may
be thought to vary according to the type of transport involved and the mode of
operation being undertaken.
The standard method of examining for scale effects is to look at the cost
elasticity of output. Evidence presented in a survey of studies of the rail, road
haulage, and aviation sectors in the United States (Winston, 1985) suggests that
scale effects exist in some modes with respect to output but not in others. All
studies examined, irrespective of the modeling framework used, indicate scale
effects in rail freight transport, but the empirical evidence is contradictory for
road haulage – although the more recent of the studies, employing state-of-the-
art econometric techniques, suggest constant costs. Studies of United States
domestic aviation indicate there are no, or few, scale effects in the conventional
sense, irrespective of how the cost function is specified (Johnston and Ozment,
2013). More specific studies tend to consider features of transport industries, fleet
size, infrastructure, etc., and these are looked at in turn.

Economies Associated with Larger Vehicle Size

One approach is to look at costs with respect to size of vehicle or vessel. The
classic example of this is in shipping, where capacity (in terms of volume)
increases much faster than surface area, but this is also a feature of most other
modes of freight transport. Thermal processes (such as engine size) also gener-
ally exhibit scale economies in all forms of transport, and crew numbers do not
increase proportionately with the size of mobile plant. Figure 5.4 gives some
rough indication of economies of scale in bulk carriers. The figure takes four size
classes – Handy (up to 40,000 dwt), Handymax (between 50,000 and 60,000 dwt),
Panamax (over 65,000 dwt), and Capesize (over 150,000 dwt) – and looks at the
cost savings per tonne as size increases. The differentials are given for rough guid-
ance. The exact economics of individual ships of any class varies according to
such factors as age, design, speed, and so on. Similar general patterns emerge not
only for other types of shipping – for example, Figure 5.5 gives data on container
vessels – but size economies also exist, at least up to a point, for most modes of
transport.
Diseconomies of vehicle size can be associated with such as the state of
technology (there were limits to the size of ships until metal vessels emerged) and
the quality of complementary services, such as transport infrastructure. Evidence
produced in the early 1970s by Edwards and Bayliss (1971), however, suggests that
there may be limits to the cost advantages of using extremely large road haulage
vehicles, at least in the United Kingdom. Diseconomies of scale appear to begin

BUTTON 9781786435668 PRINT.indd 139 27/04/2022 11:07

 140 TRANSPORT ECONOMICS, 4TH EDITION

18

Handy to
16
Handymax
$3.7/tonne
14 (22%)
$ cost per cargo tonne
Handymax to
Panamax
12
$2.7/tonne
(20%) Panamax to
10 Capesize
$3.6/tonne
(36%)
8

4
20 40 60 80 100 120 140 160 180
Ship size (thousand dwt)
Source: Stopford (2009).

Figure 5.4 Economies of scale related to ship size for bulk carriers

25

20
$/TEU/Day

15

10

0
0 5,000 10,000 15,000 20,000 25,000 30,000
Ship capacity in TEUs

Figure 5.5 Economies of scale related to ship size for container carriers

setting in after about 11 tonnes carrying capacity has been reached. Evidence
from Britain’s Armitage Inquiry into heavy trucks, however, indicates that there
are still economies of scale to be enjoyed by using commercial road vehicles in
excess of the 32.5 tonnes gross maximum weight permitted in 1980 (see Table 5.2).
Scale may also be approached in a different way, namely in terms of the
number of power units used to move a load. Ships, and many types of truck,
have a physically fixed engine to capacity ratio, whereas railroads have separate

BUTTON 9781786435668 PRINT.indd 140 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 141

Table 5.2 
Estimated savings from increasing the maximum weight of trucks in the United
Kingdom from 32.3 tonnes

Maximum weight Savings (%)

35 tonnes, 4 axles 5–7

38 tonnes, 5 axles 5–9
40 tonnes, 5 axles 7–14
42 tonnes, 5 axles 9–13
44 tonnes, 6 axles 11–13

Source: UK Department of Transport (1980).

locomotives pulling independent wagons. There are exceptions to this, includ-

ing the Australian road trains whereby a single powerful (‘prime mover’) truck
pulls a number of trailers, but they are limited (Zhang et al., 2020). Many of the
economies of platooning (or ‘flocking’) come from a reduction in the overall
weight of the locomotive unit needed to pull a large load and from the use of
a single driver. Platooning of wagons (or even automobiles) also reduces the
infrastructure required because of the strict physical regulation of wagons. This
allows them to travel closer together than would occur with individual cab/
wagon units. Uniform speeds have fuel saving potential. A possibility of auto-
mated, driverless trucks offers the prospect of greater economies of consolida-
tion as more wagons can be strung together and controlled remotely (Zhang
et al., 2020).
While for engineering reasons for various economies of scale clearly do exist
in utilizing larger units of mobile plant at least up to some point, it should be
noted that it is often impossible to take advantage of them even when demand
for transport services is high. In many instances there may be physical limitations
associated with complementary inputs, such as loading bays, and the associated
infrastructure cannot always handle the larger vehicles (ports and airports pose
specific problems in this context), while in others the consignments of traffic are
small and can be more efficiently handled in smaller vehicles. Returning to our
maritime example, large ships on liner routes often incur heavy costs as frequent
landings and embarkation of individual consignments require rearrangements of
the entire cargo. This is one reason why improved hub-and-spoke operations are
gaining in importance – see Section 5.3.

Economies of Scale in Infrastructure Provision

Transport demand is not spread evenly over space, but tends to be concentrated
on links between specific trip-generating and trip-attracting points, for example
journey-to-work trips are concentrated along links between residential and indus-
trial estates; holiday air transport demand is highest between the main cities in
the United Kingdom and popular recreational resorts in Spain, Greece, etc. Thus,
there is a tendency in transport, as we saw in Chapter 4, for demand to be con-
centrated on certain portions of the network and, ipso facto, certain parts of the

BUTTON 9781786435668 PRINT.indd 141 27/04/2022 11:07

 142 TRANSPORT ECONOMICS, 4TH EDITION

Table 5.3 Economies of large port operations

Vancouver Seattle

Traffic 88,000 377,000

Number of berths 3 11
Wharfage $23.63 $33.75
Handling & through-port charge $81.27 $40.00
Vessel service & facility charge 0 $26.60
Tailgate loading $22.28 0
Port charges per container $127.18 $95.35

Source: Heaver (1975).

static infrastructure. Technically this concentration of traffic is possible because

of substantial scale economies in infrastructure provision.
An examination of the costs associated with handling container traffic in an
average port (Vancouver) and a large port (Seattle) in Table 5.3 (which assumes
the charges paid by container traffic reflect port costs) reveals the economies
enjoyed from scale. It seems in port operations that a fourfold increase in size
reduces costs by approximately one-quarter. Similarly, with internal transport,
John Tanner (1968) found some years ago that the cost of four-lane motorways
was on average 78 percent of those with six lanes. However, it should be said that
evidence of scale economies in road provision is not altogether conclusive. The
findings of Keeler and Small (1977), looking at 57 freeway segments in the San
Francisco area, favor constant returns, while Alan Walters (1968), again employ-
ing United States sources, finds evidence ‘that there are increasing costs of (road)
construction in urban areas’.
The difficulty here, as with other forms of infrastructure costing, is to
isolate comparable construction costs from those costs associated with specific
locations and geographies. In terms of railways, the move from a single- to a
double-track system involves roughly a quadrupling of capacity by eliminating
conflict between directions. Quadruple tracking should more than double capac-
ity by permitting segregation by speed. Indications of such gains can be seen in
the British estimates in Table 5.4 of the costs of these options (at 1967 prices) for
similar geographies. As is often done, to spread costs of large investments over
time, interest is taken as the cost of capital. We see that the average rail costs rise
much less rapidly than increases in potential capacities. But there are wide varia-
tions in these costs dependent upon whether existing track is upgraded or entirely
new track is laid, the nature of the traffic, the gradients and curves involved, the
need for bridges and tunnels, etc.

Economies from Large Fleet Size

Larger fleets of vehicles may offer economies in maintenance, standardization (or,

in some cases, the availability of a mixed vehicle fleet to meet variable demand),
crew scheduling, etc., although administrative problems and remoteness of

BUTTON 9781786435668 PRINT.indd 142 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 143

Exhibit   The container and world trade

A major driver of the sevenfold increase in world trade from the early 1960s to the 1990s
was the introduction of containerization. This was developed in the mid-1950s in the United
States by Malcom McLean. Containers are robust, with steel-reinforced corners allowing
stacking without causing damage. They are uniform in size, theft-proof, easy to load, and can
be moved by ship, barge, truck, train, or plane (although air containers are smaller).
Prior to this, loading and unloading general cargo involved the handling of individual items
of different dimensions and weights. There was some unitization using pallets, sacks, and so
on, but this did little to reduce costs and ships spent two-thirds of their productive time
in ports. The costs of bulk breaking meant that most maritime movements involved simple
connectivity and limited hub-and-spoke movements. Containerization also saw pilfering pre-
container reduced considerably, which brought down insurance costs.
Containerization is capital-intensive, requiring specialized infrastructure not only at ports
and at inland origins and destinations, but also in trucks and railroad wagons, as well as
investment in the boxes themselves. But productivity gains were huge – the first purpose-
built crane in 1959 could load a 40,000 lbs. box every three minutes – more than 40 times
the productivity of a longshore gang. These and other effects are seen in the table which
assesses the effects of containerization in the United Kingdom and Continental Europe over
the first five years of containerization.
Pre-container: 1965 Container: 1970–71

Productivity of dock labor 1.7 tons per hour 30 tons per hour
Average ship size 8.4 gross registered tonnage 19.7 gross registered tonnage
Port concentration (number of 11 ports 3 ports
 European ports, southbound
Australia)
Insurance costs (Australia–Europe £0.24 per ton £0.04 per ton
trade for imports)
Capital locked up as inventory in transit £2.00 per ton £1.00 per ton
(Hamburg–Sydney)

Bernhofen et al. (2016) looked at a large panel of product-level trade flows between 1962
and 1990 to examine the role of containerization on twentieth-century globalization. They
assumed 1966 to 1983 as the period in which significant containerization occurred, with
the four-year period before being ‘pre-container’ and the seven years after being ‘post-
adoption’. It was assumed other technologies were pretty stable over the 18 years.
Focusing on 22 industrial countries in which the uptake of containerization was seen as
quicker and deeper, they compared the causal effects of containerization on international
trade using non-containerized bi-lateral pairs as a counterfactual. Adjustments were made
to allow for some countries trading more in products not amenable to containerization.
Developments in institution trade arrangements were captured by variables reflecting
membership of a free trade bloc and by both partners in trade being General Agreement on
Tariffs and Trade (GATT) members.

BUTTON 9781786435668 PRINT.indd 143 27/04/2022 11:07

 144 TRANSPORT ECONOMICS, 4TH EDITION

Greater effects were found in trade between developed economies than either between
developed and developing economies or between two developing economies. More traded
goods were suitable for containerization. Regarding inter-developed economies’ trade, the
cumulative average effects of containerization over a 20-year time period were also much
larger than the effects of free trade agreements or the GATT.
See also: D.M. Bernhofen, Z. El-Sahli, and R. Kneller (2016) Estimating the effects of the
container revolution on world trade, Journal of International Economics, 98, 36–50.

Table 5.4 Average cost per mile of rail track in the United Kingdom (1967, current prices)

Number of tracks
1 2 3

Interest £3,020 £4,260 £7,900

Revenue of track & structure £1,400–2,840 £1,986–3,442 £3,422–4,474
Signaling £2,600 £4,030 £8,070
Total £7,020–8,460 £10,190–11,730 £19,380–20,430

Source: Foster and Joy (1967).

decision-maker from customer may temper these advantages. Evidence on the

presence of fleet economies of scale, however, is far from conclusive for all modes
of transport and indeed there seems to be a gradual emergence of constant
returns for many with limited scale economies in specialized forms of transport.
Historically, the analysis regarding bus operations has been rather confusing.
The oft-cited United Kingdom statistical analysis by Lee and Steedman (1970),
looking at the accounts of 44 United Kingdom municipal bus undertakings for
1966–67, suggests the notion of constant returns to scale, a situation agreeing
with Williams’ (1981) study of 11 American publicly owned operations. Wabe
and Coles’ (1975) examination of 66 United Kingdom operators, however, ‘pro-
vides evidence that diseconomies of scale exist in the provision of bus services’.
Koshal’s (1970) work in India offers international support for the notion of con-
stant returns in bus operations. The problem with most of these studies, however,
is that they use econometric models which assume that the existence of scale
effects does not extend over all levels of output.
The 1980s witnessed the development of more ‘flexible’ model forms in
econometrics (a topic we address in more detail later in this chapter) which allow
for variation in scale effects as output changes – for example, they allow for testing
for a U-shaped cost curve. Button and O’Donnell (1985), applying such tech-
niques to United States urban bus data, found that economies of scale do seem to
exist up to a point, although diseconomies set in for the larger undertakings. This
broadly conforms with studies both of Israeli bus fleets (Berechman, 1983) and
for the United States inter-city bus industry.
Alan Walters’ work on the trucking sector produced evidence of constant
returns in the United Kingdom, although the large number of owner-drivers

BUTTON 9781786435668 PRINT.indd 144 27/04/2022 11:07

 145
DIRECT COSTS OF TRANSPORT ­

working in this sector makes exact costing difficult. Regarding United Kingdom
aviation, the Edwards Committee found evidence of scale economies when there
was fleet standardization but generally concluded that the optimal size of a
fleet depends upon the task in hand. Looking at shipping, Tolifari et al. (1986)
found evidence of scale effects in bulk fleets – see Figure 5.6. The difficulty with
empirical work in this field, as well as in aviation, is the diversity of the market
conditions that are encountered and the support that is often forthcoming from
government to finance ‘the nation’s flag carrier’. As can be seen in the figure, the
costs of a fleet with traditional registry (in one of the major industrial states) tend
to be higher than when open registry (the adoption of a ‘flag of convenience’)
is favored, because of the standards demanded by the former. As we see below,
these scale effects also become rather clouded with the introduction of economies
of density.
One indication of constant returns to fleet size is sometimes thought to
be diversity in the scale of operators in a sector of transport: large companies
competing directly against one or two vehicle firms suggesting that there is little
advantage in large fleets. Road haulage is possibly the most extreme example of
this phenomenon, as we see regarding the United Kingdom in Table 5.5, but
throughout transport – except for areas directly regulated by government – one
finds large and small firms competing.
Differential managerial skills may permit some firms to grow larger than
others, but this does not imply technical economies of scale exist. They may also
be supplying slightly different services, perhaps in qualitative terms, which are not

12
Open registry
Average cost per thousand ton-miles

10 Traditional registry

0
1 2 3 4 5 6 7 8 9 10 11 12
Hundred million ton-miles
Source: Tolifari et al. (1986).

Figure 5.6 Economies of scale in bulk shipping

BUTTON 9781786435668 PRINT.indd 145 27/04/2022 11:07

 146 TRANSPORT ECONOMICS, 4TH EDITION

Table 5.5 
United Kingdom O-license holders and number of vehicles grouped by size of
fleet (2015)

Fleet size Number of licenses Number of vehicles

0 8,023 0
1 31,576 31,576
2 to 5 24,112 69,305
6 to 10 5,616 42,512
11 to 20 3,141 45,538
21 to 50 1,909 59,642
51 plus 858 99,487
Total 75,235 348,060

easy to quantify, and thus strict cost comparisons become difficult. Evidence of
this type, therefore, may be capturing differential demands rather than any strict
scale effect.
The variable costs of transport – those related to the rate of output – are gen-
erally considered to be dominated by labor and fuel items. They are usually seen as
variable in the long term because the inputs involved are largely tied to the outputs
delivered, and thus suppliers of transport can cut their costs if demand for their
services declines. But often the costs of these inputs can also change dramatically
in the short term, independently of actions by transport suppliers. Oil crises are
one example of such instability but civil unrest and natural events like floods also
pose challenges. In this case the transport supplier is suddenly confronted with a
cost increase necessitating a rise in fares or freight rates being charged.
The importance of variable relative to fixed costs can vary considerably, even
in the short term as seen by the increased importance of fuel costs for modes such
as aviation in 2007 to early 2008, when oil prices rapidly rose to record levels.
Since infrastructure costs are relatively fixed, it is, therefore, the cost of the mobile
plant that is normally treated as marginal. Table 5.6 offers data on international
airlines in a more stable period before the atypical fuel price fluctuations that
began in the 1970s.
It should be noted that even in the long term the costs of mobile plant
are, in aggregate, likely to exceed those of infrastructure, although individually
vehicle costs in many cases are relatively small. The costs of aircraft operations
come within a sector with very high costs per unit of mobile plant, while road
haulage is a sector in which the total operational costs per vehicle are low. The
importance of both direct and indirect labor costs and fuel costs, however, is
apparent at both ends of the spectrum.
Remaining with airlines, while Table 5.6 offers average figures, the costs
incurred vary considerably according to such things as type of aircraft, and their
configurations, lengths of routes, traffic flow, flights per day, and so on.
Table 5.7 represents a broad average across many different types of airline
operations. But variable costs differ not simply with the level of vehicle usage
but also with the type of transport operation undertaken. Vehicles, for example,

BUTTON 9781786435668 PRINT.indd 146 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 147

Table 5.6 Components of operating costs of scheduled international airlines (percent)

1980 1990 1999

Direct operating costs

Flight operations 44 32 35
  Flight crew 8 7 8
   Fuel & oil 28 15 8
   Airport & en route charges 5 4 7
   Aircraft rental, insurance, etc. 3 6 9
Maintenance 11 11 11
Depreciation: aircraft & ground facilities 6 7 7
Total direct operating costs 60 50 53
Indirect operating costs
Station/ground services 11 12 11
Passenger services 9 10 11
Ticketing, sales, & promotions 14 16 14
Administration & other costs 6 11 12
Total indirect operating costs 40 50 47

Source: International Civil Aviation Organization, Digest of Statistics, ICAO, from years
indicated.

Table 5.7 United States airlines’ Boeing 757–200 flight operating costs (FOCs) in 2009

Airline Number Seats FOCs per FOCs per Utilization Stage length
of aircraft block hour seat hour (hours/day) (miles)

American 101 188 $2,568 $13.66 10.3 1,460

Continental 34 179 $2,568 $14.35 12.1 1,860
Delta 101 188 $2,357 $12.95 1.6 984
America West 12 190 $2,065 $10.87 13.1 1,167
Northwest 48 191 $2,260 $11.83 11.7 1,137
Trans World Air 20 179 $2,656 $14.84 11.8 1,405
United 98 186 $2,684 $14.43 11.2 1,281
USAir 34 182 $3,069 $16.87 11.1 1,254

Source: https://www.icao.int/mid/documents/2017/aviation%20data%20and%20analysis%20
seminar/ppt3%20-%20airlines%20operating%20costs%20and%20productivity.pdf.

have an optimal speed above or below which fuel costs tend to rise steeply; con-
sequently, operations involving continually stopping and starting will, ceteris
paribus, increase variable costs. Maintenance costs can also vary considerably
with the type of terrain over which journeys are made. To some extent these varia-
tions may be off-set by employing specialized vehicles whose variable cost profiles
conform most closely to the type of operation undertaken. In the airline context
there exists a whole range of different aircraft designed to meet the needs of dif-
ferent operational patterns – airbuses for short-haul and large-volume traffic,
wide-bodied jumbos for long-range operations, and so on.
Equally, Table 5.8, from one of the most detailed studies on the subject, gives
some indication of the running and standing costs of different forms of operation

BUTTON 9781786435668 PRINT.indd 147 27/04/2022 11:07

 148 TRANSPORT ECONOMICS, 4TH EDITION

Table 5.8 
United Kingdom trucking, running and standing costs per mile (1977, current
prices)

Fleet Bulk Tipping Smalls & Long Other Other All

size tankers parcels distance general sectors sectors

Running 100+ 0.20 – 0.12 0.15 0.21 0.14 0.18

Standing 100+ 0.27 – 0.30 0.18 0.31 0.42 0.28
Running 21–100 0.17 0.13 0.12 0.15 0.16 0.16 0.15
Standing 21–100 0.22 0.19 0.28 0.16 0.22 0.26 0.21
Running 1–20 0.19 0.15 0.12 0.13 0.12 0.16 0.14
Standing 1–20 0.31 0.12 0.19 0.19 0.14 0.25 0.18

Source: UK Price Commission (1978).

in the United Kingdom’s trucking sector in the late 1970s. While these figures
are not strictly variable and fixed costs, and they are very much case-specific in
that trucking hauls are relatively short in Britain compared with, say, the United
States, they do convey a general impression of how costs vary, even within sectors,
with the type of transport operation performed.
The cost profiles also vary with the type of firm controlling the operations.
Evidence from statistical studies of United States urban public transit systems,
for instance, suggests that public ownership can lead to higher costs of provi-
sion. Karlaftis and McCarthy (1999), for example, found that Indianapolis
experienced an annual 2.5 percent reduction in operating costs after privatizing
the management of its public transit system. Ian Wallis (1980), in his study
of urban bus operations in major Australian cities, gave a list of reasons why
private operators enjoy lower costs in certain areas than their publicly owned
counterparts:

• greater flexibility and efficiency in use of labor;

• relatively small proportions of maintenance and administrative costs;
• lower basic rates of pay; and
• lower wage/salary on-costs (taxes, pensions, etc.).

While in part cost variations may be explained in terms of either the size of
the operator or the type of operation undertaken, cost differences may also reflect
alternative operational objectives. There is ample evidence that large national
airlines often employ high-cost modern equipment to enhance their image. But
even at the level of local public transport, similar indications of X-inefficiency
exist. Teal et al. (1980), for example, cite instances of local authorities preferring
to operate their own para-transport system rather than make use of established
private operators, despite demonstrably higher costs.
Labor costs, although flexible, are usually much less variable than fuel
costs. This is not simply because of imperfections in the labor market (for
example, fixed working hours, union agreements on redundancies, training costs,
etc.), which often make it difficult to increase or reduce the size of the labor

BUTTON 9781786435668 PRINT.indd 148 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 149

force – even in the sense of dividing up public transport crews to conform with
daily peaks in travel demand – but also because of the nature of many types of
transport operation. Once a form of transport operation has been decided upon,
and capital invested, there are high labor costs associated with maintaining and
servicing this equipment irrespective of the traffic carried. Further, once an
undertaking is committed to a scheduled service, labor becomes a fixed cost in
providing this service.
One of the major problems in this latter context is the technologically unpro-
gressive nature of many forms of transport operations, which makes it difficult to
substitute one factor input for another as their relative prices change. In the case
of the mercantile marine, it has proved possible to substitute fixed for variable
factors (notably capital for labor) as labor costs have risen, but this is much less
easy in areas such as public transport provision. It is difficult to see how the basic
operations of traditional taxi-cabs, for example, could be retained with a substan-
tial reduction in labor input. Attempts to reduce labor costs in the urban public
transport sphere in the United Kingdom, by introducing one-person-operated
vehicles, did seem to have some limited effect on costs (for example, Brown and
Nash 1972 estimated a 13.7 percent cost saving), but this should be a once-and-
for-all step rather than the prospect of continual factor substitution.

5.3 Economies of Scale, Scope, Density, Experience, and

Commonality

In recent years it has become increasingly appreciated that while economies of

scale in their strictest form may be of considerable importance in many transport
activities, there are instances where it is not simply the sheer size of the under-
taking or activity that is the prime determinant of cost variation. Several other
factors are of importance in the transport context.
Many transport undertakings provide a variety of outputs. Taking aviation
as an example, at the most basic level these may be scheduled and charter services,
but, in more detail, there is the matter of outward and return services, and, at
a more detailed level still, each flight may be viewed as a specific product. The
economic question then becomes one of deciding whether there are cost savings
in one supplier producing this range of services rather than there being several
suppliers each specializing. Where there are cost economies from multi-product
production then economies of scope are said to exist. Just as with conventional
scale effects, it is possible for economies of scope to exist at some levels of output
but not at others. As a city expands, for example, there may exist economies of
scope in bus service provision favoring a monopoly supplier, but as it gets larger
and demand rises so these economies may disappear and several much smaller
operators prove more cost-effective.
Technically, economies of scope are assessed when:

S 5 {[C(Q1) 1 C(Q2)] – C(Q1 1 Q2)}/{C(Q1 1 Q2)} (5.1)

BUTTON 9781786435668 PRINT.indd 149 27/04/2022 11:07

 150 TRANSPORT ECONOMICS, 4TH EDITION

where:

C(Q1) is the cost of producing Q1 units of output one alone;

C(Q2) is the cost of producing Q2 units of output two alone; and
C(Q1 1 Q2) is the cost of producing Q1 plus Q2 units of together.

Economies of scope exist if S > 0. There are economies of scale if C/Q falls as Q
expands.
There are also instances where there are cost economies from serving larger
markets – it effectively allows the more intensive use of capital. Again, aviation
provides an example of these so-called economies of density where larger markets
enable higher load factors to be enjoyed and hence a lower unit cost per passenger.
Economies of scale, scope, and density can be obtained in various ways using
a variety of network types. While we discuss each of these below, separately, in
practice most transport undertakings combine them in a variety of ways.

Point-to-point Services

Point-to-point services simply offer services between two points – that is, connec-
tivity as opposed to inter-connectivity. This is seen in Figure 5.7, where there are
five cities that are all linked to each other by, for example, rail services offered by
separate suppliers. This involves services over ten separate lines. There are no real
economies of scope because each train carries passengers who are only concerned
with getting to the destination of an individual service. What economies of scope
do exist are on the demand side and come from some travelers wanting different
qualities of service, for example, those who are willing to pay a premium for first-
class service.

Linear Networks

Linear transportation services such as buses and railways that collect and
drop passengers en route seek to capture economies of density and scale. For

Point-to-point network Hub-and-spoke network

10 different routes 4 different routes

Figure 5.7 Point-to-point versus hub-and-spoke networks

BUTTON 9781786435668 PRINT.indd 150 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 151

example, in the United States, Amtrak’s north-east rail service from Washington
DC to Boston makes stops on the way at other major cities such as Baltimore,
Philadelphia, and New York, to drop and collect passengers. One reason for the
fares scale Amtrak uses, and the schedules adopted, is to maximize revenues by
maintaining a high-load factor by moving various groups of passengers between
a variety of destinations. The challenge is that overtaking on railways is difficult
without expensive provision of sidings. Buses, on the other hand, have more flex-
ibility when operating linear route structures, as, for example, do airlines to some
extent in Norway and Chile, where linear networks are common because of the
geography of those countries.

Hub-and-spoke Networks

The coming together of these economies of scope and of density has also been
characterized by the adoption of ‘hub-and-spoke’ operations. While United
States domestic aviation is the most cited example of this phenomenon, with all
the main airlines basing their services on flights to and from a limited number of
hubs it is also found in shipping (with traffic into Europe coming by large ships to
a small number of large ports to be distributed to other ports by smaller vessels)
and in some spheres of bus operators (such as the importance of Victoria Coach
Station in London as the hub for long-distance inter-city bus services in the
United Kingdom).
Figure 5.7, as we have seen, offers an indication of the cost advantages of
hub-and-spoke operations. With direct services between each city pair, as in the
point-to-point operation shown in the left-hand side of the figure, there would
need to be ten services provided, and many of those may be relatively thin with
limited traffic. If, as in the right-hand side of the figure, traffic is hubbed through
one of the cities, with passengers being consolidated at this hub to go onto their
final destinations, then only four services are required. This allows larger aircraft
to be used, and potentially higher-load factors enjoyed, thus reducing costs and
fares. The penalty is that people traveling through, rather than stopping at, the
hub airport will have longer flight times and spend some time waiting for their
connecting flights.
For longer-distance travel, often international routes, there may be several
changes. Figure 5.8 illustrates a typical ‘dog-bone’ or ‘dumb-bell’ route structure.

x a
y b
z
A B
c
j
i
Figure 5.8 The ‘dog-bone’ pattern of international hub-and-spoke operations

BUTTON 9781786435668 PRINT.indd 151 27/04/2022 11:07

 152 TRANSPORT ECONOMICS, 4TH EDITION

Here a passenger wanting to travel from x to c would be routed as x ⇒ A ⇒ B ⇒

c, where A and B are major hub airports that consolidate traffic from cities x, y,
z, … , j traveling to a, b, c, … , i. An example of these types of trips are found on
North Atlantic air routes where, for example, an individual may fly with United
Airlines from a small city in the United States to Washington DC, change plane
and fly on to London, change planes and fly to a small European city.
Competition in this system is largely between inter-connected airline net-
works, often involving alliances. Figure 5.9 again takes the long-distance move-
ment x ⇒ c, but now looks at the options that may be available when comparing
different air cargo alliances rather than just the services of a single airline. The
routing x ⇒ A ⇒ B ⇒ c is retained as the base case, but the airline offering this
may also have alliance ties with other carriers and push traffic through one or
more of their hubs; this may make x ⇒ C ⇒ A ⇒ B ⇒ c an option, or x ⇒ C ⇒
D ⇒ c. A competitive alliance may offer x ⇒ F ⇒ G ⇒ c. And so on.
Depending on schedules and routes served, there are thus numerous pos-
sibilities for moving cargo from x to c. From a costing perspective, the costs of
these various combinations will differ for the combinations supplying airlines
involved. In the context of a discussion of fuel price rises, each overall routing
would require a different surcharge that would need to not only reflect the weight
of the consignment but also such things as the different distances involved and the
forms of equipment used by the various members of an alliance.

Radial Networks

In practice, we find that in many cases the large inter-connected network transport
undertakings often compete not only with each other but with alternative forms
of networks. In some cases this involves competition with networks that have
little or no connectivity – they are, rather, a set of non-connecting services. Tramp
shipping may offer this type of competition to the maritime liner networks. But
more structured competition may be found in the airline sector whereby low-
cost carriers operating ‘radial’ networks compete with the hub-and-spoke model
of the traditional airlines. Looking again at Figure 5.7, radial networks would
involve scheduled services between a ‘base’ and the four connected airports but
no inter-lining of services at the base.

C D

x c
A B

F G

Figure 5.9 Network competition in the air cargo sector

BUTTON 9781786435668 PRINT.indd 152 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 153

When radial networks are adopted, they are generally accompanied by consid-
erable commonality (or standardization). Remaining with the low-cost airline
example, Ryanair and Southwest use a common fleet of derivatives of the Boeing
737 aircraft, while EasyJet and Frontier Airlines use planes from the Airbus 320
family. This concentration on a limited number of aircraft type produces a variety
of advantages:

• The smaller the number of aircraft types, the lower the number of reserve
crews needed (to cover for illness, etc.). Flexible swaps of crews are possible
with one type, with less crew training being required.
• Fewer spare parts are need with common fleets, lower costs for storage,
capital intensity of stock, and costs of obsolescence. Standardized mainte-
nance processes lead to reduced labor costs, with mechanics only needing to
be familiar with only one type of aircraft.
• Standardization means ground handling can be simplified and more cost-
efficient. Carriers benefit from economies of scale in commonality of ground
handling equipment and from less training of labor.
• Generally, ordering several planes of a single type will lower their unit price
compared to ordering a mixture of planes from different manufacturers.

While a standardized fleet can produce significant cost advantages, there can
be drawbacks. Carriers aiming at maximizing efficiency and their cost advantages
by using a uniform fleet generally limit the types of services they provide. As
noted earlier when discussing hub-and-spoke networks, airlines require types of
aircraft that best suit their networks, and when offering a wide range of diverse
services may be best served with a more diverse fleet. In the low-cost carrier case,
there are limited variations in demand and routes and commonality may thus
be advantageous. This contrasts with airlines such as Lufthansa, Air France,
or Delta, which operate global hub-and-spoke systems involving high-demand
trunk routes as well as lower-demand regional connections. These airlines require
a more varied fleet. The more alike are routes and demand for the services, the
more important will fleet commonality be in attaining maximum efficiency.

Economies of Experience

A further recent development is the appreciation that there is a ‘learning-by-

doing’ element in many activities that can reduce costs, and in some cases also
be a factor affecting demand. These ‘economies of experience’ tend to be under-
researched in the transport sphere but technically exist when unit costs decline as
more is produced or as a supplier stays in a market for an extended period. There
may be several dimensions involved in this:

• Goodwill. When confronted with several carriers, potential users of avia-

tion have varying levels of information, and especially about the quality of
services. Risk aversion encourages a ‘better the devil I know’ mentality that

BUTTON 9781786435668 PRINT.indd 153 27/04/2022 11:07

 154 TRANSPORT ECONOMICS, 4TH EDITION

favors incumbent suppliers. The need to combat this with advertising and
promotions pushes up the costs of new entrants.
• Knowledge. Incumbent suppliers have more information about a market they
serve and can tailor their services to specific customer groups. New entrants
must sink resources into acquiring such information.
• Organization. New entrants must assimilate the requirements of new services
within their networks and this entails learning costs throughout the remain-
der of their organizations.

As for evidence, one early study of United States domestic airline regulation
found that after the opening of routes, incumbent carriers had a larger impact
on markets than newly established carriers (Baker and Pratt, 1989). Similarly,
analysis of Compass Airlines’ entry into the Australian domestic market high-
lights the problems of establishing goodwill and brand loyalty (Nythi et al., 1993).
Switching mode, in the United Kingdom it has been argued that these experience
effects may have been advantageous to National Express, the largest bus opera-
tor, in the period following the deregulation of the inter-city bus network after
1980. National Express was the incumbent supplier and enjoyed, along with
other things, considerable experience in the market that enabled it to dominate
newcomers.
The natural way for a transport undertaking to enjoy the various cost advan-
tages of scale, scope, and so on is for it to develop its own network and services
on those networks efficiently. In many cases, however, suppliers try to accelerate
this process by merging or taking over an existing network to add to their own.
Mergers may take place for several reasons. In some cases – and we shall return
to this in Chapter 7 – it may be to remove competition from a market to enjoy
additional market power. Mergers involving overlapping transport networks are
often seen as motivated in this way. In other cases, there may be synergies between
non-overlapping (or at least involving limited overlap) that can add to economies
of scope and density, and bring in economies of experience because they do not
involve developing routes from scratch. Added to this, the enlarged network may
generate economies of greater market presence on the demand side by offering a
wider range of services to transport users.
The decision to merge networks occurs, therefore, when the incentives exist
for the businesses involved, be they driven by rent-seeking or cost-saving motiva-
tions. For simplicity we assume that the benefits of, say, a shipping network are
proportional to the number of customers (n) who use it, and that the constant of
proportionality is unity. Then, following Metcalfe’s Law, the value of the network is
n2. A simple calculation shows that combining two networks of size n1 and n2 yields:

∆v2 = n2 ( n1 + n2 ) − n12 = n1n2 (5.2a)

∆v2 = n2 ( n1 + n2 ) − n22 = n1n2 (5.2b)

BUTTON 9781786435668 PRINT.indd 154 27/04/2022 11:07

 155
DIRECT COSTS OF TRANSPORT ­

Each network of sailings gets equal value (v2) from the inter-connection. Those
on a smaller network (2) each get considerable value from linking with the large
number of nodes on the larger network (1), while many consignors of the latter
may each get a smaller additional utility, but there are a lot of them. This offers
scope for reciprocation, with the networks having free access to each other. The
problem is that the large supplier may need to keep its market power to allow
adequate price discrimination to recover costs and make an acceptable long-term
return. In this case, the larger shipping company may merge with the smaller one
and attain twice the value of offering inter-connecting services:

∆v1 = n1 ( n1 + n2 ) − n12 − n22 = 2 n1n2

2
(5.3)

5.4 Specific, Joint, and Common Costs

When dividing out costs, according to the groups of services produced, it is

often useful to allocate responsibility to specific users or consignors. While the
fixed/variable cost dichotomy poses problems about the relevant time period to
consider, cost responsibility raises issues of the traceability of costs. Some costs
are very specific and can, therefore, be allocated quite easily – the stevedore costs
of loading and unloading a cargo onto and off a ship is a case in point. In other
cases, a degree of averaging may be necessary but, nevertheless, costs can gener-
ally be traced to specific groups or classes of user.
But there is also a whole range of other costs that may be either ‘joint’ or
‘common’ to a few users and are difficult to trace directly to any specific group.
It is sometimes said that fixed costs may generally be treated as joint or common
while variable costs may be treated as specific, but this is too simplistic. Many
variable costs are, in practice, joint (for example, the fuel costs incurred in moving
a train in one direction and bringing it back are joint to both movements) or
common (for example, the basic maintenance costs of retaining a freight and
passenger rail link), while certain fixed costs are clearly specific (for example, the
capital costs of freight wagons have no connection with passenger demand).
In some cases, the technology is such that two products can be delivered at
the same time: the textbook wool and mutton case. An example of this in trans-
port is the combination of passengers being carried on the upper deck of com-
mercial aircraft and freight on the lower deck. Allocating the fixed costs of the
aircraft between these two uses, other than for specific features like seats, etc., is
challenging. Nevertheless, airlines do carry both on many aircraft, as can be seen
in the revenue composition in Figure 5.10.
Strictly, joint costs exist when the provision of a specific service necessarily
entails the output of some other service. Jointness is a technical feature and exists
at all points in time, that is, both before as well as after any investment decisions
are made. Return trips (or ‘back-hauls’), where the supply of transport services
in one direction automatically implies the provision of a return service, are the
classic examples in transport economics. The fact that true joint products are

BUTTON 9781786435668 PRINT.indd 155 27/04/2022 11:07

 156 TRANSPORT ECONOMICS, 4TH EDITION

60
50
40
30
20
10
0 A es

Ja na nes

irl s
or A ys

Si at As es
ap y P na

C Air fic

A o
fth s
Fr a
A ce
iti rth a

A nite irw t
er A ays
D A nes

es
U sh A wes
LM Lu line

a ine
ir ns

Br No itali
n arg
n in
in

in
K na wa

an
i
ng a ia
or ac
hi li

an li
-A a
ea irl
irl

elt irl
pa C
ir

ic ir
hi ir

A
C A

m d
A

h
EV

K
Figure 5.10 Percentage of airlines’ revenues from cargo

produced in fixed proportions means there can be no variability in costs, making

it logically impossible to specify the cost of, say, an outward journey when only
the overall cost of the round trip is known. Joint costs are, consequently, non-
traceable. This further implies that joint costs can only be escaped jointly, with
services in both directions being withdrawn together.
In a market situation, joint costs pose few problems in practice (Mohring,
1976). If there is a competitive road haulage service offering a round trip between
A and B and back again each week using M trucks then equilibrium rates would
soon emerge for each service (that is, from A to B and from B to A). Although
there are specific delivery, terminal, pick-up costs, etc., little difference exists
between the costs of running the lorries fully loaded or empty and hence prices
would be primarily influenced by the differences in the demand in each direction.
In the short term, the combined revenues from the A to B and the B to A ser-
vices may not be enough to cover joint costs, but in such a situation the number
of trucks offered would soon fall below M, increasing the price of trips in both
directions until joint costs are recovered. Excess revenue above joint costs would
have the opposite effect. The key point is that differences exist in the demands
for the out and return services and that different prices should be charged for
each in equilibrium. Consequently, knowledge of the relevant demand elasticities
together with that of joint cost permits the problems of traceability to be avoided.
We return to consider this problem in more detail in our discussion of pricing in
Chapter 6.
Common costs are like joint costs, in that they are incurred as the result of
providing services to a range of users, but differ in that the use of resources to
provide one service does not unavoidably result in the production of a different
one. The classic example of common costs in transport is the provision of track
facilities. A road may be used in common by lorries and cars, but the withdrawal
of rights for hauliers still leaves costs to be borne by motorists. With several
classes of user, it is often possible to trace certain components of cost to those

BUTTON 9781786435668 PRINT.indd 156 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 157

responsible, but there is still usually a large proportion that is untraceable. We

now turn to look at some attempts that have been made to allocate common track
costs across different categories of traffic.

5.5 Problems of Common Cost Allocation:The Road and Rail

Track Cases

The allocation of common track costs among users poses practical problems in
transport and deserves specific attention. As Arthur Hadley put it in 1885, before
becoming President of Yale University, ‘God Almighty did not know the cost of
carrying a hundred pounds of freight from Boston to New York’. The ways costs
are allocated varies between countries in part because some facilities are publicly
owned and some private. The inter-city road network in the United Kingdom,
for example, is almost exclusively the responsibility of central government. Users,
since the effective abolition of the Road Fund in 1937 and its legal death in 1955,
make no direct, hypothecated payments to use the network – save for a small
number of toll bridges and roads – but do pay considerable sums to government
each year in the form of fuel tax, value added tax, car tax, and vehicle excise duty.
The United States has had a national federal fuel tax since 1956, which in 2021
amounted to 18.4 cents per US gallon of gasoline and 24.4 cents per gallon of
diesel fuel. The revenue goes into a Highway Trust Fund. Revenues go into the
Highway Account which funds road construction and other surface transporta-
tion projects, and a Mass Transit Account to support mass transit. In addition,
individual states have a variety of mechanisms for financing their road networks.
When deciding upon the desirability of making a road journey, potential
users are to some extent influenced by these taxes. Therefore, attempts have been
made, on the grounds of economic efficiency, to allocate accurately the public
costs of road provision (both the construction and maintenance of the track) to
users. The European Union, for example, wishes members to ensure that all road
users pay at least their allocated short-run marginal costs of track provision and
that the full long-term cost is recovered in total.
There are problems of deciding exactly what constitutes the total cost of
road track provision in any period – for example, should the maintenance costs
be estimated in the same way as depreciation in nationalized industries or a com-
mercial business, or simply considered as they are incurred? And what exactly
constitutes the capital cost of any one year?
National comparisons of annual expenditures against tax revenue suggest
that most governments of industrial countries recover from road users more
than is spent on road track provision. There is also certainly no reason why, as
some have advocated, the ratio should approximate to unity. Part of the revenue
raised from road users must be considered as a ‘pure’ tax in the same way as
there are taxes on other expenditures. Also, there are social costs associated with
road transport (see Chapter 6) and motoring taxes may, in part, be seen as a
method, albeit a very imperfect one, of making road users aware of such costs.

BUTTON 9781786435668 PRINT.indd 157 27/04/2022 11:07

 158 TRANSPORT ECONOMICS, 4TH EDITION

Additionally, if prices in other sectors of the economy deviate from costs, there
are sound economic reasons for this to be also the policy on roads.

Development of the United Kingdom’s Road Track Costs Allocation Method

The experiences of the United Kingdom in developing mechanisms for the allo-
cation of transport track costs in the 1980s offers an insight into the economic
challenges involved and into the need for pragmatism when putting methodolo-
gies into practice.
While there is no sound reason why expenditure should match revenue in
aggregate, it may still be desirable for each class of road vehicle to cover its allo-
cated track costs. This is the view, for example, of the European Union (1998). It
has also become increasingly important as public–private partnerships (PPPs) are
playing a greater role in highway provision. The partners need to make contrac-
tual agreements covering the tolls levied on various types of uses (Button, 2016).
Allocation of track costs to vehicle categories is, therefore, still important. The dif-
ficulty is that roads provide a common service to a variety of modes of transport
(cyclists, motor cars, light vans, heavy lorries, buses, etc.) and the exact appor-
tionment of marginal costs is, therefore, far from easy. The method of allocation
developed by the UK Department of Transport is based upon a refined version
of an approach pioneered in the 1968 ‘Road track costs’ study which crudely
attempts to allocate long-run marginal costs to different classes of road users. The
methodology is similar to that developed in the ‘Federal highway cost allocation
study final report’ which has been used in the United States since the late 1990s.
The calculations are done for the four standard categories of road: motor-
way, trunk, principal, and other. Of total capital expenditure (made up of the
average expenditure over the previous year, the current year, and the forecast
for the following year), 15 percent is allocated directly to heavy goods vehicles
according to their maximum vehicle weight times the kilometers run, on the
grounds that they necessitate higher engineering design standards. The remaining
85 percent is allocated out on passenger car unit (pcu) kilometers (pcu being an
estimate of the amount of road required to accommodate a vehicle expressed in
terms of car equivalents), on the argument that capital expenditure is determined
by changes required in the physical capacity of the network.
Current maintenance expenditures are allocated according to a series of
ad hoc calculations that attempt to relate the various component items (such as
resurfacing, grass cutting, lighting, road markings, traffic signs, drainage, etc.) to
different vehicle characteristics, namely their size, number of standard axles (high
axle loadings doing considerable damage to road surfaces), and the use made of
roads. The criteria used to decide how costs are affected by the vehicle character-
istics are based upon ‘expert advice from traffic engineers and research scientists’.
Special items such as policing and car parks are treated separately.
Table 5.9 compares the cost allocations with revenues for different broad
categories of road users in the United Kingdom, for 1989–90. When aggregated,
road users cover their allocated cost, a situation which is also common in many

BUTTON 9781786435668 PRINT.indd 158 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 159

Table 5.9 
Allocated taxation/revenue ratios by vehicle type in the United Kingdom
(1989–90)

Vehicle class Tax/cost ratio

Cars/light vans/taxis 3.4:1

Motorcycles 2.3:1
Buses & coaches 1.1:1
Trucks over 1.525 tonnes gvw
Not over 3.5 tonnes gvw 1.3:1
Over 3.5 tonnes gvw 2.4:1
Other vehicles 2.4:1
All vehicles 2.6:1

Source: UK Department of Transport (1989).

other industrialized countries although not the United States. A similar picture
is seen within vehicle classes, although again the United States is something of
an exception, with many classes falling short in paying their fully allocated costs.
In 2000, for example, while automobiles and buses broadly covered their costs,
trucks did not. Additionally, even in countries where all sizes of heavy goods
vehicle pay tax in excess of the public road costs attributable to them, there is
considerable variation in the revenue/cost ratio and large vehicles with a small
number of axles tend to have the lowest ratios. This may be particularly unde-
sirable if these large vehicles are also responsible for generating a high level of
external costs such as pollution and noise.
The UK Department of Transport’s method of cost attribution has come
under some criticism. At one level, the detailed allocation within the Department’s
framework in the 1970s was biased against heavy goods vehicles because of:

• the excessive emphasis on vehicle weight in the allocation of capital costs

when much of the network’s design is determined by vehicle speed;
• the assumption in the calculation that the network is of optimal size;
• the rather dubious nature of passenger car units as a measure of road capac-
ity; and
• the ‘standard axle’ measure used in the calculations.

In particular, there may be an element of double counting when calculating

heavy goods vehicle costs. Cars are accredited with a small allocation of mainte-
nance costs, but this is only possible because of the high engineering standards of
roads required to carry heavy goods traffic. Hence, goods vehicles are allocated
both the additional costs of high design standards and the bulk of maintenance
costs whereas with lower design specifications, suitable for cars only, cars would
be allocated much higher maintenance costs.
This last point has been examined in some detail by Ken Small et al. (1989)
in a study of the United States’ road investment and pricing policy. The evidence
seen in Table 5.10 strongly suggests that under the United States regime, where

BUTTON 9781786435668 PRINT.indd 159 27/04/2022 11:07

 160 TRANSPORT ECONOMICS, 4TH EDITION

Table 5.10 
Examples of actual and optimal road vehicle taxation in the United States
(1988)

Vehicle Taxes (cents per mile)

Current Marginal cost Marginal cost
(existing roads) (improved roads)

2-axle single unit

28,000 lb 2.5 9.2 3.6
33,000 lb 3.0 23.8 9.3
5-axle tractor unit
33,000 lb 4.0 1.2 0.5
80,000 lb 7.2 41.3 16.2

Source: Small et al. (1989).

certain federal road taxation revenues are partly hypothecated to road building
and maintenance, the amount of investment has historically been suboptimal,
resulting in pavements that are, from an economic prospect, too thin. The result
is the need for high maintenance and reconstruction outlays relatively soon after
a road is opened.
The economic evidence is that, if higher taxes had been introduced initially
and higher-standard roads constructed, then subsequent taxation to finance
maintenance would, on average, have been much lower than it inevitably will have
to be if the network is not to deteriorate. The structure of the taxation should
also have been adjusted to more closely correspond to the damage associated with
various classes of user.
At a more fundamental level, Christopher Nash (1979) has suggested that the
traditional road track cost approach is really asking the wrong question and that
track costs should be allocated along altogether different lines. He suggests the
more appropriate way is to adopt a sequential approach where greater emphasis
is placed upon differing demand elasticities among road users. Specifically, he
advises:

• Forecasting traffic growth rates by vehicle type using alternative assumptions

about future taxation levels and structures.
• Estimating the full costs of catering for different traffic growth rates.
• Identifying the level and structure of taxes at which the revenue obtained
from an incremental slice of traffic matches incremental costs both for traffic
as a whole and for individual traffic types.
• To the extent that the resulting taxes fall short of government revenue
requirements, raising taxes on vehicle classes along second-best lines (that is,
according to demand elasticities).

Such calculations obviously place greater demands on informational sources,

but they do offer a rather more realistic basis for track cost allocation as well as
pricing consistent with principles adopted elsewhere in the transport sector.

BUTTON 9781786435668 PRINT.indd 160 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 161

Allocating Costs in a State-owned Railroad

Railway track cost allocation is carried out in a slightly different context because
the railways, unlike the roads, have traditionally been responsible in most coun-
tries for providing both track and rolling stock. This has changed somewhat
in some European countries, not only because in countries such as the United
Kingdom rail operations have been separated from track provision, but because
European Union regulations require open access to their networks and this
necessitates track cost estimations for charging purposes. In countries such as
the United Kingdom, the necessity for devising a method of allocation stems
not simply from designs for internal efficiency but also to permit the allocation
of common fixed costs between those services that are operated on commercial
criteria (for example, inter-city services) and those that are operated on social cri-
teria (for example, commuter services) and are given central government subsidies
(Bowman, 2015).
Here we focus on the situation where, as is still common, state ownership of
track and operations remains, and we look at some of the challenges faced in the
past by British Rail – a largely multi-purpose passenger system. The intensity with
which this subject has been studied provides insights of a more general nature.
In countries such as the United Kingdom, where track is often used by
several types of service, the necessity for devising a method of cost allocation pre-
privatization of rail track in 1993 stemmed not simply from a desire for internal
efficiency but also from the need to allocate common fixed costs between those
railway services that were operated on commercial criteria (for example, inter-city
services) and those that were operated on social criteria (for example, commuter
services) and were subsidized by the state. The situation in the United States is
different. There, the tracks are largely owned by the freight railroads and Amtrak,
the large national passenger carrier, and local services’ short-haul providers. The
various railroads largely allocate costs to meet their specific objectives. Class 1
freight railroads are, however, legally obliged to allow access to Amtrak services
and only to charge a marginal cost for this.
One of the difficulties with railway operations is that common costs (which
include signaling, termini, etc., in addition to track) form a substantial part of
the total cost. Normal commercial practice would be to use a ‘cost-plus’ method
of pricing so that each customer would pay a rate covering his/her specific costs
plus a contribution to overheads. Provided this results in all costs being recov-
ered in aggregate, the problem of common cost allocation is not a serious issue.
Unfortunately, for the reasons mentioned above, plus the difficulty of devising a
sufficiently sensitive price discrimination regime given the diversity of services
offered (see Chapter 7), the railways have found it important to be able to allocate
their track costs.
A major difficulty in this area is that the railways’ ‘jargon’ often does not
conform to conventional economic definitions. The railways talk of ‘direct costs’
and ‘indirect costs’, but the former (which embraces haulage costs, maintenance,
marshaling, booking, insurance, and collection and delivery by road) is clearly

BUTTON 9781786435668 PRINT.indd 161 27/04/2022 11:07

 162 TRANSPORT ECONOMICS, 4TH EDITION

different from the economic notion of short-run escapable costs (and may or may
not exceed them). As the United Kingdom’s Select Committee on Nationalized
Industries stated in 1960, ‘[t]he direct costs ascertained by traffic costing methods
are not the same thing as short run marginal costs. Nor do they correspond with
the savings that would flow immediately from the discontinuance of a small part
of railway activities’. Equally, indirect costs, as defined by the railways (that is,
track, signaling, and general administration) are not sufficiently fixed costs that
are common to all traffic. While certain costs (for example, earthworks) are invar-
iant with traffic, it is often possible to allocate track and signaling costs to services
according to causation. The type and density of traffic determine whether a
single-track route is operated with no signaling or a multi-track, multiple-aspect
signaling system is provided.
Stewart Joy (1973), the chief economist at British Railways, highlighted
almost 50 years ago how these costs can vary with the quality of service – an
express Category A service on double track with 12 trains a day cost £8,250 per
mile per annum in track costs (at 1961 prices); a less frequent, Category B service
cost £7,250; heavily used non-express Category C services cost £6,250; and slow,
Category D services cost £3,500. It is possible with poorer quality services to have
more basic signaling and lower track maintenance standards. Further, there are
quite significant differences associated with the costs of track used exclusively
for passenger services compared with track used only for freight. The Beeching
Report, which brought about major reforms in the British rail system in the
1960s, found that a single track maintained to passenger standards costs at least
£3,500 per mile per annum, but if it were only required to conform to freight
standards it would cost £2,000 per mile per annum, and it has been argued this
could have been reduced further.
The improvement in costing in the United Kingdom came about in part
because of the 1968 Transport Act and the introduction of social service subsi-
dies for specific routes – the system required identification and costing of those
services and facilities whose cost should properly be borne or aided by the com-
munity. The common costs were allocated according to the so-called ‘Cooper
Brothers’ formula (which was essentially an average, rather than marginal, cost
type of framework) that endorsed the idea of allocating track costs in terms of
gross tonne-miles and signaling costs on the basis of train miles. With homogene-
ous traffic flows evenly spread this is reasonable, but with mixed traffic and peaks
in use the allocation technique is unlikely to match causation with costs.
British Rail moved in the late 1970s to a system of ‘contribution account-
ing’, which entailed breaking down revenue and costs into some 700 major sub-
sectors (or ‘profit centers’). These profit centers, which are composed of single
traffic flows, groups of flows, or specific passenger services, are defined so that
resources allocated to them can be specifically identified with a minimum of con-
troversy. Even so, not all common costs could be so allocated and thus British
Rail accounts revealed the surplus of revenue over directly attributed expenses
that are a ‘contribution’ to the indirect costs. The sum of all avoidable costs
recovered may not cover all business costs, however, and a ‘basic facility cost’ is

BUTTON 9781786435668 PRINT.indd 162 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 163

likely to remain. British Rail argued though that this approach, given the high
proportion of indirect costs, ‘ensures a high level of certainty in profit assess-
ment’ (Dodgson, 1984).
The privatization of railways in the United Kingdom and the deregulation
of private networks in countries such as the United States has meant that there
has been more effort put into cost modeling of the type discussed elsewhere in this
chapter. The American railroads, a largely freight mover, have traditionally varied
in the ways that they have allocated their costs and generalization is not easy
(Poole, 1962). The removal of most economic regulation under the 1980 Staggers
Rail Act provided the Class 1 United States railroads with greater flexibility in
their pricing and the ability to close or sell off unprofitable lines. With this came
the need for more specific costing. The growth in importance of container traffic,
with its own unique set of costing issues, has provided a further stimulus for more
accurate costing, as has the common use of track with Amtrak.
Prior to 1980, the US Interstate Commerce Commission (ICC) regulated
rates on a rate-of-return basis – essentially a cost-plus model. For this it assigned
to each service those costs that could be directly and unambiguously attributed
to it. In addition to these directly attributable costs, each service was assigned a
portion of common costs in such a way that all such costs were allocated some-
where. This is what is often called a ‘fully distributed cost approach’. In other
words, the total allocated costs of C of a service were defined by an affine func-
tion of the form C = F + my, where F represents fixed costs insensitive to traffic
volume, m is a constant marginal cost, and y is the level of output.
John Meyer (1958) and others criticized this approach over the years, point-
ing to the arbitrary allocation of common costs across services. In doing this they
were resurrecting a point made in the late 1800s by Frank Taussig (1891), who
argued that ‘attempts have indeed been made at various times … to apportion the
expenses, and to assign to each item of traffic the sum which it cost … . Yet surely
the division is purely arbitrary’. Meyer also pointed to the inappropriateness of
the ICC’s assumption about output costs which assumed, for example, that it cost
as much to haul 100 tonnes one mile as to haul one tonne 100 miles. While any
attempt to allocate all costs to a particular output is unquestionably difficult, as
we shall show later in this chapter, the degree of arbitrariness has been reduced
somewhat with the introduction of econometric costing analysis.
From the two examples, road and railroad track costs, the problems of allo-
cating costs common to several services are, therefore, seen to be difficult ones.
Economic principles advocate the notion of seeking avoidable costs associated
with specific users and then allocating these accordingly. The problem is in defin-
ing the base from which to begin the series of allocations; in the case of roads,
are roads mainly designed for cars with lorries imposing additional costs, or are
they there to provide a quality of service with the faster car traffic necessitating
higher engineering standards? We have seen that it is possible to allocate many
items on an avoidable cost basis although practical application may necessitate a
high degree of averaging.

BUTTON 9781786435668 PRINT.indd 163 27/04/2022 11:07

 164 TRANSPORT ECONOMICS, 4TH EDITION

5.6 Transport User Costs and the Notion of Generalized Costs

Simple observation shows us that different types of transport are associated,

in a very broad way, with the distances of trips. Walking and cycling, together
with the riding of mopeds, are largely reserved for local travel, whereas air trans-
port dominates long-distance movements. Similar generalizations can be given
about freight transport, although in this case it is shipping that dominates long-
distance movements, at least in terms of physical movements. Of course, there
are cases where this very broad picture does not pertain. For example, where
there are stretches of water to be covered, it is often by boat or aircraft if there
is no road or rail infrastructure; and equally in mountainous regions where it is
costly to build surface infrastructure, air transport can play an important role.
The boundaries also tend to vary slightly with culture; the United States is a car-
oriented society and many people, even for the shortest of journeys, would not
consider walking or cycling.
Figure 4.1 in the previous chapter provided a very stylized picture of the
boundaries of the various transport modes. The cross-over points represent
‘distant boundaries’ over which the modes compete. This competition takes a
variety of forms. From the traveler’s point of view, or that of a consignor of
freight, a multiplicity of factors influences this type of decision.
Travelers in particular take notice of the time it takes to make a trip and
the money costs involved, and, frequently, also the quality of the service offered.
Consignors are concerned not simply with the financial costs of carriage but also
the speed, reliability, and timetabling of the service. The demand for transport
is not, therefore, simply dependent upon financial costs but rather on the overall
opportunity costs involved. Transport is not unique in this, but it does differ from
other services in that money costs may only form a relatively small part of overall
costs. In terms of decision-making, the money cost of a trip may have minimal
influence over whether it is undertaken or the transport mode preferred; a fact
that may explain the considerable use of private motorcars even when ‘cheaper’
alternative modes are available.
User costs are also, according to Herb Mohring (1976), important in the
urban public transport context for another reason. He argues that, in the context
of very frequent public transport in cities, the main scale economy effects are
associated with savings in passengers’ time. These economies exist when the
service is such that people do not arrive at a stop intending to catch a scheduled
bus, but, rather, know that headways are so close that at any time the wait will
be short. If say the average headway is ten minutes then the average wait, assum-
ing random arrivals of passengers, will be five minutes. A reduction of headway
to five minutes will reduce the average wait to two and a half minutes. Increased
output of bus services, therefore, reduces average waiting time and thus the users’
cost of traveling by bus.
In analysing transport demand or when forecasting future consumer
response, it is often possible to assess responses to the individual components of
overall cost, that is, money costs, time costs, inconvenience costs, and so on. In

BUTTON 9781786435668 PRINT.indd 164 27/04/2022 11:07

 165
DIRECT COSTS OF TRANSPORT ­

some situations, however, it has proved useful to have a composite measure. This
may be true in situations where multi-dimensional cost functions are unwieldy or
when a simple uni-dimensional measure, by focusing attention on general trends
in cost, permits a clearer understanding of changes in the demand for transport
services. A pragmatic device to reduce the wide range of costs involved in travel
is to employ a single index expressing ‘generalized cost’. This was a concept
developed by Leon Warner (1962) as part of the Chicago Area Transportation
Study.
The generalized cost of a trip is expressed as a single, usually monetary,
measure combining, generally in linear form, most of the important but disparate
costs that form the overall opportunity costs of the trip. On a rare few occasions,
a generalized time-cost measure has been advocated to replace the financial
index (Goodwin, 1974). The characteristic of generalized costs is, therefore, that
it reduces all cost items to a single index and this index may then be used in the
same way as money costs are in standard economic analysis. Simply, generalized
costs can be defined as:

G 5 g(C1, C2, C3, . . ., Cn) (5.4)

where G is the generalized cost and C1, C2, C3, … , Cn are the various time, money,
and other costs of travel. This permits the demand for trips to be expressed as a
function of a single variable – that is, QD = f(G). While, in simple indices, general-
ized cost is formed as a linear combination of time and money (or distance) costs,
in most applied analysis the time and money components are divided into several
elements (for example, walking time, waiting time, in-vehicle time, etc.). This
results in an expression of the general form:

G = ∑M i + ∑Tj (5.5)
i j

where M are the actual money costs of a journey (for example, fare or petrol
costs) and Tj are the time costs (for instance, in-vehicle time and waiting time
made up of the amount of time involved multiplied by the monetary value per
unit of time). These were discussed in detail in Chapter 4.
For expositional ease, we take a fairly simple specific form of the general-
ized cost function that was used some time ago in the United Kingdom as part
of the South-east Lancashire, North-east Cheshire (SELNEC) transport study
conducted in the late 1960s (Wilson et al., 1969), which provides a useful illustra-
tion of actual application. It is dated, but it makes the key points rather neatly.
The generalized cost index used in the combined trip-distribution–modal-split
element of the analysis was of the form:

GijK = a1tijK + a2 eijK + a3 dijK + p Kj + ζ k (5.6)

BUTTON 9781786435668 PRINT.indd 165 27/04/2022 11:07

 166 TRANSPORT ECONOMICS, 4TH EDITION

where:

GijK is the generalized cost of travel by mode K between points i and j;

tijK is the travel time from i to j by mode K (in minutes);
eijK is the excess time (for example, waiting time for public transport) for the
journey from i to j (in minutes);
K
dij is the distance from i to j that acts as a surrogate for the variable money
costs of trips (that are assumed proportional to distance);
pjK is the terminal cost (for example, parking charges) at j (in pence);
ζK is a modal penalty reflecting the discomfort and lesser convenience associ-
ated with public transport journeys; and
a1, a2, and a3 are parameters which, since pjK and ζK have unit coefficients,
value other cost items in monetary terms.

An important issue is how these costs should be treated in analysis of trans-

port decisions. Economics is concerned with costs that influence behavior in the
short term and with those that affect long-term decisions. In the short term,
people may well only perceive a limited range of costs or not fully appreciate
the full magnitude of some cost items. Nevertheless, it is this set of costs which
influences their immediate actions. The problem of perception is generally associ-
ated with the external costs that travelers generate by ignoring their actions (see
Chapter 6), but here we are concerned with the misperception of the costs they
bear themselves.
People misperceive the costs of their journeys (or of moving goods) for
several reasons:

• This is the first time the trip is being made and the trip-maker has incomplete
information of the costs involved.
• The money or time cost may be so small that it is not worth taking into
account.
• Certain variable costs may be regarded wrongly as fixed costs; included here
would be the tendency for car users only to take account of petrol costs of
journeys and ignore depreciation of the vehicle and its maintenance.
• Users may be unaware of the connection between a particular action and the
costs to which it gives rise; for example, a fast driver may be unaware of the
additional fuel costs he/she incurs.
• Habit can make regular trip-makers unaware of changing cost conditions
over time even if they were fully cognizant of the full resource costs of their
actions at some earlier point in time. This is more likely to be a problem
encountered by car users than public transport travelers who face regular
ticket purchases.

While the final three reasons for misperception result from poor or inadequate
information, the first represents a departure from the conventional economic idea
of maximizing behavior. While this poses interesting theoretical questions, there

BUTTON 9781786435668 PRINT.indd 166 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 167

are reasons for arguing that the last three are likely to be of greater quantita-
tive importance for transport economists. Lack of good information is likely to
result in different travel behavior to that anticipated in full information situations.
Whereas perceived generalized costs offer a basis for travel behavior analysis, it is
actual resource costs that are appropriate for investment decision-taking. Where
people accurately perceive the costs of their travel there is no difference between
the perceived and resource generalized cost. Where there is m ­ isperception,
however, resource costs, being the full opportunity cost of trip-making, will
exceed the perceived costs and this may result in over-investment in transport
facilities if adjustments are not made. (Of course, we are still ignoring external
costs such as pollution, but these complicate rather than change the argument.)
The social welfare gains associated with an investment are assessed by com-
paring the resource costs with the benefits generated. The difficulty is that the
actual traffic levels using the facility depend upon perceived costs. In Figure 5.11
we have a linear demand curve for use of a road with an initial perceived general-
ized cost of usage equal to P1. A widening of the road speeds traffic, causing the
perceived generalized cost to fall to P2.

Exhibit   Measuring perceived costs of driving

While accountancy calculations may be used to estimate the resource costs of making a car
trip, for the monetary costs perceived by the driver (which are needed for examining travel
motivations) either revealed- or stated-preference methods have to be used. A stated-
preference study conducted in 2001 involving questioning 416 Israeli drivers considered a
variety of factors influencing drivers’ perceptions of trip costs. These were:
• Fuel consumption is a variable cost affected by travel distance (F).
• Insurance is a fixed cost (I).
• Maintenance is a variable cost covering routine maintenance including tires and oil (M).
• Depreciation includes the reduced value of the car and interest rates. This is usually
considered a fixed cost, but part of it is also a variable cost as the future value of the car
is also a function of its mileage (D).
• Repairs refer to mechanical and accident repairs to the extent that they are not covered
by insurance. They are distance-based (R).
• Licensing is a fixed cost (L).
It was found that the person vehicle size affected perceptions of the cost of using it, as
did whether drivers were asked about daily costs or monthly costs. The first table below
provides the various response rates and the values of vehicle costs as perceived by owners
who answered both questions. The standard deviations are large, indicating rather poor
perception among drivers.
In terms of which costs drivers consider when trip-making, the table below shows clearly
the importance of fuel costs, although these are often combined with other factors. Very
few respondents considered all four possible elements.

BUTTON 9781786435668 PRINT.indd 167 27/04/2022 11:07

 168 TRANSPORT ECONOMICS, 4TH EDITION

Average perceived cost per kilometer by vehicle size

Number of Response Average Standard
observations rate (%) (NIS)a deviation

All vehicle sizes (monthly estimate) 360 87 1.01 1.19

All vehicle sizes (daily estimate) 332 80 1.45 2.69
Small vehicles (monthly estimate) 110 88 1.01 0.97
Small vehicles (daily estimate) 103 82 1.03 0.76
Medium vehicles (monthly estimate) 204 86 1.03 1.23
Medium vehicles (daily estimate) 186 78 1.71 3.49
Large vehicles (monthly estimate) 46 90 0.77 0.55
Large vehicles (daily estimate) 41 80 1.30 1.07

Note: a. NIS = new Israel shekels.

Comparison of cost elements

Element mentioned Number Response Average Standard
of cases rate (%) NISa/km deviation

At least fuel (F) 400 96 1.03 1.21

Fuel (F) & maintenance (M) 73 18 0.94 0.57
Fuel (F) & insurance (I) 88 21 0.85 0.57
Fuel (F) & depreciation (D) 82 20 0.92 0.88
F & at least one other element 159 38 0.92 0.75
F & M, but not I or D 25 6 0.99 0.59
F & I, but not M or D 31 7 0.88 0.66
F & D, but not M or I 34 8 1.08 1.20
F & M and I 38 9 0.91 0.57
F & M and D 29 7 0.89 0.61
F&I&D 38 9 0.78 0.53
All four elements 19 5 0.87 0.63
None of them 14 3 0.70 0.65

Note: a. NIS = new Israel shekels.

Feeding the results into a regression model to examine factors affecting these variations in
the perceived monetary costs of driving produced the following:

Perceived costs (NIS/km) = 1.00E + 00

+ 1.97E – 04 Engine size (cc)
+ 1.95E – 0.2 Vehicle age (years)
– 1.40E – 04 Vehicle use (km/month)
– 9.50E – 0.2 Gender
– 6.32E – 03 Person age (years)
+ 1.79E – 03 Commuting time (minutes)

The importance of the vehicle age and engine size is clear from the size of their coefficients,
and supported by high t-statistics. The usage coefficient is negative, large, and highly

BUTTON 9781786435668 PRINT.indd 168 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 169

significant, supporting the idea that the more a driver uses a vehicle the lower the perceived
costs. It appears that longer-distance commuters have a higher perception of motoring
costs. Overall, the results are suggestive of much lower perceived costs than the resources
cost of car use in Israel.
See also: Y. Shiftan and S. Bekhor (2002) Investigating individual’s perception of auto travel
cost, International Journal of Transport Economics, 29, 151–66.

Generalized
costs ($)

a
F2
b c

P1
d e
F1
f g h

P2 Demand
i j

0 t1 t2 Traffic flow
Figure 5.11 Welfare gains from a cost reduction with misperceived transport costs

If, however, the actual resource costs of trip-making along the road are F1 and P2
for the respective pre- and post-investment situations, then there will be ‘dead-
weight’ welfare ‘losses’ generated at both the t1 and t2 traffic levels. (At the pre-
investment traffic flow, t1, this loss is equal to area c and at the post-investment
flow, t2, it is h.) If no account is taken of this, however, the apparent consumer
surplus gain from the road widening is equal to (d + e + f + g). In fact, since
the genuine resource costs are measured by F1 and F2, the investment will result
in a net benefit of (b + c + d + e – h). The area (b + e + d) represents a straight
resource cost saving under the demand curve by reducing the resource costs of
travel, while (c – h) reflects the change in deadweight welfare loss between the
two traffic flow situations. Henry Neuberger (1971) generalized this calculation
to take account of the effects of policies that alter costs and travel patterns over
a network of roads.
The adoption of a single index idea of transport costs has permitted signifi-
cant advances in transport forecasting and project appraisal to be made. This does
not mean, however, that the concept is without its critics, nor that those choosing
to ignore its existence have not made other advances. First, the inherent constraints
implicit in the aggregating of the various cost components into a unique index
restricts the separate elasticities of demand with respect to each individual cost

BUTTON 9781786435668 PRINT.indd 169 27/04/2022 11:07

 170 TRANSPORT ECONOMICS, 4TH EDITION

component. One ends up with an elasticity with respect to generalized costs but
cannot, for example, assess the particular effect of a reduction in travel time costs.
Second, there is concern about the long-term stability of money as the
numerator. Because income rises over time, it is argued, the utility of money will
fall relative to other items, especially time which is fixed in quantity. McIntosh
and Quarmby (1972), therefore, argue that time should be used as the basis of
measurement and the operational concept should be generalized time costs.
Additionally, time is equally distributed (in the sense that everyone has 24 hours
in a day) which circumvents some of the distributional difficulties of using money
values of travel cost components. Third, even if the basic notion of general-
ized cost is accepted, there are critics who oppose the use of a ‘universal’ index
for application throughout a country, for example of the form McIntosh and
Quarmby (1972) put forward for the United Kingdom some years ago. A dif-
ficulty is that there is little evidence to support the universality of any weighting
scheme employed. Although the use of official time valuations and formulae
ensure consistency in approach, they may lead to inconsistencies in the results if
the overall index is only accurate in certain sets of circumstances.
Generalized cost is, despite these criticisms, a useful tool in helping us to
understand, in broad terms, how variations in travel cost can influence travel
behavior. Above all, it is an extremely useful pedagogical device that can help
policy-makers articulate their ideas and plans to a more general audience. It
also serves as a pragmatic device for assisting in certain types of modeling and
decision-making where, otherwise, no information would be forthcoming at all.
In this context the index is likely to be an imperfect instrument, but, when used
with sufficient circumspection, it can yield useful insights into the possible effects
of alternative transport policies.

5.7 The Bunching of Public Transport Services

There often seems to be a bunching of buses on the roads, or a concentration of

flights leaving at about the same time for a destination. In the bus case, this may
be simply caused by traffic conditions that do not allow buses, with the frequency
of their stops and their relative lack of dexterity, to keep up a schedule that on
paper has more regular headways between them. In the airline case it can some-
times be explained by airport noise curfews or ‘banks’ of flights arriving and
leaving to facilitate passengers making connections. But in many cases bunching,
and similar service features, results from deliberate actions on the part of sup-
pliers of services. This may be because demand peaks at certain times of the day
and, therefore, that is the time that customers are willing to pay for having the
‘bunch’ of vehicles available.
This logic extends to the large number of taxi-cabs that are at certain times
waiting at airports, railway stations, and other terminals. But the nature of supply
and the product on offer can also lead to situations in which the bunching is less
desirable. This can also occur in terms of where transport infrastructure, such

BUTTON 9781786435668 PRINT.indd 170 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 171

as gas stations, is located. Setting aside planning controls, even where there is a
relatively free choice, gas stations tend to be physically bunched.
The explanation for this phenomenon, and the account of the costs that
it creates, can be traced back to the work of Harold Hotelling (1929) nearly a
century ago. While his analysis was really designed to explain business location
decisions and why manufacturers in markets with a limited number of large
­producers – known as oligopolies these days – tend to produce very similar goods,
it also has relevance to the issue of transport service bunching. The analysis can
be explained by recourse to Figure 5.12.
The distances between points A ⇒ B and B ⇒ C in the figure represent ten-
minute intervals. If potential users are arriving continually, and at an even rate,
then this flow can be taken as the horizontal line of uniform thickness joining A,
B, and C. We introduce two bus companies, X and Y, that offer identical fares
and an infinite capacity to carry passengers. Thus, passengers are indifferent as to
which company’s bus they make their trip in. Initially they both operate their ser-
vices at 20-minute headways, each offering seats midway along a time period. The
average wait for a passenger is thus five minutes: if someone just misses X’s bus
and must wait for Y’s bus then they have a ten-minute wait, but someone arriving
coincidentally in time for Y’s bus has no wait. However, while this is an optimal
situation, it is not an equilibrium because there is an incentive for the bus compa-
nies to reschedule. X, for example, may reschedule their service to run just prior to
Y’s bus, say at X*. This will allow them to pick up virtually all the passengers who
were waiting for Y’s bus. Company Y, to remain in business, will then reschedule
their service to, say, Y*, and thus take away most of the passengers that would be
waiting for the X* service. This goes on until the companies time their 20-minute
headways to coincide with each other – at X† and Y†.
This is an equilibrium because there is no incentive for either supplier to
change their services; alterations will mean smaller numbers of passengers. It is
not, however, optimal from a passenger’s perspective because now the average
waiting time is ten minutes: missing the X bus means a wait of 20 minutes for

A B C
X Y

X Y* X* Y

Y† X†

Figure 5.12 The bunching of transport services

BUTTON 9781786435668 PRINT.indd 171 27/04/2022 11:07

 172 TRANSPORT ECONOMICS, 4TH EDITION

either X’s or Y’s next service, with no wait for those that just turn up when a bus
is due.
The nature of the competition and common costs for the two suppliers effec-
tively cause an external cost for potential bus users, a topic we return to in Chapter
6. Of course, the example is excessively simple, but there are records from the
early part of the twentieth century of bus companies engaging in a whole series of
practices akin to this – for example, missing out passengers at some stops to move
ahead of another company’s buses to collect a large line of customers later on in
a route (Foster and Golay, 1986). The degree of welfare loss due to a high level of
waiting time is likely to be dissipated because consumers will not arrive at even
intervals but will play their own ‘games’ with the bus companies to minimize their
waits. Also, the equilibrium conditions tend to break down with more than two
bus operators and when complex networks of services come into play.

5.8 Economic Performance

It is generally assumed that firms seek to minimize their costs, and this is a
common assumption when considering transport markets. In practice, firms
perform with varying degrees of efficiency. Further, performance can be measured
in many dimensions; it may be in terms of whether costs of production are being
minimized (X-inefficiency), whether output is at an optimal level (allocative effi-
ciency), or whether resources are being wasted in protecting a particular, normally
legal monopoly, market position (the so-called ‘Tullock rent-seeking inefficiency’).
We take the example of an airport and an airline and look at the potential
implications for these various forms of efficiency when there are different market
structures. The interest is in how actors seek to acquire rent (or profit) when
there is no genuine capacity problem; in other words, when they can exercise
market power. Figure 5.13 represents, for simplicity, the short-run costs (MC)
and demand (D) for take-off slots associated with a single-runway airport.
For ease of presentation the demand curve is drawn as linear, and the cost
curve is horizontal. In most natural monopolies, because of various scale and
scope economies, the cost curves would almost certainly be downward-sloping.
Introducing this, however, while more realistic, adds complexity but not sub-
stance to the argument.
There is assumed to be adequate capacity to meet all effective demand cur-
rently in the market. If there were a binding physical capacity constraint at a level
before QC, as strict classical theory would require, then the MC curve would
become vertical at the point where capacity is reached. This does not affect the
fundamental tenure of the rent-seeking argument as presented here, and is thus
ignored. But in these circumstances, charges become a rationing device for the
limited capacity and a reflection of the costs incurred in providing the existing
capacity, as well as a means of extracting economic rent.
The issue is the simple one of the level and distribution of benefits from
alternative allocating mechanisms assuming that only airlines and an airport

BUTTON 9781786435668 PRINT.indd 172 27/04/2022 11:07

 173
DIRECT COSTS OF TRANSPORT ­

$
e

d
PM*
a
PM
g
C* MC* = AC*
f
b
PC MC = AC
c

MR D

0 QM* QM QM+ Qc Output

Figure 5.13 I llustration of potential inefficiencies in a market for airport

take-off/landing slots

are involved. It is assumed, unless stated otherwise, that both the airlines and
the airport are seeking to be economic rent maximizers. This may be a strong
assumption for those in public ownership where there may be valid arguments for
considering other possibilities such as ‘satisficing’ with a series of lower objectives
being sought, but it does provide the ability to keep the analysis manageable.
The ultimate outcome will depend on the nature of the market and the power
of buyers and sellers:

• Competitive airlines/competitive airport. This yields the standard Pareto first-

best outcome of a price of PC and an output of QC. The combined rent of
buyers and sellers is maximized as the area ebPC in Figure 5.13. The airlines
enjoy the economic surplus from the use of the airport in this case because
of the horizontal nature of the cost curve; with a positive slope the airport
would also enjoy some rent, but with a negative slope it would be losing
money. The same outcome in terms of overall slot use would emerge if the
airport were a monopoly that is regulated, requiring it to marginal cost price
its slots or allocate them using a Vickrey style of auction. The highest bidder
wins but pays an amount equal to the second-highest bidder, although here
the airport would extract the rents from the airlines.
• Monopoly airline (or strong strategic alliance)/competitive airport. Since the
airport has no market power, the airline, or cartel of airlines, will be able to
force it to price its slots down to the airport’s marginal cost of providing the
slots, PC. The airlines will again gain the economic rents, again ebPC.
• Competitive airlines/single-pricing natural monopoly airport. This is the
traditional neoclassical case of a monopoly with slot prices reaching PM
and the number of slots limited by the airport to QM. The gains from the

BUTTON 9781786435668 PRINT.indd 173 27/04/2022 11:07

 174 TRANSPORT ECONOMICS, 4TH EDITION

airport’s activities are shared between the airlines (ePMa) and the airport
(PMacPC). But there is a welfare loss equal to the area abc – the Harberger
(1959) ‘deadweight loss triangle’. This is the sort of institutional struc-
ture that one could imagine under a regime where the airport sets its slot
prices at PM and then allows any airline to take up slots at this rate with
subsequent slot-swapping allowed. If, once the slot fees have been paid,
the airlines can buy and sell slots, then there would be transfers of benefits
between them to the extent of aPCd with no effect on either the airport’s
rent or aggregate welfare.
• Competitive airlines/single pricing institutionalized monopoly airport. Here,
the price and output combination is the same as the above, but the airport
will potentially enjoy a smaller rent and overall social welfare will be less.
There is likely to be rent dissipation in this type of situation. The airport, to
retain its institutional monopoly, will be willing to expend up to PMacPC in
potential economic rent to defend this position – for example, through lob-
bying, political support, advertising, or legal actions. Even a small amount of
rent is preferred to none. Since such actions are essentially non-productive, at
least in a non-Keynesian world, not only does it cost the airport money but
it also represents a loss of social welfare.
• Competitive airlines/perfect price discriminating natural monopoly airport.
In this case, through such actions as the auctioning of individual slots in a
manner that allows first-degree price discrimination, the airport can extract
slot prices down the demand curve. This action produces for the airport
economic rent of ebPC with the airlines having none. The outcome is Pareto
efficient and maximizes welfare. De facto this means that the demand curve
represents the airport’s marginal revenue curve with price differentiation.
The outcome is Pareto optimal.
• Competitive airlines/single natural monopoly airport with X-inefficiency.
Harvey Leibenstein (1966) argued that monopolies have limited incentive
to minimize their costs but rather they often operate in an ‘inert area’ where
there is little pressure to seek maximum efficiency. (Figure 5.14 illustrates the
notion of an inert area. Managers can make trade-offs between the profits
they may make by affecting costs and the efforts they put into the business.
The additional profits tend to taper away with additional effort because more
problems are encountered. At some point, E*, the additional effort (EOp –
E*) is just not seen to justify the additional profit of (ΠMax – Π) required
to maximize profits. Hence costs are not minimized.) This inertia may be
brought on by a variety of factors such as the considerable managerial
effort required to negotiate labor contracts when confronted by labor union
demands. If this is the case then the cost curve will rise to MC* in Figure 5.13
and the single profit-maximizing monopolist will limit the supply of slots to
Q*M and extract rent amounting to P*MdfC* The airlines will be left with
edP*M.
• Competitive airlines/perfect price discriminating natural monopoly airport with
X-inefficiency. The situation here is that the airport will provide Q+M and

BUTTON 9781786435668 PRINT.indd 174 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 175

Profit

Max

0 E* EOp Effort

Figure 5.14 The notion of inert areas

enjoy economic rents of egC*, with the airlines taking none. Compared with
the single-pricing monopoly situation where there is no X-inefficiency, the
aggregate social welfare is, however, reduced by PMagC*.

This categorization does not consider all possibilities – for example, there
is no consideration of X-inefficiency among the airlines or of collusion between
the parties, but it does indicate the importance of the institutional, as well as
technical, characteristics affecting efficiency in determining the levels of profits
­generated and the recipients of this rent.

5.9 Costs and the Measurement of Economic Efficiency

One can consider whether a transport undertaking is efficient, or whether its

efficiency changes over time, by looking at either the costs of producing its
outputs, which is the way we have discussed the subject so far, or the output that
is generated from the inputs used: productivity. These measures are mirror images
of each other. Cost functions relate costs to output quantities and input prices;
production functions relate output quantities to input quantities.
Productivity improvement can come about in several ways. It could reflect
reductions in inefficiency using existing technology, a shift in technological knowl-
edge and capabilities, and/or differences in environmental or operating circum-
stances that affect input/output use. An example of the latter is a transport firm
operating in adverse terrain. It will require more inputs per tonne of cargo moved
than firms operating in easy terrain; this is separate from questions of technologi-
cal prowess. The concept of productivity of greatest importance is often that of
technological change or ‘shifts’ in productive abilities. Improving efficiency using
existing knowledge will encounter limits. Major long-term advances in industries
and society’s wealth are caused by changes in productive abilities.

BUTTON 9781786435668 PRINT.indd 175 27/04/2022 11:07

 176 TRANSPORT ECONOMICS, 4TH EDITION

Returning to the strict cost side of the equation, the traditional method of
assessing whether an undertaking is efficient was to rely on accountancy proce-
dures that involved compiling data in relation to the various cost categories of
interest and connecting these to the firm’s outputs in a largely intuitive manner,
generally assuming a linear relationship between the various costs and outputs.
While useful for short-term decision-making, the technique lacks any explicit
causal link between inputs and outputs, which limits its usefulness for long-term
decision-making.
Separating the impacts of fixed and variable costs is particularly problem-
atic. In the transport context, a somewhat more advanced accountancy frame-
work is described by Ken Small and takes the form:

C 5 c1RM 1 c1PV 1 c1VH 1 c1VM (5.7)

where:

RM is route-miles;
PV is peak vehicles in service;
VH is vehicle-hours; and
VM is vehicle-miles.

The accounts imply constant returns to scale, with total costs (C  ) increasing
by the same factor (c1) as any common increase across all the right-hand-side var-
iables. If, however, RM is kept fixed and the other factors increase by a common
factor, then the increase in costs will be less than this, implying economies of
density.
There are two broad approaches to quantitative performance ratings that
are now more widely used than accountancy measures. The first are statistical
or econometric estimations of production or cost functions and the second are
non-parametric or programming approaches. Before looking at these in detail,
there is also a more basic productivity indicator that entails ‘benchmarking’. This
is a generic concept referring to a process of identifying similar organizations
or production activities, gathering quantitative and impressionistic data about
the similarities and differences across the enterprises, and drawing conclusions
about which are the most effective, and hence what other enterprises can learn
from them. This may include quantitative performance comparisons but is not
restricted to that. It often includes management organizational structures and
practices. We do not discuss this in detail, but it is often used in practice (for
example, by EUROCONTROL, the body overseeing European air traffic control,
when comparing the efficiency of individual centers) and focuses on the more
rigorous econometric and program approaches.

BUTTON 9781786435668 PRINT.indd 176 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 177

Statistical Cost and Production Functions

Statistical functions, with parameters normally estimated using regression analy-

sis, are much more widely used in economic analysis. A statistical production
function relates output to levels of inputs, but because it is difficult to estimate
a production function when firms produce more than one output, cost function
approaches have been developed based on duality theory. A cost function is the
dual to production technology in that it translates in a one-to-one fashion. The
cost function can also be applied to multiple-output situations. Duality theory
also recognizes that input prices replace the need for explicit measurement of
input quantities. A cost function specifies the minimum cost of producing a given
output obtainable from an input vector given the production technology, that is:

C = C(w, y, t) (5.8)

where:

C is the cost;
w is a vector of input prices;
y is a vector of outputs; and
t is the state of technology. (The production function approach relates
outputs to inputs.)

Because the estimation involves the costing of producing, y, given the input
vector, there is also the need to specify the motivations of the transport sup-
plier and the technical relationship between the inputs and outputs involved.
Traditional econometric methods for estimating cost or production functions
implicitly assume that all firms are successful in reaching the efficient frontier, and
deviate only randomly. If, however, firms are not always on the frontier, then the
conventional estimation method would not reflect the efficient production or cost
frontier against which to measure efficiency. For this reason, many studies now
estimate frontier production or cost functions which recognize that some firms
may not be on the efficient frontier. The reasons for this we explore in more detail
in Chapter 6.
Logarithmically differentiating the cost function with respect to time decom-
poses the rate of growth of total cost into its sources: changes in input prices,
growth of output, and rates of cost reduction due to technical progress:

∂lnC  N ∂lnC ∂lnwn   ∂lnC ∂lny   ∂lnC 

= ∑ + +  (5.9)
∂t  n =1 ∂lnwn ∂t   ∂lny ∂t   ∂t 

The rate of technical progress equals the negative of the rate of growth of
total cost with respect to time, holding output and input prices constant. In a
regression, this is the parameter measuring the shift in the cost function over
time. When it comes to introducing technology into estimations, the accountancy

BUTTON 9781786435668 PRINT.indd 177 27/04/2022 11:07

 178 TRANSPORT ECONOMICS, 4TH EDITION

approach, as seen in equation (5.7), assumes a linear relationship between inputs

and outputs, often referred to as a Leontief technology. This means that there is
no possibility of substituting one input for another in the production process.
The Cobb–Douglas cost function, while being limited by having constant
returns to scale, has been widely used in transport studies; for example, histori-
cally in the analysis of economies of scale in the British bus industry by Lee and
Steedman (1970), and American railroads by Keeler (1974). It has the advantage
that it does allow some factor substitution in production, and it is relatively
easy to estimate the coefficients in the model using regression techniques when
expressed in natural logarithms. Its general form is:

lnC = β0 ∑ βi lnyi + ∑ γ j lnw j (5.10)

i j

where i denotes the inputs and j the outputs.

There may be systematic differences between firms otherwise employing the
same technology, for example, differences in terrain or market location. Network
variables could allow for differences in terrain, weather, or exogenous character-
istics of the market area served, such as more favorable directional flows of cargo
and/or shorter average lengths of haul. Firm dummy variables (which by taking
the value of unity or zero distinguish whether a firm is in a specific grouping or
not) and firm-specific trend variables are sometimes incorporated to measure
differences in productive efficiency across firms and over time. By controlling for
these factors, the estimate with respect to time is a measure of the rate of techni-
cal change, that is, how the cost function is shifting over time. Bitzan and Keeler
(2003) further separated out specific productivity changes (reduction in crew size
and elimination of caboose) and show that there were sustained productivity gains
in addition to the specific advances after railroad deregulation in the United States.
A major constraint in using stochastic frontiers such as the Cobb–Douglas
specification (and others such as the constant elasticity of substitution model) is
the need to specify the detailed form of relationship between inputs and outputs.
Additionally, none provides functions akin to the ‘U-shaped’ functions that are
the norm in economics as initially costs fall with production but ultimately rise
as inputs become scarcer and thus more expensive. To circumvent this problem,
variable ‘flexible’ forms of function have been developed. Perhaps the one most
widely used in transport analysis is the translog model. This takes the general
form in the case of total cost of:

ln C = β0 ∑βi lnyi + ∑γ j lnw j + ∑∑δij lnw j ln y j

 i j i j
(5.11)

+ 0.5 ∑∑eik ln yi ln yk + 0.5 ∑∑Φ jl lnw j ln wl

i k j l

with the following cost share equations also being required:

δ lnC
Sj =
 = γ j + ∑δij lnyi + ∑Φ il lnwl
δ lnw j i l

BUTTON 9781786435668 PRINT.indd 178 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 179

The flexible form approach does not need any prior specification of technol-
ogy; it emerges from the estimation process. For example, in early work using
the framework, Allen and Liu (1995) found that traditional ideas that American
trucking was characterized by constant were incorrect and that results using
a translog model showed they enjoyed increasing returns to scale; and Anne
Friedlander (1993), by holding capital inputs constant, found increasing returns
to density on the United States railroads, as did Caves et al. (1984) for airlines and
Singh (2005) for Indian urban buses.
The usual emphasis in cost function estimation is on the overall results, that
is, the parameter values such as the degree of scale economies and measure of
technological change. These are based on the common relationship revealed by
statistical estimation across the set of firms and years. For performance compari-
sons, one also looks at individual firms relative to the estimated cost function. In
estimating a cost function from combined cross-sections of firms and time-series
data, it is common to include firm- and time-specific variables to remove the mean
effect of systematic departures from the industry norm of firms or years. This
is done to correct for possible omitted variables in the regression. For example,
a trucking firm might enjoy unusually low costs because of some overlooked
network advantage – perhaps the cities it serves are closer together than those for
other firms. The cost and output data for this firm could distort the coefficients
being estimated across the industry.
This practice, though, can sometimes prove to be counterproductive.
Expressed as a ‘fixed effects model’, dummy variables are used to remove the
mean deviation of the firm’s observations from the common regression estimate.
The firm in question may, however, be the largest in the industry. It might be scale
economies explaining the firm’s superior performance, and one would want to
have left the firm’s data unadjusted precisely to help measure scale economies. In
terms of productivity comparisons, the firm dummies are capturing some of what
is being sought: how does the firm compare to others? If it has systematically
lower costs and this cannot be explained by known variables, the residual may
be taken as an indicator of managerial performance. The value of a firm-specific
dummy variable could be a performance indicator, but again, it could be that
there are other factors explaining the firm’s apparently superior performance. It
is, therefore, important to examine the deviations of firms and years to see if there
are possible explanations that need to be taken into account.
The total cost function formulation assumes that firms adjust all inputs
instantaneously as outputs change. However, they may not be able to adjust them
all (especially capital stocks) as outputs change. To account for the short-run dis-
equilibrium adjustment in capital stock, Gillen et al. (1990) in their work on air-
lines, and others, estimate variable cost functions, in which capital stock is treated
as a fixed input, that is, the capital stock enters the cost function rather than the
service price of capital. Separating out the influence of capital stocks and capital
utilization is an important element in productivity studies, that is, distinguishing
between productivity gains from capital investment or increased utilization of
capital, as distinct from actual shifts in the function (technological change).

BUTTON 9781786435668 PRINT.indd 179 27/04/2022 11:07

 180 TRANSPORT ECONOMICS, 4TH EDITION

Productivity and cost measures, irrespective of the way they are derived,
assume that the quality of service provided is constant. This is a long-recognized
shortcoming of most productivity measures, and the criticism remains. In the
absence of acceptable quality measures, productivity measurement is biased
because it measures quantity changes but not quality. Improving quality absorbs
inputs, but the higher-quality output is not recognized except partially by a shift
in weights if prices for higher-quality services rise relative to others. The inability
to incorporate service quality is a major weakness of productivity measures.
An issue related to quality is capital utilization. If the flow of capital services
is measured as strictly proportional to capital stocks (that is, the capital depreci-
ates with the passage of time rather than actual use), then productivity gains
can be obtained via higher utilization of the fixed or indivisible capital stocks.
But high utilization of capital may be accompanied by deteriorating service
such as congestion delays. Insofar as the deterioration of service is manifested
by increases in the use of other inputs, this will off-set the seeming productivity
gain. But if congestion manifests itself in decreased service quality, standard pro-
ductivity measures do not capture this. High utilization rates of capital may thus
appear to imply high productivity, but this might, in part, be misleading if there
is deterioration in unmeasured service quality.

Programming Measurements

Management programming methods directly compare quantities of outputs with

quantities of inputs. The most widely used is data envelopment analysis (DEA),
which is a mathematical programming technique that generates quantitative
relative performance scores across ‘decision-making units’ such as bus companies
or shipping lines, where multiple outputs and inputs are involved but there is
no basis for assigning relative weights to the various inputs and outputs. Index
numbers compare the growth in aggregate output to the corresponding changes
in aggregate inputs. The aggregates are weighted sums of the growth rates of
respective outputs and inputs, with economic weights (for example, revenues and
costs) assigned to the outputs and inputs.
Programming measures of efficiency involve index number procedures that
construct a ratio-type productivity/efficiency measure from measures of outputs
and inputs. They are non-parametric; that is, they are a direct numerical calcula-
tion, in contrast to the statistical estimation used in the production econometric
cost function approach. They may take several forms, including partial produc-
tivity ratios, such as DEA, and total factor productivity (TFP). Once a measure
of productivity is developed, it can be ‘decomposed’ into sources of productivity
gains to isolate pure technological change in contrast to other influences on pro-
ductivity, such as X-inefficiency.
Partial productivity measures relate a firm’s output to a single input factor
such as revenue tonne-kilometers divided by the number of employees to measure
labor productivity. This is tracked over time or compared with other companies
or operations. However, the productivity of any one input depends on the level of

BUTTON 9781786435668 PRINT.indd 180 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 181

other inputs being used; high productivity performance in one input may come
at the expense of low productivity of other inputs. Despite this limitation, partial
productivity measures remain in wide use, and can provide useful insights into
causes of differential performance across firms operating in similar operating
environments or over time within a firm when the operating environment and
input prices remain relatively stable.
DEA becomes useful when firms employ several inputs and generally
produce numerous outputs. It can be used for ranking the relative efficiency of
decision-making units, say different urban bus fleets or airports, in this context.
With multiple outputs and inputs, weights are usually needed to aggregate the
various output and input categories to construct an index to rate different units.
DEA uses linear programming to solve for a set of weights that will maximize
each unit’s or firm’s performance rating relative to others in the dataset. In its
most basic form, we can take a case where W and Z are the inputs in an airline-
wide sample of n carriers and w and z are the corresponding observations of a
typical airline. The airline’s efficiency index, Θ, assuming free disposability and
variable returns to scale, is the solution to the linear program:

Choose {Θ, λ} to: min Θ such that: Θw ≥ λʹ W

z ≤ λʹ Z
λi ≥ 0, Σλi = 1, i = 1, 2, ..., n(5.12)

An illustration is seen in Figure 5.15, which depicts production, rather than

cost, frontiers. We consider the efficiency with which several bus companies with
identical vehicle fleets use their labor input to provide peak (X) and off-peak (Y)
services. The black dots represent the combinations of peak and off-peak passen-
ger seat-miles each operator provides. Those enterprises on the ‘old frontier’ are
deemed to be the most efficient users of their labor over the two periods, although
no judgment can be made about whether company M, which offers a relatively
large amount of peak services is preferable to N, which focuses on using its labor
for off-peak services. To make that judgment one needs to know the demands for
the different types of services. The filled stars indicate what may happen to the
frontier over time as technology improves.
DEA can be a useful procedure for establishing relative performance
scores where multiple outputs and inputs are present yet their relative impor-
tance (weights) are not evident. But it has its limitations. While other suppliers
inside the frontier are relatively inefficient, being on the frontier does not imply
maximum efficiency; all the companies may be inefficient but differ in their
degrees of inefficiency. There are methods of measuring the extent to which sup-
pliers fall behind those on the frontier, but these may be sensitive to the ways in
which measurements are made; for example, is the measurement the vertical dis-
tance from the frontier, the horizontal distance, or some average of the two? The
results are also sensitive to outliers. The shape of the frontier can differ consider-
ably if there is, say, one supplier that is well away from the remainder. This can be
seen in Figure 5.15 if instead of producer K in the second period being located

BUTTON 9781786435668 PRINT.indd 181 27/04/2022 11:07

 182 TRANSPORT ECONOMICS, 4TH EDITION

Output Y

N
New frontier

K*
K

Old frontier

0 Output X
Notes: • denotes performance of firms in the first period; ★ denotes performance of firms in the
second period.

Figure 5.15 Production frontier for two outputs, observed at two different time periods

as indicated, it is located at K*. Finally, the technique does not in itself allow for
any assessment of causality: it basically provides a ranking mechanism indicat-
ing levels of cost (or production efficiency), offering insights as to the underlying
causes of the differences between suppliers.
Given the availability of easy-to-use software, there are numerous examples
of applications of DEA in transport, for example Bookbinder and Qu (1993) who
calculated DEA scores for seven large North American railroads and explored the
effects of different numbers of outputs and inputs as well as reducing the number
of firms. Another case study is by Oum and Yu (1994), who analysed OECD rail-
ways for the period from 1978 to 1989, employing two types of output measure
and evaluating and interpreting the relative efficiency scores. Button and Costa
(1999) examined the efficiency of local transport in several European cities, and
Fare et al. (2007) looked at airlines, just to provide a few examples. Some studies
have used both parametric and non-parametric techniques to confirm their find-
ings, as, for example, did Good et al. (1965) when comparing the efficiency of
European and United States airlines.

References

Allen, W.B and Liu, D. (1995) Service quality and motor carrier costs: an empirical analy-
sis, Review of Economics and Statistics, 77, 499–510.
Baker, S.H. and Pratt, J.B. (1989) Experience as a barrier to contestability in airline
markets, Review of Economics and Statistics, 81, 352–6.
Berechman, J. (1983) Costs, economies of scale and factor demand in road transport,
Journal of Transport Economics and Policy, 17, 7–24.

BUTTON 9781786435668 PRINT.indd 182 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 183

Bitzan, J.D. and Keeler, T.E. (2003) Productivity growth and some of its determinants in
the deregulated U.S. railroad industry, Southern Economic Journal, 70, 232–53.
Bookbinder, J.H. and Qu, W.W. (1993) Comparing the performance of major American
railroads, Transportation Research Form, 33, 70–85.
Bowman, A. (2015) An illusion of success: the consequences of British rail privatization,
Accounting Forum, 39, 51–63.
Brown R.H. and Nash, C.A. (1972) Cost savings from one-man operations of buses,
Journal of Transport Economics and Policy, 6, 281–4.
Button, K.J. (2016) Public–private partnerships: a review of economic considerations
with particular reference to transportation projects, Transportation Planning and
Technology, 39, 136–61.
Button, K.J. and Costa, A. (1999) Economic efficiency gains from urban public transport
regulatory reform: two case studies of changes in Europe, Annals of Regional Science,
33, 425–38.
Button, K.J. and O’Donnell, K. (1985) An examination of the cost structures associated
with providing urban bus services in Britain, Scottish Journal of Political Economy, 32,
67–81.
Caves, D.W., Laurits, R., and Tretheway, M.W. (1984) Economies of density versus econo-
mies of scale: why trunk and local airline costs differ, Rand Journal of Economics, 15,
471–89.
Dodgson, J.S. (1984) Railway costs and closures, Journal of Transport Economics and
Policy, 18, 219–36.
Edwards, S.L. and Bayliss, B.T. (1971) Operating costs in road freight transport, Report,
UK Department of the Environment.
European Union (1998) Fair payment for infrastructure use: a phased approach in
a common transport infrastructure charging framework in the EU, White Paper,
Com(96)466.
Fare, R., Grosskopf, S., and Sickles, R.C. (2007) Productivity of US airlines after deregula-
tion, Journal of Transport Economics and Policy, 41, 93–112.
Foster, C. and Golay, J. (1986) Some curious old practices and their relevance to equilib-
rium in bus competition, Journal of Transport Economics and Policy, 20, 191–216.
Foster, C.D. and Joy, S. (1967) Railway track costs in Britain, Report, Development of
Railway Traffic Engineering, Institution of Civil Engineers.
Friedlander, A. (1993) Rail costs and capital adjustment in a quasi-regulated environment,
Journal of Transport Economics and Policy, 24, 9–34.
Gillen, D.W., Oum, T.H., and Tretheway, M.W. (1990) Airline cost structure and policy
implications, Journal of Transport Economics and Policy, 24, 9–34.
Good, H., Röller, R.D., and Sickles, R.C. (1965) Airline efficiency differences between
Europe and the US: implications for the pace of EC integration and domestic regula-
tion, European Journal of Operations Research, 80, 508–18.
Goodwin, P.B. (1974) Generalized time and the problem of equity in transport studies,
Transportation, 3, 1–24.
Hadley, T. (1885) Railroad Transportation: Its History and its Laws, G.P. Putnams.
Harberger, G. (1959) Using the resources at hand more effectively, American Economic
Review, 59, 134–47.
Heaver, T.D. (1975) The routing of Canadian container traffic through Vancouver and
Seattle, Report, WESTMAC.
Hotelling, H. (1929) Stability in competition, Economic Journal, 39, 41–57.
Johnston, A. and Ozment, J. (2013) Economies of scale in the US airline industry,
Transportation Research Part E, 51, 95–108.
Joy, S. (1973) The Train that Ran Away, Ian Allan.
Karlaftis, M. and McCarthy, P. (1999) The effect of privatization on public transit costs,
Journal of Regulatory Economics, 16, 27–44.

BUTTON 9781786435668 PRINT.indd 183 27/04/2022 11:07

 184 TRANSPORT ECONOMICS, 4TH EDITION

Keeler, T.E. (1974) Railroad costs, returns to scale, and excess capacity, Review of
Economics and Statistics, 56, 201–208.
Keeler, T.E. and Small, K.A. (1977) Optimal peak-load pricing investment and service
levels on urban expressways, Journal of Political Economy, 85, 1–25.
Koshal, R.K. (1970) Economies of scale in bus transport II: some Indian experiences,
Journal of Transport Economics and Policy, 4, 29–36.
Lee, N. and Steedman, I.W. (1970) Economies of scale in bus transport I: some British
municipal results, Journal of Transport Economics and Policy, 4, 15–28.
Leibenstein, H. (1966) Allocative efficiency vs ‘X-efficiency’, American Economic Review,
56, 392–415.
McIntosh, P.T. and Quarmby, D.A. (1972) Generalized costs and the estimation of move-
ment costs and benefits in transport planning, Highway Research Record, 383, 11–23.
Meyer, J.R. (1958) Some methodological aspects of statistical costing as illustrated by the
determination of rail passenger rates, American Economic Review, 48, 209–22.
Mohring, H. (1976) Transportation Economics, Ballinger.
Nash, C.A. (1979) The track costs issue: a comment, Journal of Transport Economics and
Policy, 14, 113–16.
Neuberger, N.L.I. (1971) Perceived costs, Environment and Planning, 3, 369–76.
Nythi, M., Hooper, P., and Heusher, D. (1993) Compass Airline; 1 December 1990 to 20
December 1991 – what went wrong? Transport Reviews, 13, 112–49 and 185–206.
Oum, T.H. and Yu, C. (1994) Economic efficiency of railways and implications for public
policy: a comparative study of the OECD countries’ railways, Journal of Transport
Economics and Policy, 28, 121–38.
Poole, E.C. (1962) Costs: A Toll for Railroad Management, Simmons-Bordman.
Singh, K. (2005) Costs, economies of scale, in urban bus transport: an analysis of munici-
pal undertakings in India, International Journal of Transport Economics, 32, 171–94.
Small, K.A., Winston, C., and Evans, C.A. (1989) Road Work: A New Highway Pricing and
Investment Policy, Brookings Institution.
Stopford, M. (2009) Maritime Economics, Routledge.
Tanner, J.C. (1968) An economic comparison of motorways with 2 or 3 lanes in each direc-
tion, Road and Research Laboratory Report LR.203.
Taussig, F.W. (1891) A contribution to the theory of railway rates, Quarterly Journal of
Economics, 5, 438–65.
Teal, R.F., Marks, J.V., and Goodhue, R. (1980) Subsidized shared-ride taxi service,
Transportation Research Record, 778, 25–32.
Tolifari, S., Button, K.J., and Pitfield, D. (1986) Shipping costs and the controversy over
open registry, Journal of Industrial Economics, 34, 409–27.
UK Department of Transport (1980) Report of the inquiry into lorries, people and the
environment, HMSO.
UK Department of Transport (1989) The allocation of road track costs 1989/91,
Department of Transport.
UK Price Commission (1978) The road haulage industry, House of Commons Paper HC
698, HMSO.
Wabe, J.S. and Coles, O.B. (1975) The peak and off-peak demand for bus transport: a cross-
sectional analysis of British municipal operations, Applied Economics, 7, 25–30.
Wallis, I.P. (1980) Private bus operations in urban areas: their economics and role, Traffic
Engineering and Control, 22, 605–10.
Walters, A.A. (1968) The Economics of Road User Charges, Johns Hopkins University
Press.
Warner, S.L. (1962) Stochastic Choice of Mode in Urban Travel: A Study in Binary Choice,
Northwestern University Press.
Williams, M. (1981) The economic justification for local bus transport subsidies,
International Journal of Transport Economics, 8, 79–88.

BUTTON 9781786435668 PRINT.indd 184 27/04/2022 11:07

 DIRECT COSTS OF TRANSPORT ­ 185

Wilson, A.G., Hawkins, A.F., Hills, G.J., and Wagon, D.J. (1969) Calibration and testing
of the SELNEC transport model, Regional Studies, 3(3), 337–50.
Winston, C. (1985) Conceptual developments in the economics of transportation: a Journal
of Transportation interpretive survey, Journal of Economic Literature, 23, 57–94.
Zhang, L., Chen, F., Ma, X., and Pan, X. (2020) Fuel economy in truck platoon-
ing: a literature overview and directions for future research, Journal of Advanced
Transportation, 2020, 2604012.

BUTTON 9781786435668 PRINT.indd 185 27/04/2022 11:07

 6 The External Economic Costs
of Transport

6.1 Introduction

Chapter 5 was concerned with showing the types of financial costs directly con-
fronting transport users and those providing transport services. It is clear from
our everyday experience, however, that there are other costs associated with trans-
port that are not directly borne by those generating them. Some affect those not
traveling, such as when air travelers impose noise costs on those living below air-
craft flight paths, and when road travelers inflict dirt and vibration on those living
adjacent to major trunk routes while, at the same time, impeding the progress of
pedestrians in towns. Maritime transport frequently pollutes bathing beaches
with their oil discharges, and the construction of ports disrupts local breeding
grounds for birds and sea life.
These are often seen as issues associated with mechanized transport, and
especially automobiles and trucks. But this has, albeit on a different scale and in
different forms, been an issue back into antiquity. In Roman times, road traffic was
limited at night in cities because of noise nuisance, in the middle ages and later
river flows were used to ‘transport’ dung and excrement, as well as offal and other
waste products, away from cities. Draft animals were the cars and trucks of the
time, and they left their mark on urban streets and on the ‘fragrances’ in the air.
These are external costs generated by transport users. They are inflicted
on the non-traveling public, often those living nearby, but in the case of global
warming emissions the victims may be thousands of miles away. They are now at
the forefront not only of transportation debates but more generally as societies
have become more concerned with quality of life and the larger issue of climate
change.
But they are not the only form of negative externality. There is also traffic
congestion. A car entering a congested stream of traffic slows other cars already
in the traffic, an effect external to the driver but different from environmental
effects in that the cost is contained within transport – strictly, it is a ‘club bad’.
This chapter focuses initially on the former type of external effect – the
imposition of uncompensated effects created by transport users on the general
public or some part of it – and then moves on to look at the challenges imposed
by various forms of traffic congestion. The latter being about managing the inter-
nal workings of the transport system more effectively by optimizing the external
costs one set of users imposes upon another. But we begin with some definitions
and some clarifications.

BUTTON 9781786435668 PRINT.indd 186 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 187

6.2 Externalities

Formally, in economics externalities exist when the activities of one group (either
consumers or producers) affect the welfare of another group (consumers or pro-
ducers) without any payment or compensation being made. They may be thought
of as relationships other than those between a buyer and a seller, and do not
normally fall within the ‘measuring rod of money’ or within a complete market.
There are also external benefits as well as costs, although these are generally
thought less important in the transport sector. In the context of transport, wide
streets often act as fire-breaks in addition to serving as transport arteries, and
this may be thought of as an external benefit associated with urban motorways.
Likewise, the drainage effects of canals are often an external benefit for agricul-
ture. One reason why external benefits are less common than their negative coun-
terparts is that there is an incentive for those producing them to bring them into
the market and charge for them. A simple example of this may be when railway
enthusiasts (often called ‘foamers’ in the United States and ‘grisers’ in the United
Kingdom) enjoyed standing on railway platforms watching trains – a benefit gen-
erated by the railway company and enjoyed free by the enthusiasts. There are now
often charges, however, for accessing platforms simply to view.
A vast theoretical literature has grown up since the Second World War,
refining the rather complicated concept of external costs. There are, for example,
often confusions between externalities and such things as public goods, which
we return to later. While much of the detail of this work has a greater or lesser
importance in a transport context, there are two major distinctions that need to
be highlighted.

Pecuniary and Technological Externalities

The formal difference between these two categories of externality is that when the
latter effects occur in production (or consumption) they must appear in the pro-
duction (or utility) function, while this is not the case with pecuniary externali-
ties. Pecuniary effects occur when, say, a firm’s costs are affected by price changes
induced by other firms’ actions in buying and selling factors of production. An
example can help to clarify this. A new motorway may block or destroy a pleas-
ant view formerly enjoyed by the residents of an area. The fact that this directly
enters the residents’ utility function means that if no compensation is paid, this is
a technological externality. If this new motorway also takes business away from a
local garage and transfers it to a new motorway service station, then the reduced
income suffered by the garage proprietor is a pecuniary externality since the effect
is indirect, namely through changes in the prices charged by the two undertakings.
The distinction is a fine one – particularly since in practice both forms
of externality usually occur simultaneously – but it is an important one.
Technological externalities are real resource costs that strictly should be taken
into account in decision-making if optimal efficiency is to be ensured. Pecuniary
externalities do not involve resource costs in an aggregate sense, but they do

BUTTON 9781786435668 PRINT.indd 187 27/04/2022 11:07

 188 TRANSPORT ECONOMICS, 4TH EDITION

normally have important distributional implications (in our motorway example,

the service station gains while the local garage loses).
The fact that there may be pecuniary externalities associated with a project
does not reduce the aggregate net benefit, but, rather, reveals that there are adjust-
ments in the economy which influence who is to enjoy the gains and who is to
suffer the costs. The distinction between technological and pecuniary externalities
is, therefore, important in the appraisal of public sector transport investments
where one is concerned with the incidence of the costs and benefits in addition
to their overall level (see Chapter 11). The pecuniary external effects can, for
example, raise matters of environmental justice (Haynes, 2004).

The Distinction Between Pollution and Congestion

Conventional welfare economics distinguishes between varieties of externality

categories according to the different types of agents involved. Jerome Rothenberg
(1970) offers a simple dichotomy that is possibly of more use in the transport
context than some of the more complicated categorizations that have emerged
subsequently. He distinguishes between two forms of what he calls ‘generic con-
gestion’. The underlying idea is that externalities result from attempts by different
agents to share a common service that is not provided in discrete units earmarked
for each (in a technical sense, it has ‘public good’ characteristics). The presence of
other users already affects the quality of service that is rendered to each. Generic
congestion may, in technical terms, be divided into:

• Pure pollution. ‘The essence of pollution … is that there are some other users
who do abuse the medium – the polluters – while others are relatively passive
victims of such abuse – the public. Jet planes make the noise, housewives are
forced to submit to it’.
• Pure congestion. ‘If highway traffic is the classic example of congestion,
then the central inter-personal distributive fact about it is that all users are
using the medium (the public good) in much the same way, each is damag-
ing service quality for both others and himself, and the ratio of self/other
damage is approximately the same for all users … The whole user group loses
homogeneously by their self-imposed interaction’.

Another way of looking at externalities, and one developed by Alain

Bonnafus (1994), is according to the domain they affect. The alternative typology
is illustrated in Figure 6.1. Here we see a succession of overlapping spheres that
address the question of ‘external to what’? The innermost sphere is essentially
the firm (which may be any producer of transport services, say a bus company or
a shipping line, but may also be treated as a driver of a car supplying transport
services to him or herself). The agent is the perpetrator of the externalities.
The second sphere reflects what are often called ‘Marshallian externalities’
and captures what we have termed pure congestion: a truck entering a congested
road and slowing down others further. The third sphere reflects what Alain

BUTTON 9781786435668 PRINT.indd 188 27/04/2022 11:07

 189
THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­

Biosphere

Personal satisfaction

Market or social good

Market sphere

Firm or consumer

Marshallian external economies

and diseconomies
External costs funded by the
community

Inter-personal external effects

External effects on the

environment

Figure 6.1 The economic relationships between external effects

Bonnafus sees as the external costs of a transport company paid for by the com-
munity. These are subsidies for operations and infrastructure maintenance not
captured in user fees. The penultimate sphere reflects inter-personal external
effects, such as lack of safety or noise. Finally, the outermost sphere embraces the
quality of the larger, potentially global, environment that is affected by the emis-
sions of the truck.
From an economic perspective, the advantage of this type of approach is
that it provides an indication of where the market failings are in the broader
scheme of things, and where the incidence of the external costs are being borne.
The movement from conceptualization to practical valuation and analysis of
alternative ways of dealing with the various forms of externality is, however, made
no simpler. We do not, therefore, attempt to look at all the domains illustrated,
but focus on those particularly relevant for transport, namely those involving

BUTTON 9781786435668 PRINT.indd 189 27/04/2022 11:07

 190 TRANSPORT ECONOMICS, 4TH EDITION

pure pollution and pure congestion. We have already dealt with Marshallian
externalities in Chapter 5.

6.3 Transport’s Implications for the Environment

Transport pollutes the physical environment in several ways that can also be seen
in a stylized manner in relation to their broad geographical and temporal inci-
dence in Figure 6.2.

• It imposes many local environmental costs on those living, working, or taking

recreation near major pieces of transport infrastructure. These include such
factors as noise, visual intrusion, local air pollution (for example, particu-
lates, lead, and carbon monoxide), and the disposal of obsolete vehicles. A
major problem here is that, unlike many other forms of environmental intru-
sion, it is generally difficult to move transport facilities away from sensitive
areas simply because users demand easy access and proximity to roads and
to public transport terminals.
• There are trans-boundary effects such as emissions that contribute to acid
rain (such as nitrogen oxides (NOX)) and maritime spillages that have
impacts some distance from the transport activities themselves. This poses
challenges of ­inter-jurisdictional authority at the national level, and often
within federal systems at the state level.
• There is the contribution of transport to the global environmental problems
of global warming – such as emissions of carbon dioxide (CO2) – and to
upper-level ozone depletion – notably, chlorofluorocarbons (CFCs). As we
see later in Chapter 8, these pose major challenges because of their long-term

Time

CO2

Aromatics
Chlorofluorocarbons
(CFCs)
NOX

Water contamination

Lead
Noise
0 Impact radius
Figure 6.2 The time and spatial coverage of exhaust gases and other environmental
intrusions

BUTTON 9781786435668 PRINT.indd 190 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 191

effects and the need for coordinated, global approaches to handle them.
Strictly, these CO2 emissions are not pollution, CO2 being a trace gas that
appears in the atmosphere naturally with sources in groundwater, rivers and
lakes, ice caps, glaciers, and seawater.

A very general indication of the contribution of transport to some of the

atmospheric problems associated with light mechanized transport in the United
States is offered in Table 6.1. Table 6.2 offers a similar breakdown for the United
Kingdom, although covering all transport modes in this case. It is noticeable that
many of the pollutants are becoming less of a problem per mile traveled, but of
course, transport use is growing and partly off-sets this.
Many environmentalist groups argue for substantial reductions or total
elimination of these adverse environmental effects, but this ignores the cost
associated with removing such nuisances. While some people suffer from the
environmental intrusion associated with transport, others clearly benefit from
being able to travel more freely or move goods more cheaply. In almost all cases
environmental improvements would reduce the net benefits enjoyed by transport

Table 6.1 
United States light gasoline and diesel vehicle exhaust emissions rates (grams
per mile)

2000 2005 2010 2015 2018

Light-duty gasoline vehicles

  Total HC 1.617 1.020 0.786 0.499 0.350
  Exhaust CO 16.028 9.759 7.121 4.898 3.941
  Exhaust NOX 1.681 1.079 0.901 0.518 0.289
  Exhaust PM2.5 0.037 0.023 0.017 0.011 0.008
Light-duty diesel vehicles
  Total HC 2.919 1.915 0.939 0.232 0.183
  Exhaust CO 47.204 28.016 13.604 3.205 2.663
  Exhaust NOX 2.906 1.691 1.008 0.248 0.153
  Exhaust PM2.5 0.066 0.052 0.023 0.005 0.004

Source: US Bureau of Transportation Statistics.

Table 6.2 Pollutants emissions from transport in the United Kingdom (thousand tonnes)

1995 2000 2005 Percent of national

emissions in 2005

CO 4,180 2,500 1,124 46.5

NOX 1,098 818 549 33.8
PM10 53.6 38.6 33.7 22.5
Benzene 28.4 5.6 3.4 20.8
1,3 butadiene 7.4 3.8 1.6 60.8
Pb 1.049 2.2 2.1 1.7
SO2 51.8 6.7 3.0 0.4

Source: UK Department for Transport.

BUTTON 9781786435668 PRINT.indd 191 27/04/2022 11:07

 192 TRANSPORT ECONOMICS, 4TH EDITION

users. Economists tend, therefore, to think in terms of optimizing the level of pol-
lution rather than ‘purifying’ the environment entirely.
If we look at Figure 6.3, we see plotted on the vertical axis the money value
of the costs and benefits of reducing the noxious fumes emitted by motorcars
and, on the horizontal, the environmental improvements that accompany a
reduction in such fumes. The marginal costs (MC) of reducing the emissions are
likely to rise quite steeply. While more sophisticated filters may be fitted and fuel
subjected to more extensive refining, both become increasingly costly to apply as
the toxicity of the exhaust is reduced. Additionally, they reduce the efficiency of
vehicles and may, in the case of improved refining, impose higher levels of pollu-
tion on those living around refineries.
The marginal benefits (MB) of ‘cleaner’ road vehicles, in contrast, are likely
to fall with successive improvements. The public is likely to be relatively less con-
scious of lower levels of emission and be aware that many of the seriously toxic
materials (for example, lead) are likely to be amongst the first to be removed in the
clean-up program. Consequently, the marginal cost and revenue curves associated
with improved emission quality are likely to be of the form seen in Figure 6.3.
There is quite clearly an optimal level of improvement (that is, 0E1), beyond
which the marginal costs of further emission reductions exceed the marginal ben-
efits. If the clean-up program reduced emissions to the point where further reduc-
tions would yield no additional benefit (that is, exhaust fumes would be considered
‘pure’, although this may not mean zero toxicity if individuals’ perceptions are
faulty), then the situation is not optimal. Improvements beyond 0E1 to 0E2, in fact,
result in a net welfare loss equal to the shaded area ABE2 in the diagram.
Consequently, when talking about the excessive environmental harm caused
by various forms of transport it is important to remember that this is an excess
above the optimal level of pollution, not above zero pollution or some per-
ceived ‘pure’ environment. We return to this topic and methods of attaining the
optimum in Chapter 8.

MB
MC
B

0 E1 E2
Environmental improvement

Figure 6.3 The optimal environmental improvement

BUTTON 9781786435668 PRINT.indd 192 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 193

6.4 The Valuation of Externalities

Physical measures of environmental damage are helpful, and a necessary input

into analysing their economic implications, but the diversity of measurements
used and impacts generated limit their usefulness. To compare the external costs
and benefits of transport with other features of transport it is often found useful
to convert such costs and benefits into monetary terms. This is no easy task but
economists have developed several procedures that, at least in the case of some
externalities, do provide reasonable guidance to the value of these external effects.
In recent years the level of sophistication used in this process has risen consider-
ably and only a very brief outline of some – the more common techniques – is set
out below.
The various stages in valuation are seen in Figure 6.4. As can be seen, it is a
sequential process embracing many disciplines beside economics; indeed, without
good physical measurement and scientific understanding of the consequences
of environmental intrusion, the economic analysis would not be possible. The
immediate focus will be on the various techniques that go into ‘monetization’ of
the various externalities, and in the following section we shall consider some of
the empirical work that has been conducted.

Precedents

Consistency over time is the prime reason for suggesting that historical prec-
edents could be used as a means of valuing certain aspects of the environment.

Activity

Emissions

Transport &
chemical conversion

Concentration/
deposition

Response of
receptors (humans, Changes in utility
flora, materials,
ecosystem) Welfare loss

Physical impact
Monetization

Figure 6.4 The stages of putting a money value on an externality

BUTTON 9781786435668 PRINT.indd 193 27/04/2022 11:07

 194 TRANSPORT ECONOMICS, 4TH EDITION

Precedents in this context largely comprise legal, judicial rulings on compensation

for inflicting environmental damage. While having some superficial attractions,
the procedure, however, has severe limitations.
The main applications of this approach in transport have been in terms of
valuing injury and death in transport accidents although there are numerous
instances of transport suppliers, and especially shipping companies, having to
compensate for spillage of toxic pollutants and there are often rulings on com-
pensation on such things as the additional noise nuisance associated with an
airport expansion.
Precedents only exist where there are established rights and these extend to
very few environmental attributes. Even without this practical limitation, the use-
fulness of the techniques is restricted by the nature of most legal systems. They
normally apply to the need for victims (including relatives of people killed) of the
incident to be cared for during the remainder of their lives. Consequently, where
the environmental damage causes death, the ‘cost’ to the deceased is not consid-
ered. Equally, damage to flora and fauna is generally outside of the scope of legal
rulings on compensation. Finally, where evidence has been produced looking at
legal precedence, it tends to show little by way of a consistent pattern.

Averting Behavior

Many adverse environmental consequences of transport can be ameliorated

by insulation. Noise nuisance can be reduced by double-glazing windows, the
adverse effects of air pollution by installing air conditioning, and accident risk by
the adoption of safer engineering design standards for transport infrastructure
(for example, relating to road surfaces and air traffic control radar), the ways in
which it may be used (for example, speed limits), and the design of the vehicles
that use it (for example, air bags and seat belts in cars). A widely deployed tech-
nique for assessing the costs of environmental damage is to equate them with the
cost of avoidance, often called the ‘prevention expenditure’ method.
The main problem is the difficulty of isolating specific expenditures made
for environmental reasons from the implicit joint expenditure on other benefits
accompanying, for example, double-glazing (such as reduced heating bills, etc.)
or air conditioning (for example, a cooler room temperature). Noise insulation
is also only partial in that it does not offer protection when in the garden or
when windows are open. More fundamentally, there are questions about the opti-
mality of the level of avoidance adopted. In terms of safety, for example, the avia-
tion industry provides an extremely safe product but only at a tremendous cost.
In terms of potential lives saved, each is implicitly valued more highly than, say,
a life saved on the roads where the per capita safety expenditure is much lower.
This type of approach is the mirror image of looking at the environmental costs
of lost production or the damage done in repairing the problems caused by envi-
ronmental damage – for example, adding lime to water polluted by ­traffic-related
NOX emissions. The costs of pollution on health can be related to the days of work
lost by, for example, more frequent asthma attacks due to fuel additives or from

BUTTON 9781786435668 PRINT.indd 194 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 195

lack of sleep due to aircraft noise. Isolating these costs can be difficult; there is a
need to make a full life assessment of their impacts on individuals or production.

Revealed Preference: Hedonic Prices

There are circumstances where consumers of environmental resources, through

their actions, implicitly reveal the values that they place on them. They make
trade-offs involving sacrificing some monetary benefits to limit the use of envi-
ronmental resources, or to gain some environmental benefit. The classic case is the
willingness of people to pay to live away from noisy airports or roads, or to pay
a premium for a hotel room away from a busy street. But these choices, as we saw
earlier with the discussion on generalized costs, also extend to travel time values
that involve trade-offs between such things as external and internal attributes of
modes and travel speeds.
The underlying theory can be discussed in terms of Figure 6.5, which plots
the welfare enjoyed by an individual at various levels of wealth. The diminishing
marginal utility of money gives, for example, the trade-off curve I for an indi-
vidual living in a quiet, rural setting. The construction of an airport adjacent to
the house imposes measurable noise costs on this person and, for every level of
wealth, this pulls the trade-off curve down, that is, it becomes II. If the person
was initially at a point A on I then the imposition of the noise will reduce welfare
to level B. To get them back to their original welfare level, compensation of BC
would be needed, this being sufficient to move the person around II until the
original level of welfare was restored.
This approach assumes, however, that there is a finite level of compensation
that satisfies the individual. If, however, one starts at an initial position A*, then
it is not altogether clear that this is so. This type of problem, for example, arose

Welfare

I
A*

A
II
C
B

0 Wealth

Figure 6.5 The basic trade-off model

BUTTON 9781786435668 PRINT.indd 195 27/04/2022 11:07

 196 TRANSPORT ECONOMICS, 4TH EDITION

as early as the late 1960s when the Roskill Commission in the United Kingdom
was assessing sites for a Third London Airport and were trying to value the noise
costs of aviation effects; eventually an arbitrary value was adopted for some indi-
viduals. Second, the onus of the technique as described above is on compensation.
One would normally get a different value by taking the amount adversely affected
individuals would be required to pay to bribe the authorities not to construct the
airport (that is, the amount necessary to get back to the higher trade-off curve for
the new level of welfare B).
In practical terms, revealed preference techniques normally require sophis-
ticated econometric analysis. This is because most goods involve a variety of
attributes of which environmental elements represent only a subset. In conse-
quence, the normal approach is to use a hedonic price index that puts values on
the diverse attributes of the good being examined (for example, the various fea-
tures of houses in the noise case mentioned above). In general terms this means
estimating:

Ch = F(c1, c2, c3, … , cn) (6.1)

where Ch is the consumption of housing services c1, c2, c3, … , cn.

The specifications of individual models differ, indeed one of the problems
with hedonic indices is that of model specification; but whatever form they take,
they seek to isolate the ‘price’ of each characteristic in the equation.

Exhibit   The economic costs of CO2 emissions

The social cost of carbon (SCC) is calculated using integrated assessment models (IAMs) of
climate and the economy. Because of the longevity of many pollutants and the duration of
their effects, the latter estimate the global damage resulting from greenhouse gas emissions
over a century or more. They seek to capture the complex interplay between climate and
the economy. Future climate damage is discounted to a current SCC value using a social
discount rate.
Three widely used estimation model frameworks used to estimate the SCC of CO2 are:
• The dynamic integrated climate change (DICE) model. This is essentially a neoclassical
global economic growth model with CO2 emissions treated as function of global GDP
and the carbon intensity of economic activities, with the latter declining over time with
technical progress. The damage function is calibrated on the sum of the damage to
agriculture, coastal areas, energy use, health risk, recreation, cities, and the ecosystem. It
assumes that climate damage increases more than linearly with temperature increases.
• The framework for uncertainty, negotiation, and distribution (FUND) model is a
nine-region model of the world economy and its interactions with climate, running
in time steps of one year from 1990 to 2200. It consists of scenarios for economy
and population, which are perturbed by climate change and greenhouse gas emission

BUTTON 9781786435668 PRINT.indd 196 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 197

reduction policy. Policy variables are energy and carbon efficiency improvement, and
sequestering CO2 in forests.
• The policy analysis of the greenhouse effect (PAGE) model uses simple equations to
simulate the results from more complex specialized scientific and economic models.
The damage caused by temperature increases includes economic and non-economic
categories and are considered over eight regions. The model assesses lost output from
global temperature changes.

Model Study SCC per ton CO2 SCC per ton CO2 according
according to the study to the US Department of
in the 2nd column Energy ($ in 2010)

DICE Nordhaus (2008) $6a $28

FUND Anthoff et al. (2011) $8b $6
PAGE Stern (2007) $8a $30
Hope (2006) $5a –
Hope (2013) $106b –

Notes: a.Value in 2000. b.Value in 2010.

Source: Van den Bergh and Botzen (2015). This paper contains the full references to the studies cited.

Studies using these methods have produced the results seen in the table. A higher SCC per
ton indicates a greater gain from reducing carbon emissions at the margin. There is diversity
between results even using the same method. The final column provides results from a
US Department of Energy comparative study of the three models, using five emissions and
socioeconomic scenarios, and assuming the same discount rate value of 3 percent. Each
element in the column is average across the outcomes of the five scenarios.
Concerns have been expressed over the wide range of values of the SCC found not only
across the three broad modeling frameworks, but also between studies using the same
approach. Aside from differences in the technical modeling, there are also differences in the
variables included and the discounts that are used. The functions used in the work do not
handle ‘tipping points’ or extreme changes in climate well, nor do they allow for individual
risk aversion to climate risk.
See also: J.C.J.M. van den Bergh and W.J.W. Botzen (2015) Monetary valuation of the social
cost of CO2 emissions: a critical survey, Ecological Economics, 114, 33–46.

This leads on to a further problem. It is necessary to have a substantial amount

of information on the determinants of, in our example, housing selection pro-
cesses just to gain an insight into the value of one environmental influence. Many
hedonic studies employ a wide variety of variables. It is also important that the
characteristics used are the ones perceived to be important to house occupiers
and buyers: the perceived housing market attributes. It is not the actual set of
characteristics that determine hedonic prices, but rather the characteristics as

BUTTON 9781786435668 PRINT.indd 197 27/04/2022 11:07

 198 TRANSPORT ECONOMICS, 4TH EDITION

seen by those active in the housing market. This often leads to the implicit, but
seldom accurate, assumption that all those in the market have complete informa-
tion of all housing attributes. The assumption is also that all such attributes are
sufficiently tangible and measurable to be incorporated into the econometric
estimation.

Travel Cost Method

The travel cost method is a particular form of revealed preference analysis. New
transport infrastructure can destroy recreation sites such as parks and fishing
facilities that have been provided at a zero price. People, however, travel to such
locations to make use of the natural amenities and thus incur a measurable travel
cost both in terms of time and money. Use can be made of this information to
gain some idea of the value of such facilities. This is a special case of the more
general revealed preference approach (Smith, 1989).
Figure 6.6 offers guidance to the simplest travel cost approach. Surveys find
that the number of visits to, say, a park from an origin A amounts to Xa, and
from B amounts to Xb. Further, the actual average generalized travel costs (that is,
including travel time costs) for these trips amount to Pa and Pb respectively from
the two origins. A succession of further surveys looking at other origins enables
the distance decay function to be derived. From this, the consumer surplus
derived from visiting the park and enjoyed by an individual living at A is seen
to be area (A + B). Total surplus for those originating from A is then found by
multiplying (A + B) by the number of trip-makers originating from there. Similar
calculations can be carried out for each origin to get the aggregate surplus.

Exhibit   The economic costs of the Amoco Cadiz oil spill

When the supertanker Amoco Cadiz ran aground 200 miles off the coast of Brittany in 1978
it resulted in the largest oil spill of its kind – 223 thousand tonnes – up until that date.
Subsequent legal actions resulted in economists trying to place a value on the resultant
environmental damage. This was a challenge, given the lack of data and the sensitivity
of their results to the assumptions made, for example, did the authorities use the best
knowledge available in their reactions?
Two types of case were heard, one initiated by the French government concerning the
damage and clean-up costs of the incident, and the others by local authorities, individuals,
and businesses adversely affected by the spillage. Five categories of cost are considered
under two headings:
Costs of remedial actions
• Direct costs of clean-up
• Costs of restoration

BUTTON 9781786435668 PRINT.indd 198 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 199

Other physical effects

• Costs of lost tourism
• Costs of damage to amenities
• Damage to marine resources
Marginal effective analysis was used for evaluating the clean-up and restoration costs.
Pumping of the oil, for example, was initially focused on the largest and most accessible
deposits with efficiency per worker falling as the area became less polluted. Optimal
pumping is when the marginal social benefit of doing so was equated with the marginal
clean-up cost.
The question in defining the marginal cost of clean-up and of restoration by the French
military was one of whether it should be valued as if a third party were brought in to do
the work or as the additional cost of temporarily taking naval manpower from other duties.
Essentially, this is a matter of whether the cost curve is long-run embracing the costs of
having a reserve available to undertake these actions or a short-run cost curve of diverting
existing manpower. Put another way, is the aim to calculate the option price that society is
willing to pay to avoid an irreversible change in marine resources in the face of considerable
uncertainty about the likelihood of a disaster, its implications, and the costs of dealing with
it? The outcome was based on the former, and clean-up costs of between $103 and $114
million were awarded.
A different methodology was used for evaluating the second group of costs, the loss of
tourism revenue for both the polluted and unpolluted areas affected by an adverse halo
effect. Tourism effects were diverse and exact costs difficult to define. Indicators such as
levels of flour deliveries were used to calculate the decline in the number of tourists. It was
estimated that the spill effects lasted until 1979, equal to 2.9 million tourist days. Wage and
profit rates were then applied to estimate the costs.
Amenities losses, being non-market effects, were valued using both contingent methods
and travel cost approaches, although they were considered limited proxies in the court. In
terms of implications for marine life, this was measured in lost fishing time and the damage
done to boats and equipment (which was estimated by surveying fishermen). Lost income
to the fishermen used estimates based on the market price of the types of fish caught
and estimated lost catch. Similar methodologies were used to assess the costs to oyster
farmers.
See also: F. Bonnieux and P. Rainelle (1993) Learning from the Amoco Cadiz oil spill:
damage valuation and court’s ruling, Industrial and Environmental Crisis Quarterly, 7, 169–88.

The main application of the technique is in evaluating specific types of environ-

mental impact, but is of less use where there are several environmental factors
involved and one wishes to evaluate them individually. Perhaps a bigger problem,
however, is the need to specify the generalized cost function which itself should

BUTTON 9781786435668 PRINT.indd 199 27/04/2022 11:07

 200 TRANSPORT ECONOMICS, 4TH EDITION

Travel cost

B Distance decay function

Pb

Pa

0 Xb Xa Trips

Figure 6.6 The basic travel cost method

include a monetary value of travel time. While, as we have seen, the work on the
value of travel time is extensive, the subject is itself at least as controversial as that
in the field of environmental evaluation.

Stated Preference

Stated preference techniques (often called ‘contingency valuations’ in the envi-

ronmental literature), which have their basis in experimental economics (Smith,
1982), do not involve attempting to place values on environmental costs by
observing actual trade-offs, but, rather, seek to elicit information on the trade-offs
individuals would make when confronted with particular situations. It provides
individuals with a mental experiment, a pseudo-market that involves decision-
making within a controlled environment as far as the information they have and
the options open to them.
The most widespread approach is asking, through questionnaires and
surveys, or interactive computer games, a relevant group of individuals either
what compensation they would need to keep them at their current level of welfare
if some pre-defined transport-induced environmental degradation took place, or,
alternatively, what amount would they be willing to pay to prevent this occurrence
(Spash, 2008). The questions are set in an institutional context (for example, to
make it clear what methods of finance are involved), and, to provide a market
framework, the questioner initiates the process by suggesting an opening ‘bid’ to
which the respondent reacts. The questions need to be couched carefully to ensure
that the hypothetical trade-offs are clear and that the potential problems with the
techniques are minimized.
Inevitably questions arise regarding the extent the information gained
through a stated preference approaches is that which would emerge if an actual

BUTTON 9781786435668 PRINT.indd 200 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 201

market existed. Strict comparisons are not possible since, by definition, no actual
market exists in the stated preference context, but comparisons with other tech-
niques can, at least, give some indication of consistency. In a number of case
studies (which were not transport-specific), where stated preference techniques
were deployed alongside other methods of evaluation, Pearce and Turner (1990)
found that there was a ‘reassuring’ degree of overlap in the findings reported.
Differences do, however, still exist and it is difficult to decide whether this reflects
variations in the quality of individual studies or is a reflection of the usefulness
of differing techniques.
One of the main problems in using the above set of procedures is that they
do not all have the same theoretical underpinnings and this makes comparisons
of results difficult. Is it valid, for example, to compare a value for noise pollution
derived from an averting study with a value of air pollution derived from a stated
preference study? It may also be argued that stated preference is more likely to
yield meaningful results for local environmental effects such as traffic noise and
road accidents because these affected people are more generally familiar with, and
are thus capable of making, trade-offs in experimental situations.

6.5 The Magnitude of the Environmental Externality Problem

The discussion so far has provided a general account about the nature of the
external environmental cost of transport and an account of some ways in which
these may be usefully expressed for economic analysis. For example, Figure 6.3
presented hypothetical marginal cost and benefit curves associated with reducing
motor vehicle exhaust emissions, but to make good practical use of these con-
cepts, as we have emphasized, it is first necessary to measure physically the levels
of pollution and then to put a monetary value on the units of pollution generated.
Here we look at the measurement problem and consider some of the ways
in which pollution has been evaluated in practice and the results obtained. There
is also an assessment of the economic importance of various forms of transport-
associated environmental effects. These topics embrace many complex issues and
have been subjected to major research efforts.

Noise

The noise caused by traffic is associated with many things and is increased by
heavier traffic volumes, higher speeds, and greater numbers of trucks. Vehicle noise
is a combination of the noises produced by the engine, exhaust, and tires. Defective
mufflers or other faulty equipment on vehicles can also increase the loudness of
traffic noise. From an economic perspective, it is often important to have engineer-
ing data on the courses of noise to assess optimal policies for remediation.
Noise associated with transport is a major problem. Early analysis includes
a survey conducted by Market and Opinion Research in 1972 which found that
12 percent of respondents thought that excessive noise was one of the three or

BUTTON 9781786435668 PRINT.indd 201 27/04/2022 11:07

 202 TRANSPORT ECONOMICS, 4TH EDITION

four most serious problems in Britain. In pre-unified Germany, Frenking (1988)

found 65 percent of the population was adversely affected by road traffic noise,
with 25 percent seriously affected; by way of comparison, this represented twice
the problem of noise from neighbors and three times that from industrial noise. It
is an especial nuisance in urban areas, in towns that suffer from a lot of through
traffic (for example, located astride major trunk arteries, such as rail lines,
motorways, etc.) and at locations around transport terminals, such as airports,
bus stations, and car parks. It should also be remembered that noise is not only
generated by traffic, but extremely high levels of noise are also often associated
with the construction of transport infrastructure – up to levels of 110 dB when
piles are being driven.
It has been estimated that over 110 million people in the industrial world
are exposed to road traffic noise levels of more than 65 dB(A), a level considered
unacceptable in OECD countries. While consistent data is somewhat sparse,
there is also ample evidence that there are, in large part because of the nature
of national land-use patterns, but also because of differing national legal struc-
tures, quite considerable differences between countries in terms of the popula-
tions affected by transport-related noise. Equally it is difficult, because of data
limitations, to discern exact trends in population exposure to high noise levels.
International comparisons provide tentative evidence of a decline in numbers suf-
fering from serious noise problems – that is, over 65 dB(A) – in some countries but
a rise in others, however there does seem to be a pattern of significantly increasing
numbers of people falling into the gray area of intermediate noise nuisance of
between 55 and 65 dB(65) (OECD, 1991).
Noise has several different effects on health and well-being. It affects activi-
ties such as communication (speaking, listening to radio and television) and sleep.
These effects further induce psychological and physiological disorders such as
stress, tiredness, and sleep disturbance. Noise can also contribute to cardiovascu-
lar disease, and, at high and prolonged exposure, hearing loss.
In practical terms there are, however, problems in measuring noise nuisance.
First, noise nuisance depends upon both the intensity and the frequency of the
noise. The ‘A’ weighted decibel scale (dB(A)) attempts to allow for this by offer-
ing a measure based on a weighted avenge of decibel readings where the weights
reflect the level of unpleasantness caused by different frequencies and the decibels
reflect the actual intensity of the noise. (While this measure is used in most trans-
port-related work, a slightly different set of weights is employed in the perceived
noise decibel (PNdB) used in the measurement of aircraft noise.) The dB(A) scale
is logarithmic and Table 6.3 gives some examples of dB(A) measured peak noise
of different forms of transport relative to other sources of noise.
The dB(A) scale is sometimes prefixed by a term such as L10 which means
that it relates to a specific proportion of time (that is, L10 refers to the 10 percent
peak noise level). On some occasions, decibel measures have been combined
with other indicators of noise annoyance in a composite index. The Noise
and Numbers Index (NNI) developed for the economic appraisal of the Third
London Airport, for example, combined the average peak level of noise at an

BUTTON 9781786435668 PRINT.indd 202 27/04/2022 11:07

 203
THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­

Table 6.3 Relative noise of different activities

Perception example Sound level dB(A)

Extreme jet take-off at 100 meters 120

Pop group 110
Loud car horn 100
Very loud heavy traffic 90
Noisy office 80
Loud busy street 70
Average office 60
Noisy normal conversation 50
Moderate quiet office 40
Quiet conversation 30
Quiet room 20
Very faint normal breathing 10
Threshold of hearing 0

airport (measured in PNdBs) with an indicator of the daily number of aircraft

heard. The logarithmic nature of the NNI means that a one-unit increase in the
index represents a greater increase in noise nuisance, the higher the existing level
of the index.
A scale against which noise nuisances may be measured does not offer an
economist trying to optimize noise emitted by transport much assistance in being
able to place a monetary value on the noise so that the opportunity costs of dif-
ferent policies may be assessed. There are several ways in which noise has been
evaluated, although revealed preference studies dominate in number.
In the 1960s, the Commission on the Third London Airport, in its pioneer-
ing study, considered changes in property values with higher noise levels and in
so doing established an adoption of hedonic indices as an evaluation method
(see also Chapter 11). Surveys were conducted at the existing Heathrow and
Gatwick airports seeking the actual and predicted sale prices of properties at
different distances from them. While the Third London Airport method is useful
it does present some difficulties. House prices vary for many different reasons
and not simply because of the noise levels inflicted upon them. House prices
around Gatwick, for example, tended for a variety of reasons to be higher than
Heathrow, which explains the greater fall in house values with respect to noise
levels in the former. This does not invalidate the house valuation technique
but it does suggest that values obtained by employing it should be used with
circumspection and, more specifically, that a value for noise nuisance derived
using it in one area may be inappropriate for transport studies elsewhere without
adjustment.
A typical hedonic index of the type now used, albeit probably with fewer
variables, is provided by Jon Nelson (1979):

lnPV = 1.54 + 0.03(lnX1) + 0.20(lnX2) + 1.35(lnX3) + 0.02(lnX4)

          + 0.07(lnX5) – 0.33(lnX6) – 0.01(lnX7) (6.2)

BUTTON 9781786435668 PRINT.indd 203 27/04/2022 11:07

 204 TRANSPORT ECONOMICS, 4TH EDITION

where:

PV is a measure of property value differences;

X1 is a measure of the age of the house;
X2 reflects whether there is air conditioning;
X3 is the number of rooms;
X4 is the plot size;
X5 reflects whether the house has a riverside location;
X6 is time to reach employment; and
X7 is the ‘noise exposure forecast’ (NEF).

Examination of the NEF coefficient, the models calculated on data from around
six major airports in the United States, thus suggests that a one-unit increase in
NEF results in a 1 percent depreciation in a house’s value. Table 6.4 provides some
examples of the findings of such studies regarding airport noise.
Subsequently, Nelson (2004) applied meta-analysis to the negative relation-
ship between airport noise exposure and residential property values. The analysis
focuses on 20 hedonic studies covering 33 estimates of 23 airports in Canada and
the United States that have looked at the percentage depreciation in property
values per decibel increase in airport noise, or the noise discount. The weighted-
mean noise discount across the studies is 0.58 percent per decibel, with country
and model specifications having some effect on the measured noise discount;
the cumulative noise discount in the United States is about 0.5 to 0.6 percent

Table 6.4 
Early estimates of noise nuisance on property values (percentage change per
decibel increase)

Study Year % in house price Country

Abelson 1979 0.45 Australia

Collins & Evans 1994 0.45 United Kingdom
De Vany 1976 0.80 United States
Dygert 1973 0.60 United States
Emerson 1969 0.57 United States
Gautin 1975 0.35 United Kingdom
Levesqus 1994 1.30 Canada
Maser 1977 0.62 United States
McMillan 1978 0.50 Canada
McMillan 1980 0.87 Canada
Mieszkowski 1978 0.40 Canada
Nelson 1979 1.10 United States
O’Byrne et al. 1985 0.52 United States
O’Bryne et al. 1985 0.57 United States
Paik 1972 0.65 United States
Pennington et al. 1990 0.60 United Kingdom
Price 1974 0.83 United States
Uyeno et al. 1993 1.13 Canada

Source: Johnson and Button (1997). This paper contains the full references to the studies cited.

BUTTON 9781786435668 PRINT.indd 204 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 205

per decibel at noise exposure levels of 75 dB or less, while in Canada it is 0.8 to

0.9 percent per decibel. These are findings in line with the literature review con-
ducted by David Gillen (2003), who found that aviation noise seems to reduce
the prices of similar houses by about 0.45 to 0.9 percent for each decibel.

Atmospheric Pollution

Transport is a source of many harmful gases, and is one of the major contribu-
tors of several atmospheric pollutants. It is worth emphasizing, however, that
while in some respects the environmental damage done by transport is increasing,
in others, and regarding local pollution, there are reductions in many countries.
Visual manifestations of this can be seen in the removal of the smog that used to
hang over cities such as Los Angeles.
It is also worth remembering that exhaust fumes have a time and a spatial
coverage. There is a time gap as the impacts move from one level to another.
Figure 6.2 offered a broad picture of what happens. At the higher levels, the
original impacts are connected to many other effects and systems, which
are not exclusively related to transport. A potent cocktail of transport- and
­non-transport-related emissions, therefore, often exists. For ease of exposition,
however, we deal with each of the main pollutants separately. The discussions
below provide some indication of both these long- and short-term implications,
as well as the nature of the spatial coverage, when this is ­particularly relevant.

Fuel additive emissions

To enhance engine performance, additives are added to fuels. While some are rela-
tively benign in their environmental effects, others have caused increasing concern
over time. The organic lead compounds added to gasoline as an anti-knock agent,
especially when used by automobiles in confined urban spaces, have been singled
out for particular attention. Lead is a metallic element that can be retained in the
body in the forms of its compounds and can have an adverse effect on the mental
development of children and affect the kidneys, liver, and reproductive system.
In industrialized nations, transport is the single largest source of lead emis-
sions with some 50 percent of lead associated with transport, but the figure
can approach 100 percent in confined urban spaces. However, the tightening of
laws concerning maximum lead content in gasoline (for example, in the United
Kingdom’s case, from 0.84 grams per liter to 0.40 per liter in 1981 and a further
reduction to 0.15 grams per liter in 1985) and the fostering of increasing use
of lead-free gasoline through fiscal and other measures which we discuss further
in Chapter 8, has caused major changes in most major industrialized countries.

Particulate matters
These embrace fine solids or liquid particles found in the air or in emissions such
as dust, smoke, or smog. They are normally neither smelt nor tasted, nor seen
by the naked eye, but can penetrate the body through the lungs and cause major
health problems. Sources include the fine asbestos and other particles stemming

BUTTON 9781786435668 PRINT.indd 205 27/04/2022 11:07

 206 TRANSPORT ECONOMICS, 4TH EDITION

from wear and tear of tires and brakes as well as matter resulting from engine, and
especially diesel engine, combustion.
Transport is the major source of particulate emissions in many industrialized
countries including the United Kingdom and the United States. Particulate matter
may be toxic, or carry toxic (including carcinogenic) trace substances absorbed
into its surfaces. It also imposes costs on physical structures, for example, in terms
of the need to clean and repaint buildings. There is, however, no scientific consen-
sus on the size of particles that may be most harmful – for example, matter of less
than 10 microns (PM–10) or 2.5 microns (PM–2.5) – making economic valuations
difficult, and what estimates have been made generally relate to health aspects,
and especially mortalities. In this context, however, McCubbin and Dellucci
(1999) found that particulate matter, because of the severity of its health impli-
cations, was the most serious road-transport-generated pollutant in the United
States. In terms of physical quantity emitted by transport in the States, PM–2.5
matter has fallen from 7.3 million short tons (2,000 lbs.) in 1991 to 2.61 in 2006,
but PM–10 has risen to 18.42 million short tons from 18.4 over the same period.

Carbon dioxide emissions

The environmental concern here relates to CO2 being generally seen by scien-
tists as a major contributor to the greenhouse effect and consequential global
warming: ‘the balance of evidence suggests a discerning human influence on the
global climate’ (IPCC, 1995). CO2 emissions result from the combustion of fossil
fuels. It was estimated that, globally, transport contributed about 28.6 percent of
CO2 in 2018.
The contribution of CO2 to the atmosphere varies considerably between
countries but the industrialized countries as a whole are responsible for about
80 percent. Although detailed regional apportionment is difficult, Table 6.5

Table 6.5 Share of transport CO2 emissions

1971 1990 1998

OECD
North America 25 29 30
Europe 14 20 23
Pacific 16 20 22
Non-OECD
Africa 20 18 17
Middle East 14 20 18
Europe 10 9 13
Former USSR 9 9 8
Latin America 31 33 34
Asia (excl. China) 14 16 18
China 4 6 8
World 19 22 24

Source: IEA (2000).

BUTTON 9781786435668 PRINT.indd 206 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 207

provides some information on the division between the wealthier countries within
the OECD and other countries at the end of the twentieth century. The general
trend is clear. By 2017, the major emitting regions were Asia (29.6 percent)
and North America (28.9 percent), with Africa responsible for 2.7 percent.
(Population differences cannot account for this: Africa’s population is about 1.22
billion, North America’s 579 million, and Asia’s 4.56 billion.)
Since CO2 is a natural constituent of air (although only about 0.03 percent)
it is not strictly a pollutant. Additionally, excess amounts of the gas have no
detrimental effect on personal health. The problem is that there is mounting,
although some would argue not yet conclusive, evidence that high levels of CO2
in the atmosphere, by preventing heat from escaping from the planet, will lead to
global climate changes.
The issue is not really one about the merits of the greenhouse effect per se
(without it, estimates suggest the global average temperature would fall to about
19°C), but rather about the desirability of the effects which changes in its intensity
will have. The exact geographical impacts of global warming, and its timing, are
difficult to predict, and the long-term economic consequences are even harder
to foretell. The types of problems which are feared, however, include: a rise in
sea level as a result of thermal expansion of the sea and the melting of land ice;
changes of climatic zones, for example, of desert regions and regions affected by
tropical storms; detrimental effects on water resources in many areas; and prob-
lems of adapting agricultural production.
The wide-ranging potential impacts of global warming make it particularly
problematic to place money values on them. The ‘Stern review’ (Stern, 2007) con-
cluded that 1 percent of global GDP per annum needs to be invested to avoid the
worst effects of climate change, and that failure to do so could risk global GDP
being up to 20 percent lower than it otherwise might be.

Nitrogen oxide emissions

These pose difficulties when combined with other air pollutants or in areas where
residents already suffer ill-health. In the latter case they can lead to respiratory
difficulties and extended exposure can result in oedema or emphysema. At the
trans-boundary or regional level, NOX emissions converted to nitric acid and
combined with sulfur dioxide (SO2) form a significant component of ‘acid rain’
(or ‘acid deposition’) that has serious detrimental effects on ecosystems, for
example damage to fresh-water fish stocks and deforestation. About 50 percent
of NOX emissions stem from the transport sector, and the rest from the energy
and industrial sectors, although in many countries their output is falling. In the
United States, it fell from 26.9 short tons in 1970, to 22.6 short tons by 2000, and
to 18.2 short tons in 2006. The decline in transport-associated NOX is largely due
to the widespread adoption of auto catalysts; from 1990 all new cars in Norway,
Sweden, Austria, and Switzerland had these fitted, and most other industrialized
countries quickly followed.

BUTTON 9781786435668 PRINT.indd 207 27/04/2022 11:07

 208 TRANSPORT ECONOMICS, 4TH EDITION

Carbon monoxide emissions

Carbon monoxide (CO) can have detrimental effects on health by interfering with
the absorption of oxygen by red blood cells. This may lead to increased morbid-
ity, and adversely affects fertility. There is also evidence that it affects worker pro-
ductivity. CO is especially a problem in urban areas where synergistic effects with
other pollutants mean it contributes to photochemical smog and surface ozone
(O3). Concentrations of O3 at lower levels have implications for the respiratory
system. CO emissions result from incomplete combustion, which is most likely
to occur at low air/fuel ratios in the engine. These conditions are common during
vehicle-starting when the air supply is restricted.
In the 1980s some 90 percent of all CO emissions originated from the trans-
port sector in most industrial countries, and about 80 percent were associated
with automobile use. The figure reached 100 percent in the center of many built-
up areas. CO emissions from transport more than halved in the United States
in just over 35 years from 1970 to 100.6 million short tons in 2006. As with the
United Kingdom (see again Table 6.2), in the United States transport accounts
for about half of CO emissions.

Sulfur dioxide emissions

Emissions of this colorless but strong-smelling gas can result in bronchitis and
other diseases of the respiratory system and they are, in conjunction with NOX
emissions, the major contributor to acid rain. Transport is directly responsible
for about 5 percent of SO2 emissions with diesel fuel containing more SO2 per
liter than gasoline. What is more important, coal-fired electricity generation is a
major source of this gas and thus there are further transport implications both for
electric rail transport and the manufacture of transport vehicles. Modern diesel
engines are being adopted in wealthier countries, reducing the sulfur problem,
and enhanced technology is combating the levels of SO2 associated with coal-
fired electricity generation.
There have been several studies at various levels of spatial aggregation to try
to place an economic value on the costs acid rain imposes on forestry and the ben-
efits of policies to reduce it. Estimates by Chestnut and Mills (2005), for example,
of the quantified health and environmental benefits of the US Acid Rain Program
totaled over $100 billion annually for 2010.
The trend in SO2 emissions in general, including transport, has been down-
wards in many higher-income countries. Table 6.6 provides a decade of data for
the United States showing the overall decline in SO2 emissions since 2009 by type
of activity. The direct use by transport (highway vehicles and off-highway) has
clearly fallen dramatically both in absolute terms and in relative terms compared
to the overall levels of sulfur emissions.

Volatile organic compounds

These comprise a wide variety of hydrocarbons (HC) and other substances (for
example, methane, ethylene oxide, formaldehyde, phenol, phosgene, benzene,
carbon tetrachloride, CFCs, and polychlorinated biphenyls). They generally

BUTTON 9781786435668 PRINT.indd 208 27/04/2022 11:07

 209
THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­

Table 6.6 United States emissions of sulfur dioxide (million tons)

2009 2011 2013 2015 2017 2019

Highway vehicles 0.04 0.03 0.03 0.03 0.03 0.01

Off-highway vehicles 0.21 0.13 0.09 0.06 0.04 0.04
Fuel combustion 7.97 5.52 4.02 3.20 1.81 1.44
Industrial processes 0.70 0.58 0.53 0.48 0.44 0.44
Waste disposal & recycling 0.02 0.02 0.03 0.03 0.03 0.03
Miscellaneous 0.15 0.20 0.18 0.15 0.22 0.22
All sources 9.09 6.48 4.87 3.95 2.77 2.17

Source: US Bureau of Transportation Statistics.

result from incomplete combustion of fossil fuels, although evaporated gasoline

from fuel tanks and carburetors is increasingly contributing to releases of aro-
matic HC such as benzene.
When combined with NO in sunlight, HC and some volatile organic
compounds (VOCs) can generate low-level ozone, the main component of
photochemical smog. Besides producing respiratory problems and causing eye
irritations, some of the compounds are suspected of being carcinogenic and
possibly mutagens or teratogens (which can result in congenital malformations).
Excluding methane, emissions of which largely stem from agricultural sources,
about half of VOC emissions in industrialized countries are generally associated
with road traffic and the proportion in developing countries tends to be rising.
About 30 percent of all VOC emissions are directly related to transport.

Accidents

Transport is a dangerous activity. Every year more than 1.17 million people die in
road crashes around the world, and over 10 million are crippled or injured. These
accidents can concern not just those involved in transport itself but also third
parties. The dangers inherent in the transport of dangerous and toxic substances
are, in fact, increasing this latter problem. From a purely statistical perspective,
road transport resulted in 39,888 lives being lost in the United States in 2014,
amounting to 124 deaths per million inhabitants, and only slightly less in the
European Union. There are quite wide variations at a lower level of aggregation
in the chances of being killed in a transport accident. For example, within the
European Union in 2013/14 the number of road fatalities per million inhabitants
ranged from 29 in the United Kingdom, 51 in Malta, 48 in the Netherlands, and
22 in Sweden through to 66 in Lithuania and 77 in Poland.
It should, however, be pointed out that in many high-income, developed
countries the number of fatal road accidents is decreasing – for example, road
fatalities for the United Kingdom reveal 4,753 deaths in 1991, falling to 3,743 in
1997, to 3,368 in 2004, and 1,752 in 2019 (the year before Covid effects began)
with the general pattern repeated for injuries. For Germany, the decline has been
even more pronounced, falling from 11,300 fatalities in 1991 to 5,842 in 2004,

BUTTON 9781786435668 PRINT.indd 209 27/04/2022 11:07

 210 TRANSPORT ECONOMICS, 4TH EDITION

and to 3,036 in 2019 while for Italy the figures for 1994, 2004, and 2019 are 7,036,
6,122, and 3,173 respectively.
This is not, however, the situation in many low-income countries where, as
private transport is expanding, the number of fatalities continues to rise. There
is about a 70 percent greater chance of dying in a road accident in Africa than
America, and a 185 percent greater chance than in Europe. Increased amounts
of hazardous waste are being transported each year, and the related problem of
spillage is also adding to the risks borne by third parties throughout the world,
but particularly so in lower-income countries.
If one considers the accident rates by mode then road transport incidents
dominate statistics, although, because of variations in modal split between coun-
tries, there are national variations in their relative importance. Some indication
of different accident rates by mode and over time for the United States is, for
example, seen in Table 6.7.
Interpretation of such data does, however, pose some problems: there is
the point of comparison against which numbers of accidents should be set.
Commercial aviation is, from a statistical perspective, generally cited as the
safest mode of transport, but this may not be the case viewed in terms of time
exposure.
Valuing the external accident costs of transport poses a particular problem
(Jones-Lee and Loomes, 2003). Accident risks are partly internalized within
transport in the sense that individuals insure themselves against being harmed by
them. However, many travelers have no insurance, or, where it has been taken up,
it is based on a misperception of the risks involved. Where there are also third-
party risks involved in the possibility of accidents is during the transporting of
dangerous goods or toxic waste. Attempts to devise methods for valuing accident
risk have a long history, especially regarding fatal accidents.

Table 6.7 Transport accidents in the United States by selected mode

Air Highway (thousand) Railroad Waterborne

Fatalities
1995 964 41.8 1,146 1,016
2000 764 41.9 937 888
2005 603 43.5 887 777
2015 406 35.5 749 700
2019 452 36.1 899 697
Injuries
1995 452 3,593 14,440 6,165
2000 357 3,260 11,643 5,112
2005 302 2,728 9,402 4,095
2015 283 2,455 9,130 3,357
2019 259 2,727a 7,914 2,989

Note: a. 2018 data.

Source: US Bureau of Transportation Statistics.

BUTTON 9781786435668 PRINT.indd 210 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 211

The methods of valuation currently in use, however, still differ between countries,
and agencies within countries. Some adopt cost avoidance calculations, others
use lost-production/consumption-type techniques, but the use of revealed and
stated preference methods is becoming more widespread. The lost-production (or
export) method essentially asks what output the economy forgoes if, for example,
someone is killed in a road accident – essentially a discounted calculation of the
difference between what that person could have been expected to produce over the
rest of their life and what they could have been expected to consume. The obvious
problem is that a pensioner’s death would be accorded a positive value with such
a procedure. The lost consumption (or ex ante) method avoids this problem by
assuming that the individual would gain utility by not dying and thus does not net
out lost consumption, the ability to enjoy this consumption acting as proxy for
the welfare of remaining alive.
Analysis based on microeconomic principles looks at choices that people
make when trading off safer travel against riskier options and estimates the
willingness to pay for the reduced risk. The revealed preference work focuses on
actual choices, such as driving faster, which saves time but is more dangerous, or
driving more slowly, which is more costly in time.
There is still no universally accepted value for accident prevention, and coun-
tries adopt a variety of valuations, and in some use different values according to
mode or circumstance. The United Kingdom for example, uses a figure of about
£1.93 million in 2020 for a statistical life saved in road project appraisal based on
stated preference analysis, and £216,915 and £16,722 for serious and slight non-
fatal accidents respectively. The United States used a value of up to $9.6 million
in 2017 for a fatality, and Canada used $6.31 million in 2018. Academic studies
also show some variability in their results. An early review of such studies in the
United States, the United Kingdom, and Sweden using mainly stated preference
methods concludes that the most reliable estimates from such studies give a distri-
bution of values of life in 1989 with a median of $1.1 million and a mean of $3.4
million (Jones-Lee, 1990).
While reservations must be expressed over the method of valuing lost life (in
terms of lost production) and some of the other forms of accident damage, these
types of figures can also be aggregated to give very broad national overall costs of
accidents – for example, the American Automobile Association estimated that the
costs of property damage, lost earnings, medical costs, emergency services, legal
costs, and travel delays due to road accidents amounted to $164.2 billion in 2007.
It must be remembered, however, that such figures are gross of the internalization
that takes place through insurance markets, and do not include the costs of lost
lives and injuries.

Visual Intrusion

Transport infrastructure and mobile plant is frequently visually intrusive and

often far from aesthetically pleasing. In addition, particulates and gases emitted
by transport scatter and absorb light, adversely affecting visibility. The problem

BUTTON 9781786435668 PRINT.indd 211 27/04/2022 11:07

 212 TRANSPORT ECONOMICS, 4TH EDITION

is measuring these effects. Some attempts have been made in the past to assess
the intrusion of motorways on the landscape by looking at the percentage of the
skyline obscured, but this approach only considers one dimension of a multi-
faceted problem. Transport infrastructure must be viewed in the context of its
surroundings – a new freeway located in formerly unspoiled countryside is likely
to be viewed differently from one that blots out an unsightly waste tip. Design is
also important. Also, it should be remembered that vehicles are as intrusive as
infrastructure and large trucks or buses are, for example, often totally out of place
in unspoiled areas or ‘historic towns’. Whether it is the actual size of vehicles that
is alarming or simply the level of traffic flow is difficult to disentangle.
A newer problem is that caused by the eyesores created by the difficulties
of disposing of the old hardware of transport. The problem not only embraces
disused infrastructure of road, rail, and maritime transport but also increasingly
the vehicles themselves: cars, ships, and railway wagons. Variations in the number
of scrapped vehicles over time are due to a variety of factors, including changes
in fuel prices (Jacobsen and van Benthem, 2015), but as the overall vehicle popu-
lation grows, disposal problems increase. Nevertheless, the weight of passenger
cars, vans, and other light goods vehicles scrapped in the European Union in 2019
was 6.9 million tonnes; 95.1 percent of the parts and materials were reused and
recovered, while 89.6 percent were reused and recycled.

Pollution of Water Systems

Water systems, both fresh and saline, suffer considerable pollution and other
environmental damage from transport. Maritime transport itself results in both
accidental and intentional releases of waste and oil into the seas, lakes and rivers,
and ports, especially those requiring significant amounts of dredging, and are
disruptive to wildlife. Other modes, however, can also cause damage including
the run-off from roads and airports of liquids such as de-icing fluids, and the
­diversion of natural water-courses to allow for the construction of infrastructure.
Although most oil spillage (53 percent of incidents) is the result of transfers,
major oil spillages due to maritime accidents attracts considerable attention:
for example, the Atlantic Express spilled 287,000 tonnes of oil in 1979; the ABT
Summer, 260,000 tonnes in 1991; the Castillo de Bellver, 252,000 tonnes in 1983;
and the Amoco Cadiz, 233,000 tonnes in 1978. The quantity of oil spilt is not,
however, a good indicator of the environmental damage done. The Exxon Valdez
incident in 1989, for example, because it occurred in a scenic area, Prince William
Sound on the Gulf of Alaska, is considered the most expensive in history despite
the spillage being 37,000 tonnes. The costs of this in terms of clean-up ($2.2
billion), lost production in fishing and other industries ($300 million), and lost
fishing output in south-central Alaska ($108 million), besides damages paid by
the ship’s owner, are large, although strictly difficult to calculate exactly because
of problems of double-counting (Cohen, 1995). For example, while south-central
Alaska may have lost business, other areas may have gained and some of the costs
that have been included in calculations for damages paid by the ship’s owners are

BUTTON 9781786435668 PRINT.indd 212 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 213

already reflected in the direct damage and clean-up cost estimates. Ships are also
generally insured and this means that at least part of the cost of spillage is inter-
nalized, as with many forms of accident risk.
There are also technical issues in physically assessing damage from spillage
prior to any efforts at placing monetary values on them. A spill on marshland
in winter does minimal damage to plant life, for example, because it has died
back naturally and weather conditions can affect the rate of evaporation and
toxicity of oil (Talley, 2001). There is also damage to water systems from indi-
rect transport-related events, for example oil released when ships sink and on-
shore maritime fuel storage tanks are damaged due to earthquakes and adverse
weather.

Vibrations

Low-flying commercial aircraft, heavy goods vehicles, and railway wagons create
vibrations that can affect buildings. Vibration within vehicles also adversely
affects drivers and passengers. Vibration within a ‘cabin’ starts from the engine
and the response of the vehicle to the road surface. Vibration varies in response
to the load of the vehicle; there is more vibration as the load becomes lighter. The
most widely reported injury for whole-of-body vibration is back injury.
Again, useful measures are elusive. While it is known, for example, that
ground-borne vibration is related to axle loads, it has proved impossible to relate
this effectively to any measure of structural damage. The evidence suggests,
however, that the physical damage caused may be less than is sometimes claimed.
Improved engineering techniques have reduced the damage caused by road trans-
port and much of the damage formerly thought ‘caused’ by heavy lorries is more
likely to have simply been ‘triggered’ by them. As Whiffen and Leonard (1971)
pointed out half a century ago, ‘[a]ttention can be drawn to vibration by the rat-
tling of doors, windows, lids of ornaments, mirrors, etc. The association of these
audible and visible signs with the possibility of damage to the building results in
exaggerated complaints about vibration, even though, in fact, there may be no
risk of damage’. Vibrations may still be a cost in an economic sense, however,
even if there is no structural damage to buildings. This is not a new problem:
Martin (1978) found that 8 percent of the United Kingdom’s population was con-
siderably bothered by vibrations from road traffic. Since then, our knowledge of
the implications of low-frequency noise has increased considerably, with evidence
of discomfort, irritability, and sensitivity arising from sleep disruption (Araújo
Alves et al., 2020).

Community Severance

Roads, railways, canals, and other transport arteries often present major physical
(and sometimes psychological) barriers to human contact. An urban motorway
can cut a local community in two, inhibiting the retention of long-established
social ties, and, on occasions, making it difficult for people to benefit from

BUTTON 9781786435668 PRINT.indd 213 27/04/2022 11:07

 214 TRANSPORT ECONOMICS, 4TH EDITION

recreational and employment opportunities on the other side of the barrier. A

rail line can do the same. Although it may be possible to obtain estimates of the
potential pedestrian and other delays, and reassignments resulting from the imped-
ance, suppressed trips are much harder to identify. Severance also has important
links with amenity; as Susan Handy (2003) puts it, ‘the quality of a place, the way
it looks, sounds, smells and feels … affects the way people experience a place’.
Giving the largely qualitative implications of community severance full
quantification, let alone any economic valuation, is not an immediate prospect.
Progress has, however, been made since the late 1970s when the United Kingdom’s
Jefferson Report found that ‘the overall conclusion is that no acceptable way is
seen of extending the assessment of severance beyond an individual examination
of some of the perceived effects except, perhaps, by means of subjective state-
ments in appropriate cases’. Anciaes et al. (2016), for example, consider a number
of ways in which community severance may be measured and its effects on health
evaluated. They base their analysis on four questions: Are there barriers restrict-
ing mobility and accessibility? Who is affected? How do people travel? And where
do they go? Quantification of severance is then assessed by metrics that consider
time losses, making use of examples from a variety of national approaches. What
is still missing is any quantification of the mental costs of limiting mobility or
access, or the feeling of isolation.

6.6 Energy Use

Energy use per se is not strictly an environmental concern in itself; it is more

a matter of the types of energy used and in what context. The various energy
sources have associated with them different environmental implications. Even
wind power must be generated by man-made equipment usually involving the use
of other energy sources, and wind farms can create local micro climates in their
locations. There are broader issues about the non-renewable nature of oil, the
main fuel used by transport.
To move anything requires energy. Transport, therefore, is a user of energy.
The amount and types of energy that are used, however, have varied consider-
ably over time as technology has changed. Much of it used to be in the form of
the food given to animals, beasts of burden, or simply human consumption of
calories in various forms. Renewable sources of energy, wind, and water power,
have been widely exploited and still are, although often in somewhat different
ways. Wind power, for example, is more often used indirectly as a means of
generating electricity than for propelling ships or Chinese wheelbarrows. Much
of modern transport since the advent of the steam engine has, however – and
excepting some less developed countries where the beast of burden often still
plays a major role – relied on non-renewable sources of energy such as coal, oil,
and natural gas.
The levels of current and projected use of energy by the transport industries
are particularly impressive. While there are national differences, a pretty clear

BUTTON 9781786435668 PRINT.indd 214 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 215

Table 6.8 Energy consumption by sector in the United States

Quadrillion British thermal units

1995 2000 2005 2010 2015 2019 2020

Residential 34 34.5 32.5 31 32 32.5 31

Commercial 28 27 28 27 27.5 28 24
Industrial 18 21 22 22 20.5 21 21
Transport 15 17 17.5 18 18 17.5 16

Source: https://www.statista.com/statistics/239790/total-energy-consumption-in-the-united-states-
by-sector/.

universal pattern of energy use emerges. Table 6.8 for example provides some
information on relative final demand for energy by the main economic sectors in
the United States. Transport accounted for some 17 to 17.5 percent over annual
energy use over the 25 years to 2019, after which the Covid-19 pandemic of 2020
reduced absolute consumption somewhat as the United States along with the rest
of the world’s economy went into recession. In terms of oil demand, however,
because of the need for a mobile source of power, transport has been the domi-
nant sector throughout the pre-Covid period. For example, the United States
consumed about 20 million barrels of oil products per day, of which 14 million
barrels were used for transport, 9 million of these being gasoline. But a succes-
sion of global and national initiatives are now in place, designed to, amongst
other things, reduce transport’s dependence on carbon-based fuels such as oil.
Relatively, though, transport is still expected to account for 30 percent of final oil
demand in 2030, remaining the largest single consumer.
The exact effects of these global environmental initiatives are uncertain.
Historically, a major factor that has contributed to the rise in energy demand for
transport has been technology changes and stemming from these in particular has
been the increasing use of road transport (see again Chapter 2). Future trends
are likely to change the relationship. The global number of registered gasoline-
powered cars and diesel commercial vehicles in 2003 was, respectively, about 589
and 224 million, but by 2020 the world’s vehicle fleet was estimated to be 1.06
billion passenger cars and 363 million commercial vehicles. But many forecasters
are suggesting international agreements, such as those reached at the COP26 UN
Climate Change Conference, will reverse this trend as policies associated with
containing global warming effect a disconnect between vehicle numbers and oil
consumption. For example, hybrid and electric vehicles will become a larger part
of the car park, reducing carbon emissions per mile. The challenge for the fore-
caster is to predict the number of miles driven.
But it is not just the number of vehicles that matter when it comes to emis-
sions; technical factors also come into play. For example, there is fuel efficiency.
The average new passenger car in Europe consumes about 6.5 liters of fuel per
100 kilometers, whereas the average passenger car in the United States uses over
40 percent more to cover the same distance. Part of this can be explained in terms

BUTTON 9781786435668 PRINT.indd 215 27/04/2022 11:07

 216 TRANSPORT ECONOMICS, 4TH EDITION

of distances traveled: longer trips in America may be seen to justify more ‘com-
fortable’ vehicles but in addition there are important taxation differences. Retail
gasoline prices in Europe include taxes in the range of 60 to 75 percent, compared
with only 20 to 25 percent in the United States. There may also be cultural differ-
ences in the way various societies see large, less fuel-efficient vehicles, but these are
more difficult to quantify.
Transport is not homogeneous and can be broken down in several ways to
reflect its use of energy. The focus has largely been on the use of non-renewable
energy resources, and especially oil, although electricity, an indirect energy source,
is used by many rail systems and local trams. Electricity can be generated in a
variety of ways, from oil, natural gas, coal, nuclear sources, hydro power, wind
power, and so on, and thus its environmental implications are not always trans-
parent. If, as many forecasts suggest, electric-powered personal vehicles, possibly
with many being autonomous, have economic advantages in some markets, they
will take an increasing market share from gasoline vehicles (Holland et al., 2016).
Much of the energy mix will depend on local and national policies being pursued
regarding gasoline, electric, and autonomous vehicles (Holland et al., 2021). This
is a time of transition.
Given considerable variations in the efficiency of generating plants, one
would really like an indicator of the amount of fossil fuel used to provide the
energy to produce a given unit of transportation. Additionally, most of the data
available on energy consumed in transport relate to the final movement and offer
few insights into the full costs of transport provisions that embraces the energy
needed to supply and maintain transport infrastructure and the manufacture and
maintenance of vehicles. We also have limited knowledge on the way transport
affects the use of resources in the broader economy – for example, on the effect
that transport-intensive industries such as tourism have on energy consumption
in final production such as hotels, restaurants, and the manufacturing of souve-
nirs, as well as in the movement of the tourists themselves.
In terms of its immediate effects, Table 6.9 looks at energy consumption
by various transport modes in the United States. The dominant role of gaso-
line as an energy source is clear and reflects the widespread use of automobiles
for personal travel. Other countries have somewhat different relative patterns
that depend, in part, on the nature and size of their national economies and
geography (for example, whether they produce and move large amounts of
raw materials), but also on the transport policies that have been favored (for
example, whether public transport has been strongly supported and levels of fuel
taxation).
While much of the interest in energy consumption until the 1990s focused on
its use in developed Western economies, the subsequent rapid economic expan-
sions of large developing countries, especially Brazil, India, and China, has led to
a shift in attention.
With its growth in GDP and aggressive expansion of transport infrastructure,
China, for example, saw a fourfold increase in freight traffic and a sixfold increase
in passenger traffic between 1980 and 2000. Cars for short trips and planes for

BUTTON 9781786435668 PRINT.indd 216 27/04/2022 11:07

 217
THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­

Table 6.9 Fuel consumption in the United States by the main transport modes

1980 1990 2000 2010 2016

Air
Certificated carriers
   Jet fuel (million liters) 32,249 46,228 52,631 40,990 42,273
General aviation
  Aviation gasoline (million 1,968 1,336 1,260 818 780
liters)
  Jet fuel (million liters) 2,900 2,510 3,679 5,513 5,440
Highway
Gasoline, diesel, & other fuels
(million liters)
   Light duty vehicle, short 265,683 264,067 277,375 334,474 346,319
wheelbase, & motorcycle
  Light duty vehicle, long 90,078 134,802 200,395 133,756 143,160
wheelbase
  Single-unit 2-axle 6-tire or 26,206 31,635 36,200 53,811 58,062
more truck
  Combination truck 49,350 61,070 97,155 106,677 111,876
  Bus 3,854 3,388 4,210 7,329 8,426
Transit
Electricity (million kWh) 2,446 4,837 5,382 6,534 6,604
Motor fuel (million liters)
  Diesel 1,632 2,464 2,236 2,365 2,233
  Gasoline & other non-diesel 42 129 89 383 439
fuels
Compressed natural gas n.a. n.a. 165 486 632
Rail, Class I (freight service)
Distillate/diesel fuel (million 14,778 11,792 14,006 13,949 12,814
liters)
Amtrak
Electricity (million kWh) 254 330 470 555 516
Distillate/diesel fuel (million 242 310 359 240 1,908
liters)
Water
Residual fuel oil (million liters) 33,887 23,948 24,264 17,262 11,091
Distillate/diesel fuel oil (million 5,595 7,816 8,560 8,076 8,498
liters)
Gasoline (million liters) 3,982 4,921 4,256 4,179 8,792
Pipeline
Natural gas (million cubic feet) 17,971 18,684 18,185 19,476 19,749

Note: n.a. = not available.

Source: US Bureau of Transportation Statistics.

long trips increasingly dominated passenger traffic in the country, as growing

incomes allowed more people to utilize these fast and comfortable means of travel.
From 1980 through 2002, passenger traffic grew from 228 billion to 1,413 billion
person-kilometers with highways’ share increasing from 32 to 55 percent of total

BUTTON 9781786435668 PRINT.indd 217 27/04/2022 11:07

 218 TRANSPORT ECONOMICS, 4TH EDITION

Table 6.10 Energy consumption by transport in China (1990–2000)

Mode Energy consumption (ktoe)a Consumption share (%)

1990 2000 Ratio 1990 2000 Change

Railways 14,851 13,017 0.88 27.8 13.5 –14.2

Highways 25,495 65,516 2.57 47.6 68.1 +20.5
Waterways 11,407 11,988 1.05 21.3 12.5 –8.8
Civil aviation 1,222 5,090 4.16 2.3 5.3 +3.0
Pipelines 550 605 1.10 1.0 0.6 –0.4
Total 53,524 96,214 1.80 100.0 100.0 0.0

Note: ktoe = kilotonne of oil equivalent.

Source: China Energy Research Society (2002).

person movements. Passenger aviation traffic grew more than thirtyfold, its share
of aggregate passenger movements growing 9 percent. Railway passenger traffic,
while nearly quadrupling in volume, however, saw its share of the market declin-
ing from 61 to 35 percent. Since the early 2000s, China’s expanded high-speed rail
network has, however, absorbed some of the increase in overall long-distance pas-
senger transport that would have likely otherwise used road or air modes.
As the result of the rapid traffic growth and the changing modal split, the
transport share of national energy use grew from under 5 percent in 1996 to
nearly 9 percent in 1999. The share of energy used by road transport in China
officially grew from roughly 48 percent in 1990 to 68 percent in 2000, and most of
this is in the form of oil consumption (Table 6.10). The share of civil aviation also
grew rapidly, albeit from a very much lower base.
Most of the analysis of transport energy use focuses on its importance in
moving vehicles of one form or another, but both the mobile plant used in trans-
porting goods and people, and the associated infrastructure, rely on significant
amounts of energy in their construction and maintenance. While difficult to
quantify, for example, the production of over 50 million cars, nearly 14 million
light commercial vehicles, and three million heavy commercial vehicles in 2006
obviously consumed an immense amount of energy.
As with many things, there is an intellectual curiosity about the links between
transport and energy use, but there are also important public policy issues to be
considered. Energy is used in virtually all forms of activity and there is a need
to ensure that it is used to maximum effect and in ways that ensure any external
affects are not excessive. In economic terms, the market for energy is, however, far
from perfect for a variety of reasons. These stem partly from the intrinsic nature
of the ‘commodity’ (largely associated with market failures linked to economies
of scale in supply and externalities), but can also be due to the institutional envi-
ronment in which energy is provided (especially government intervention failures
that often are seen in terms of allocating property rights and regulatory capture).
These imperfections, in turn, affect the ways in which transport users view energy

BUTTON 9781786435668 PRINT.indd 218 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 219

and the ways in which they use it, and the forms and quantities in which it is
supplied.
Much of the energy used in transport comes from finite sources: oil reserves,
coal, wood, and natural gas. In economic terms, this is not a major issue if prices
are appropriate and reflect the genuine, long-term opportunity cost of the use
of these resources. In many cases, the drawing-down of the reserves of these
resources may still be consistent with a genuinely ‘sustainable’ scenario in the
Brundtland Report sense (World Commission on Environment and Development,
1987) of ensuring that future generations enjoy the same resource base as current
generations, if at the same time alternative energy sources are being created – for
example, the creation of hydroelectric or wind capacity. In terms of the notion of
sustainable development, future generations will still have the same resource base
as the current one, albeit it in a different form.
The challenge is to ensure that there are mechanisms and signals to guar-
antee that the energy base is not diluted excessively by transport use. In the past
there have been significant shifts in the energy used in transport, with coal, and
then oil, taking over from oars and sails in shipping for example. Market forces
have largely driven these shifts; slaves became expensive as rowers, and sailing
ships became too unreliable for expanding trade networks and hence steamships
took over. One thing that has been learned, however, is that predicting the deple-
tion rate of any resource is difficult. Stanley Jevons’ famous concern in 1865 that
coal supplies would soon be exhausted and, in consequence, the rail and steam-
ship industry would, amongst others, become non-viable is a good example of
how static analysis linking non-renewable resource depletion and transport can
be misleading. But equally, the move from wood to coal and then to oil boilers on
ships showed how the market can respond to potential shortages through stimu-
lating the development of alternative technologies.
The economic problem is that for transport markets to function they must
have appropriate price signals from the energy market. The semi-cartelization of
many energy markets, with institutions such as the Organization of the Petroleum
Exporting Countries (OPEC), and of many markets that supply the hardware
of transport, such as the automobile and airframe manufacturers, coupled with
political involvement, means that these signals are far from perfect. Consequently,
the exploitation of any non-renewable resources is seldom optimal, irrespective of
any externality considerations. The issue, however, is more of a generic one rather
than being transport-specific, because market and institutional failures extend
across all uses of energy.

6.7 Introduction to Traffic Congestion

The demand for transport is not constant over time. In large cities there are
regular peaks in commuter travel while on holiday routes; both within a country
and to overseas destinations, there are seasonal peaks in demand. Transport infra-
structure, although flexible in the long run, has a finite capacity during any given

BUTTON 9781786435668 PRINT.indd 219 27/04/2022 11:07

 220 TRANSPORT ECONOMICS, 4TH EDITION

Table 6.11 Traffic congestion in major United States cities (2005 and 2017)

Large metropolitan areas Annual delay per traveler (hours)

2005 2017

Los Angeles 72 119

San Francisco–Oakland 60 103
Washington DC 60 102
Atlanta 60 73
Dallas–Fort Worth 58 67
Houston 56 75
Detroit 54 61
Miami 50 68
Phoenix 48 62
Chicago 46 73
New York–Newark 46 92
Boston 46 80
Seattle 45 78
Philadelphia 38 62

Source: Texas Transportation Institute.

period. One cannot, for example, expand and contract the size of an airport ter-
minal to meet seasonal fluctuations in demand. When users of a particular facility
begin to interfere with other users because the capacity of the infrastructure is
limited, then congestion externalities arise and time is wasted (Table 6.11). What
we see, in terms of travel in some major cities at less than the authorized speed
limit, is that large amounts of time are spent in traffic delays and that these delays
are becoming longer over time. We also see considerable variation in the economic
costs between cities, and that the ‘ranking’ of cities by their congestion levels has
changed between cities between 2005 and 2017.
One could argue about whether the measure of congestion reflects the real
economics of delays. Speed limits, for example, generally depend on the engineer-
ing standard of roads and on safety on considerations rather than on notions
of optimal congestion level. Some degree of congestion is almost unavoidable if
facilities are not to stand idle most of the time. The question is just how much
congestion is desirable. Since people accept a low level of congestion but resent
excessive congestion; because of the time and inconvenience costs imposed, there
is some implied notion of an optimal level of congestion that is not captured in
engineering criteria – a topic returned to later. There may, for example, be thresh-
olds of congestion for different journey purposes (Sweet, 2014).
It is not just roads that experience congestion. It can be found in most modes of
transport and can be either on links or at nodes. Table 6.12, for example, looks at air
traffic control delays in Europe, but there are delays at airports as well. Remaining
with Europe, for example, in 2006, 31.8 percent of flights were delayed by at least 15
minutes out of London’s Heathrow Airport, 31.3 percent from Madrid, 30.7 percent
from London Gatwick, and 28.6 percent from Paris Charles de Gaulle.

BUTTON 9781786435668 PRINT.indd 220 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 221

Table 6.12 European air traffic flow management delays

Average daily % of flights En route delays Airport delays

traffic delayeda (million minutes) (million minutes)

1997 19,658 7 15.5 5.4

1998 20,681 9 21.7 5.7
1999 22,064 12 36.3 7.0
2000 23,071 9 24.4 7.4
2001 23,001 8 20.8 6.8
2002 22567 9 11.9 6.0
2003 23,197 6 8.0 6.8
2004 24,238 4 7.6 7.3
2005 25,244 4 8.9 8.7
2006 26,286 5 10.2 8.2

Note: a. A delay is defined as 15 minutes or more behind schedule.

Source: EUROCONTROL Performance Review Unit.

One should add, with reference to previous sections, that congestion does not
only impose costs on the traveler in terms of wasted time and fuel (the pure con-
gestion cost) but the stopping and starting it entails can also worsen atmospheric
and other forms of pollution. The problem is particularly acute with local forms
of pollution because road traffic congestion tends to be focused in areas where
people work and live. Road traffic poses some of the greatest congestion problems
and offers a useful basis of analysis.
The economic costs of road congestion can be calculated using the engineer-
ing concept of the speed–flow relationship. If we take a straight one-way street
and consider traffic flows along it over a period at different speed levels then
the relationship between speed and flow would appear as in Figure 6.7. Flow is
dependent upon both the number of vehicles entering a road and the speed of
traffic. Hence, at low volumes of traffic, when vehicle impedance is zero, high
speeds are possible, constrained only by the capability of the vehicle and the legal
speed limits, but as the number of vehicles trying to enter the road increases, so
they interact with existing traffic and slow one another down. As more traffic
enters the road, speed falls, but, up to a point, flow will continue to rise because
the effect of additional vehicle numbers outweighs the reduction in average speed.
This is the normal flow situation.
At the point where increased traffic volume ceases to off-set the reduced
speed, the road’s ‘capacity’ is reached at the maximum flow. (This is the road’s
engineering capacity and differs from the economic capacity that is defined as the
flow at which the costs of extending the capacity are outweighed by the benefits
of doing so.) Absence of perfect information means that motorists often continue
to try to enter the road beyond this volume, causing further drops in speed and
resulting in the speed–flow relationship turning back on itself. These levels of flow
are known as forced flows. There is often a degree of ‘learning from experience’
that can improve the quality of decision-making and in practice, without any

BUTTON 9781786435668 PRINT.indd 221 27/04/2022 11:07

 222 TRANSPORT ECONOMICS, 4TH EDITION

Average speed

Engineering
capacity

Normal flow
Unstable
zone of flow

Forced flow

0 Flow (vehicles per hour)

Figure 6.7 The speed–flow relationship

intervention, flows would settle around the zone of instability during rush-hour
periods. A cross-sectional study of the main urban centers (Table 6.13) conducted
half a century or so ago suggests that this zone of instability occurs at speeds of
about 18 kph, but the situation is little changed today.
The actual form of the speed–flow relationship and the engineering capac-
ity of any individual road will depend upon several factors. Clearly, the physical
characteristics of the road, its width, number of lanes, etc., are of central impor-
tance and may be treated as long-term influences. Short-term factors include
the form of traffic management and control schemes in operation (traffic lights,

Table 6.13 Traffic speeds in selected cities

City Year Population City center traffic speed (kph)

(million)
Peak hour Off-peak

New York 1970 13.3 16.0 26.0

Detroit 1970 4.0 17.7 –
Salt Lake City 1970 0.9 27.0 –
London 1971 7.4 20.6 –
Birmingham 1965 1.1 22.1 –
Leeds 1965 0.5 18.0 –
Paris 1970 6.4 16.9 –
Athens 1971 2.7 15.5 24.0
Copenhagen 1967 1.7 14.5 –
Stockholm 1969 1.3 18.0 –
Calcutta 1971 7.5 11.0–16.0 19.0
Singapore 1972 2.2 21.0 –

Source: Thomson (1977).

BUTTON 9781786435668 PRINT.indd 222 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 223

Speed (mph)
4 lanes

30 2 lanes

20

10
6 lanes

0 2000 4000 6000

Flow (vehicles/hour)

Figure 6.8 Speed–flow relationships on different road categories in Sydney

roundabouts, etc.). Finally, the type and age of vehicles combined with their dis-
tribution may influence capacity.
A typical set of speed–flow relationships which illustrate these points are, for
example, offered by Neutze (1963) in his study of Sydney’s arterial road system.
Information obtained from over 400 locations on main roads in the city was used
in the exercise, the results of which are seen in Figure 6.8. As one might expect,
the capacity of six-lane roads exceeds that of either two- or four-lane roads,
although at most traffic densities the speed is slightly higher on the two- rather
than the four-lane roads. The explanation for this is that traffic management poli-
cies slow down flows of the four-lane roads because roadside parking is permitted
and thus the capacity of curbside lanes is severely restricted, and they also tend to
pass through more densely populated areas with more restrictive traffic manage-
ment controls.
The speed–flow relationship provides a key supply-side input into the
analysis but it is road space that is demanded. A theoretical framework linking
the two elements is described by Alan Evans (1992). Relaxing the basic assump-
tions of the model can lead to modifications to the speed–flow relationship, for
example Verhoef (2005) has shown that in some contexts it is not backward-­
bending at the saturation level but becomes vertical. The density function, the
number of vehicles on a road at any one time, is also important in this type of
analysis.
In Figure 6.9, element B shows the standard speed–flow relationship with
the maximum flow depicted as Fmax. This is traced round to the travel cost–flow
diagram in element C. People essentially demand to join a road and this demand
is seen as the demand curve, D, in element A of the figure. This diagram also
depicts the relationship between travel cost and traffic density – the MC being the
rising marginal cost of congestion each additional motorist imposes on others

BUTTON 9781786435668 PRINT.indd 223 27/04/2022 11:07

 224 TRANSPORT ECONOMICS, 4TH EDITION

Traffic flow
B

Fmax

0
Cost A C

MC S s d mc

0 D 2 D1 Traffic 0 Fmax Traffic

density flow

Figure 6.9 The speed–flow relationship and the demand curve for road space

using the road. The curve rises as the number of vehicles increases. The S curve
represents the cost of joining the road as seen by the additional motorist, in effect
his or her cost of trip-making ignoring the consequences of his/her actions for the
others on the road.
The curves in element C of the figure, concerned with travel cost–flow rela-
tionships, are derived from elements A and B. The s curve is the average cost
relating to congestion in a simple interaction model (see below) and the mc is the
associated marginal curve. These relate directly back to the speed–flow relation-
ship. Generalized costs (see Chapter 5) provide the vital link between physical
traffic flows and cost. Broadly, faster travel in urban areas means cheaper travel
in terms of generalized costs: vehicles are used more effectively and travel times
are reduced. The S curve in Figure 6.9 represents the average generalized cost of
trip-making at different levels of traffic flow. It is a reverse of the speed–flow curve
seen in Figure 6.7, with the positively sloped portion concerning the negatively
sloped section of the speed–flow curve; this stems from the inverse relationship
between speed and generalized cost. The mc curve is the associated marginal
curve that takes into account the congestion costs the additional user places on
the existing traffic flow. The d curve is a derived demand curve reflecting the way
in which the desired traffic flow changes as the cost of travel changes because the
number of vehicles put on the road changes.

BUTTON 9781786435668 PRINT.indd 224 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 225

The actual traffic density that will emerge without any form of traffic
restraints is where the demand for road space equals the average cost (S) of
joining the road: D1. This exceeds the optimal level, where road users take account
of the impedance they impose on others, which is where MC is equated with
demand. Moving across to the flow diagram, which is much more frequently
found in the academic literature, the optimal traffic flow is where the mc curve
intersects the derived demand curve.

6.8 The Economic Costs of Congestion

Most analysis of congestion focuses purely on segment A of Figure 6.9. It is

usually presented in the form seen in Figure 6.10, with the AC curve represent-
ing the average cost of congestion at each level of traffic flow, and the MC curve
the cost of additional traffic to existing flow. The optimal flow is, as we have seen
above, where MC and demand are equated (F2) while the actual flow, because
road users ignore the congestion that they impose on others, tends to be F1. A
further interpretation can be placed on the AC and MC curves. The curves reflect
the average and marginal generalized costs associated with different flows: they
show all the time and money costs borne by road users when trip-making. In this
sense they may be representing ‘social costs’ in the limited sense that they are the
costs to the society of road users.
However, any individual user entering the road will only consider the costs
they personally bear. They will, in most circumstances, either be unaware of or
unwilling to consider the external congestion costs they impose on the other road
users. Consequently, the individual motorist will only consider the average costs
experienced by road users and take no account of the congestive impact of their

Generalized
costs MC AC

J
C2
I
C1
K Demand

0 F2 F1 Traffic flow

Figure 6.10 The economic inefficiency of congestion

BUTTON 9781786435668 PRINT.indd 225 27/04/2022 11:07

 226 TRANSPORT ECONOMICS, 4TH EDITION

trip on other vehicles. It is frequently argued that the MC curve, therefore, relates
to the marginal social cost for the new trip-maker and existing road users of an
addition to the traffic flow while the AC curve is equivalent to the marginal private
cost curve – that is, the additional cost borne and perceived by the new trip-maker
alone. The difference between the AC and MC curves at any traffic flow reflects
the economic costs of congestion at that flow.
It is often important from a policy perspective to gain some idea of the
actual costs associated with excessive congestion. From a social point of view the
actual flow, F1, is excessive because the F1th motorist is only enjoying a benefit
of F1I but imposing costs of F1H. The additional traffic beyond the optimal level
F2 can be seen to be generating costs of F2JIF1, but only enjoying a benefit of
(F2JIF1 – HJI), where HJI is a deadweight loss. A traffic flow lower than F2 is also
suboptimal because the potential consumer surplus gains from trip-making are
not being fully exploited. Of course, this does mean that even at the optimal traffic
flow there are still congestion costs, the area between the MC and AC curves up
to traffic flow F2, but these are more than off-set by the benefits enjoyed by those
using the road.
While the work on congestion costs is extensive, estimating the overall costs
associated with excessive congestion is not simple. Work looking at the money
value of lost travel time has a long pedigree; information on such costs is of
commercial value to public transport suppliers who may trade off faster services
against higher fares. The exact cost depends upon the mix of traffic and the
reason trips are being made. Periodically crude estimates, of the type produced
by the United Kingdom’s Confederation of British Industry (CBI) in 1988, are
made of the costs of time wasted in congestion (about £15 billion per year for
commercial traffic in the prices of the time) but these estimates, and subsequent
ones based on questionnaire surveys tend to suffer from major theoretical and
measurement problems. In the CBI case the calculations were based on scaling
up the responses to a small survey of distribution companies that were asked to
assess the costs traffic congestion was imposing on their operations. Besides the
small size of the sample and the inherent dangers of aggregation, there was no
effort to define a base-line level of optimal congestion as a basis for comparison
nor was there any effort to net out the costs that the distributors were imposing
on others but not paying for.
A more rigorous approach is to consider the opportunity cost of lost travel
time (see Chapter 5). These vary by the nature of the road involved. Using the
figures adopted by the UK Department of Transport when appraising road
schemes, David Newbery (1988), as the result of careful calculations, produced
marginal congestion costs by road type in the United Kingdom (Table 6.14).
While these figures are in themselves dated, they represent solid analysis and show
the differing levels of costs imposed by an additional vehicle joining the various
traffic streams. Aggregation gives an estimated annual congestion cost in the
United Kingdom of about £12,750 million for 1989–90.
Later studies of road congestion costs in the United States include Winston
and Langer (2004), who reviewed congestion costing methods, and, using their

BUTTON 9781786435668 PRINT.indd 226 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 227

Table 6.14 Estimated congestion costs by road type in the United Kingdom (1989)

Road type Marginal cost (pence per mile)

Motorway 0.26
Urban central peak 36.97
Urban central off-peak 29.23
Non-central peak 15.86
Non-central off-peak 8.74
Small town peak 6.89
Small town off-peak 4.20
Other urban 0.08
Rural dual carriageway 0.07
Other dual carriageway 0.19
Other rural 0.05

Source: Newbery (1990).

own model, came up with annual costs of $37.5 billion annually (2004 prices), a
third of which consisted of freight vehicle delays. Weisbrod et al. (2001) evaluated
the economic productivity costs of congestion, particularly to businesses for distri-
bution, and reduced economies of scale and agglomeration, finding them to range
from $20 million to $1 billion per year in typical metropolitan regions. The costs
vary significantly by industry, with higher costs in industries that involve signifi-
cant distribution costs or rely on specialized employees. The Texas Transportation
Institute (2019) has developed a congestion index to calculate congestion costs in
major American cities and converted these into cost measures. The Institute esti-
mated that 2017 traffic congestion across the 494 urban areas examined cost $179
billion nationally, with each commuter averaging 54 hours in congestion.

6.9 Refinements on the Basic Congestion Model

The analysis of congestion, as set out, is based upon a very simple modeling
framework: a linear road, no junctions, homogeneous traffic, and drivers who
are all equally skillful. In practice, as we would expect from our discussion of the
speed–flow relationship, the total cost function varies with the details of the trans-
port system under consideration. Also, considerable traffic congestion stems from
‘incidents’ such as accidents, emergency road repair, and breakdowns that do not
fit comfortably into the simple framework. Traffic incidents account for up to an
estimated 60 percent of delay-hours. Although they are random events, they cause
significant delays when traffic volumes approach road capacity. In uncongested con-
ditions an incident causes little or no traffic delay, but a stalled car on the shoulder
of a congested road can cause 100 to 200 vehicle-hours of delay on adjacent lanes.
William Vickrey (1969) distinguishes five types of congestion relevant in this
rather more complex world. While these are couched in terms of road conges-
tion, they are equally applicable to most other modes of transport – one can quite
simply substitute air-lane or waterway for roads. The types of congestion are:

BUTTON 9781786435668 PRINT.indd 227 27/04/2022 11:07

 228 TRANSPORT ECONOMICS, 4TH EDITION

• Simple interaction. This occurs at comparatively low levels of traffic flow

where the number of mobile units is small. Delays are minimal and usually
result from slow and careful driving on the part of users who wish to avoid
accidents. Total delay tends to vary as the square of the volume of traffic, so
that each additional motorist causes a delay to each other road user roughly
equal to that which the motorist suffers. This is essentially the type of conges-
tion we have been concerned with above.
• Multiple interaction. This occurs at higher levels of traffic flow where,
although the road capacity is not reached, an additional vehicle causes
considerably more impedance to each other vehicle than with simple interac-
tion. Empirical evidence suggests that for every minute the marginal user is
delayed, other vehicles each suffer a delay of three to five minutes.
• Bottleneck situations. These occur when a particular stretch of a road (or
other piece of transport infrastructure) is of more limited capacity than
either the preceding or subsequent links in the network. If the flow is below
that of the capacity of the bottleneck, then either simple or multiple inter-
actions may occur, but once the capacity is reached, and if this is sustained
for any length of time, then queues develop. An exceptionally high level of
congestion is then likely to arise.
• Triggerneck situations. When a bottleneck situation results in queues of
traffic, these may impede the general flow of traffic even for those not
wishing to use the section of road with limited capacity. At the extreme, con-
gestion may become so severe that the traffic comes to a complete standstill
and can only flow again after some vehicles have backed up.
• Network and control congestion. The efforts of traffic engineers and man-
agers (by the introduction of different traffic control devices) may reduce
congestion costs at certain times of the day, or, for example, in the case of
bus lanes, for specific types of traffic, but increase them at other times or for
other modes. This results from the general bluntness of most traffic control
schemes, which may help solve major problems but do, at times, create other,
albeit usually less significant, difficulties. This type of congestion was not
fully appreciated until the mid-1970s and had earlier led to excessively high
estimates of urban congestion costs. Previously it was assumed that conges-
tion tended to be of the simple or multiple interactive kind. But, as succinctly
expressed in a United Kingdom 1975 policy document, ‘Transport policy: a
consultation document’, ‘[o]nce account is taken of the limitations placed
upon urban traffic speeds by factors such as the incidence of traffic lights and
the multi-purpose nature of urban road networks, traffic speeds associated
with even very low levels of congestion can be expected to be quite low –
almost certainly below 20 mph in central areas’.

In addition to these five types of traffic congestion that can arise when
the infrastructure is fixed, Vickrey also points to the more general problem of
transport congestion in the economy. In the context of urban areas, roads in the
United States take up 30 percent or more of the land area of city centers, while

BUTTON 9781786435668 PRINT.indd 228 27/04/2022 11:07

 229
THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­

in Western Europe the figure is between 15 percent and 20 percent and in Third
World countries about 10 percent. The question then becomes one of whether in
the long term the general welfare of urban society is being excessively reduced by
too much transport infrastructure congesting city centers. The acceptance of this
view makes it rather difficult to define meaningfully optimal levels of transport
provision in the traditional welfare sense.
A further problem is that many travelers, and especially road users, have a
very poor perception of their own private costs. Indeed, in the case of car users
the perceived cost of many trips may only embrace the time involved. In such
cases the perceived AC, while reflecting some of the costs to a motorist thinking
of joining a traffic stream, is an inadequate basis for calculating the MC curve
which embraces the congestion costs to other road users. The appropriate policy
curve in these circumstances is MC*, which is based upon the resource costs of
making trips rather than just the perception of the additional user. As we see
in Figure 6.11, the implication of this is that congestion may well be somewhat
higher than is sometimes estimated.
Congestion, or to be more exact excessive congestion, has been shown to
imply a ‘dead-weight’ welfare loss and to reduce the economic efficiency of any
transport system. In recent years there has been some debate, however, about
whether this welfare loss is compensated by other beneficial effects of congestion
that are not immediately apparent in the standard, static, marginal cost type of
analysis. These arguments tend to follow three broad lines: those focusing on
issues centering on the distributional effect of congestion on different groups
in society, those concerned with more straightforward efficiency problems, and
those that take other forms of cost into account.
The main costs imposed by traffic congestion are usually found to be time
costs (although there may also be fuel and other components of generalized costs

Generalized costs
MC*
MC
AC
A

C
B

Demand

0 F* FO FA Traffic flow

Figure 6.11 The problems of misperceived private costs

BUTTON 9781786435668 PRINT.indd 229 27/04/2022 11:07

 230 TRANSPORT ECONOMICS, 4TH EDITION

to be considered). Queuing up for the use of a transport facility and slowing

down its consumption takes up the user’s time. Measures to reduce the demand,
increase the supply, or the introduction of market prices to optimize congestion
(all of which are discussed later in the book) impose some form of either financial
or welfare loss which, following simple efficiency criteria, must be lower than the
congestion costs saved but still must be borne by someone.
Those who favor the retention of a high level of congestion as a method of
allocating scarce transport facilities argue that, since in the short term time is
evenly distributed to everyone – that is, there are 24 hours in every person’s day –
it is a more equitable method of allocation than many alternative techniques. If a
traveler really wants to make a journey, he or she will be willing (and able) to wait,
whereas if a high congestion-deterring charge is levied, his/her financial budget
constraint may make it impossible to make the trip.
While there seem to be some grounds for this type of argument if one accepts
that transport is unique in requiring a substantial time input for its consump-
tion (a proposition that is far from self-evident), in the longer term the wider
distributional issue is probably more effectively tackled by direct income redis-
tribution measures. There seems no reason, in the general case, for singling out
transport rather than several other economic activities for this special treatment.
In addition, even when goods have in the past been provided free of charge, there
is empirical evidence that, despite the equal distribution of time, it is the rich
who tend to obtain them and, ipso facto, a disproportionate share of the benefit
(Barzel, 1974).
Moving to the second mitigating argument in favor of allocation by conges-
tion we turn to efficiency considerations. Congestion is seen by some as a comple-
mentary method of allocating certain types of facility, supplementing rather than
competing with other, usually monetary price mechanisms. The dead-weight loss
associated with congestion may in some situations, it is claimed, be outweighed
by other forms of welfare benefit. In some instances, people – for example, those
on aircraft stacking at congested airports – use time spent in queuing produc-
tively, while in other instances the dead-weight loss associated with suboptimally
excessive congestion may be exceeded by administrative or other costs of achiev-
ing optimal utilization of the transport facility. As we see in Chapter 8 this has
been one argument used against introduction of sophisticated metering devices
for urban road pricing.
More generally, it is argued that, since transport users are far from homog-
enous, different groups of users will value time differently and hence a system
with both time-allocated and financially allocated facilities could well be optimal.
If analogies are made with other forms of economic activity, from retailing to car
manufacturing, then both money and time are used for allocation. For example,
one can get fast, personal service at a small local store, but prices are likely to be
higher than at a large, possibly distant, supermarket where queuing is normal at
checkouts.
This sort of approach is in general use for some forms of transport, with
many countries, for instance, having fast, tolled motorways running parallel to

BUTTON 9781786435668 PRINT.indd 230 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 231

slow, free trunk roads. Also, one can often choose between expensive, readily
available air services or cheap standby facilities that often involve queuing or
waiting for a flight. With a given distribution of income, this increased choice
necessarily increases welfare that may, in turn, off-set any, or at least part of, the
dead-weight loss incurred on congested parts of the system. Essentially there is
product differentiation taking place in response to variations in the opportunity
cost of time among consumers.
The difficulty with this argument is that in many cases physical factors
make it impossible to provide different types of transport service. In other cases,
economies of scale are sufficient to make the provision of alternatives excessively
wasteful. One approach, favored by theoreticians, may be to decide upon the
optimal flow, and only let that flow on to the road or facility at any one time,
leaving a queue of potential users waiting. The optimal flow in this sense being
such that the length of the queue of traffic wishing to use the road would make
the opportunity time cost of waiting equal to the money price at which the
traffic flow is optimal. It is difficult to see how this could be put into practice on
urban roads, although it may be appropriate for making optimal use of facilities
such as bridges or ferries where queuing is practicable. The information costs of
estimating optimal queue lengths may also prove an insurmountable practical
problem.
Finally, a high level of congestion may itself be optimal (even with the dead-
weight losses it imposes and where neither of the former lines of argument are
applicable) when other forms of cost are also considered. It may be, for example,
that the transaction costs of moving from an over-congested to an optimally
congested situation exceed the conventionally defined benefits of eliminating a
dead-weight loss. The transition costs involved in removing an externality such
as excessive congestion are of three broad types: the cost per unit of reducing the
externality, initial lump-sum costs of organization, and information/enforcement
costs of carrying the action through.
To remove excessive congestion would, in virtually all cases, involve costs
in one or more of these categories and it could well be that in many cases such
transaction costs could be very high. A related point is that the actual reduction
of congestion to the optimal level for transport users may mean spreading other
forms of external cost (generally noise and air pollution) to a much wider group
of non-users in the community. Raising landing fees at over-used major airports,
for example, is likely to divert traffic elsewhere and place environmental costs on
people living near other, formerly under-utilized airports.
Congestion may in these circumstances, where the demand for transport
concentrates the incidence of environmental costs on a relatively small group in
the community, be felt to offer a more acceptable use of transport infrastructure
than if congestion is reduced but this results in demand being spread geographi-
cally. This is more likely if the initial congestion is concentrated in relatively
insensitive areas, but its reduction would increase the environmental nuisance
experienced in residential or other sensitive locations.

BUTTON 9781786435668 PRINT.indd 231 27/04/2022 11:07

 232 TRANSPORT ECONOMICS, 4TH EDITION

6.10 Some Broad Aggregate Calculations

The discussion so far has focused on individual external costs and provided, in
places, some estimates of their possible magnitudes. Here we provide just a few
examples of larger studies that have sought to place monetary values on a range
of transport-related externalities. The problems with most of these studies are
that they largely rely on secondary sources and there is, in some cases, a lack of
consistency in the way various effects have been measured and evaluated. Second,
placing monetary values on individual externalities at the micro, case-study level
is difficult because such partial-equilibrium work assumes that ‘other things
remain constant’, and most especially that income remains constant and that
the prices of other goods remain constant. Clearly, at the macro level, any effort
to estimate the willingness to pay for one externality reduction will reduce the
income available to pay for the optimization of another. Equally, reducing, say,
noise nuisance to an optimal level will affect the price of noise and thus make it
difficult to evaluate the willingness to pay for, say, increased safety. Unless these
factors are embraced in the macro-level calculations, there is an inherent upward
bias in the valuation of external transport costs.
In addition, there are often important correlations in terms of the impact
of various external factors. Reducing congestion costs, for instance, not only
allows transport infrastructure to be utilized more efficiently by its users but
often also reduces environmental costs because, for example, automobiles are
not continually stopping and starting, or aircraft do not have to circle so much
before landing. The correlations are not always positive. The fitting of a catalytic
converter, while reducing NOX emissions, increases fuel consumption and thus
CO2 emissions; and while smoother tires produce less road noise, they have less
traction and can lead to more accidents.
Notwithstanding these problems, Emile Quinet (1994) first tried to use the
best information available to give a general, minimum estimate of the monetary
costs of the damage transport imposes on the environments of industrialized
countries. His figures are, however, conservative because they cover only some of
the damage done by transport, and because only lower estimates for each social
cost considered are used in the calculations.
Table 6.15 provides another assessment of the external costs (including con-
gestion) that have been specifically associated with the use of automobiles in the
United States. Table 6.16 is a table at the meso level looking at estimates of some
of the relative external costs of car use in a European city. It is not, however, a
complete list of atmospheric gas releases.
The analyses show, as is typical, that the costs of congestion per mile exceed
those of pollution and other environmental effects, although climate change
effects are excluded. It should be remembered, however, that congestion is only
external to those using transport, while the environmental costs are external to the
transport system. The data suggest, however, that the largest social gains come
from removing the imperfections within the transport system, rather than from
removing the costs that are external to it.

BUTTON 9781786435668 PRINT.indd 232 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 233

Table 6.15 Estimated externality costs of automobile use in the United States

Cents per gallon Cents per mile

Central values for marginal external costs

Fuel-related costs
  Greenhouse gases 6 0.3
  Oil dependency 12 0.6
Total 18 0.9

Mileage-related costs
  Local pollution 42 2.0
  Congestion 105 5.0
  Accidents 63 3.0
Total 210 10.0

Source: Parry et al. (2007).

Table 6.16 External costs of urban car use in Brussels (€ per vehicle mile)

Gasoline Diesel

Peak Off-peak Peak Off-peak

Congestion 1.856 0.003 1.856 0.003

Air pollution 0.004 0.004 0.042 0.026
Accidents 0.033 0.033 0.033 0.033
Noise 0.002 0.008 0.002 0.008
Total 1.895 0.047 1.932 0.068

Source: Proost and Van Dender (2001).

References

Anciaes, P.R., Jones, P., and Mindell, A.S. (2016) Community severance: where is it found
and at what cost? Transport Reviews, 36, 293–317.
Araújo Alves, J., Neto Paiva, F., Silva, L.T., and Remoaldo, P. (2020) Low-frequency noise
and its main effects on human health: a review of the literature between 2016 and 2019,
Applied Sciences, 10, 5205, https://doi.org/10.3390/app10155205.
Barzel, Y. (1974) A theory of rationing by waiting, Journal of Law and Economics, 17,
73–9.
Bonnafus, A. (1994) Summary and conclusions, in Organization for Economic Cooperation
and Development, Internalizing the Social Costs of Transport, Report, OECD.
Chestnut, L. and Mills, D.M. (2005) A fresh look at the benefits and costs of the US Acid
Rain Program, Journal of Environmental Management, 77, 252–66.
China Energy Research Society (2002) Energy policy research 2002.1, Report.
Cohen, M.J. (1995) Technological disasters and natural resource damage assessment: an
evaluation of the Exxon Valdez oil spill, Land Economics, 71, 65–82.
European Commission (2006) European energy and transport trends to 2030: update
2005, Report, Office for Official Publications of the European Communities.
Evans, A.W. (1992) Road congestion: the diagrammatic analysis, Journal of Political
Economy, 100, 211–17.

BUTTON 9781786435668 PRINT.indd 233 27/04/2022 11:07

 234 TRANSPORT ECONOMICS, 4TH EDITION

Frenking, H. (1988) Exchange of information on noise abatement policies: case study on

Germany, Report, OECD.
Gillen, D. (2003) The economics of noise, in K.J. Button and D. Hensher (eds), Handbook
of Transport and the Environment, Pergamon.
Handy, S. (2003) Amenity and severance, in K.J. Button and D. Hensher (eds), Handbook
of Transport and the Environment, Pergamon.
Haynes, K.E. (2004) Transportation location and environmental justice: a US perspec-
tive, in D.A. Hensher, K.J. Button, K.E. Haynes, and P.R. Stopher (eds), Handbook of
Transport Geography and Spatial Systems, Elsevier.
Holland, S.P., Mansur, E.T., and Yates, A.J. (2016) Are there environmental benefits from
driving electric vehicles? The importance of local factors, American Economic Review,
106, 3700–29.
Holland, S.P., Mansur, E.T., and Yates, A.J. (2021) The electric vehicle transition and the
economics of banning gasoline vehicles, American Economic Journal: Economic Policy,
13, 316–44.
IEA (International Energy Agency) (2000) CO2 emissions from fuel combustion
­1971–1998, Report, IEA.
IPCC (Intergovernmental Panel on Climate Change) (1995) IPCC second assessment syn-
thesis of scientific–technical information relevant to interpreting Article 2 of the UN
Framework Convention on Climate Change, Report, IPCC.
Jacobsen, M.R. and van Benthem, A.A. (2015) Vehicle scrappage and gasoline policy,
American Economic Review, 105, 1312–38.
Johnson, K. and Button, K.J. (1997) Benefit transfers: are they a satisfactory input to
benefit cost analysis? Transportation Research Part D, 2, 223–31.
Jones-Lee, M.W. (1990) The value of traffic safety, Oxford Review of Economic Policy, 60,
39–60.
Jones-Lee, M.W. and Loomes, G. (2003) Valuation of safety, in K.J. Button and D. Hensher
(eds), Handbook of Transport and the Environment, Pergamon.
Martin, D.J. (1978) Low frequency traffic noise and building vibration, Transport Research
Laboratory Supplementary Report, SR.429.
McCubbin, D.R. and Dellucci, M.A. (1999) The health costs of motor-related air pollu-
tion, Journal of Air Transport and Policy, 33, 253–87.
Nelson, J.P. (1979) Airport noise, location rent, and the market for residential amenities,
Journal of Environmental Economics and Management, 6, 320–31.
Nelson, J.P (2004) Meta-analysis of airport noise and hedonic property values: problems
and prospects, Journal of Transport Economics and Policy, 38, 1–28.
Neutze, G.M. (1963) The external diseconomies of growth in traffic, Economic Record, 39,
332–45.
Newbery, D.M. (1988) Road damage externalities and road user charges, Econometrica,
56, 295–316.
Newbery, D.M (1990) Pricing and congestion: economic principles relevant to road
pricing, Oxford Review of Economic Policy, 6, 22–38.
OECD (Organisation for Economic Cooperation and Development) (1991) Fighting
Noise, OECD.
Parry, I.W.H., Walls, M., and Harrington, W. (2007) Automobile externalities and policies,
Journal of Economic Literature, 45, 373–99.
Pearce, D.W. and Turner, R.K. (1990) Economics of Natural Resources and the Environment,
Harvester Wheatsheaf.
Proost, S. and Van Dender, K. (2001) The welfare impacts of alternative policies to address
atmospheric pollution in urban road transport, Regional Science and Urban Economics,
31, 383–411.

BUTTON 9781786435668 PRINT.indd 234 27/04/2022 11:07

 THE EXTERNAL ECONOMIC COSTS OF TRANSPORT ­ 235

Quinet, E. (1994) The social costs of transport: evaluation and links with internalisa-
tion policies, in Internalizing the Social Costs of Transport, European Conference of
Ministers of Transport.
Rothenberg, J. (1970) The economics of congestion and pollution: an integrated view,
American Economic Review, Papers and Proceedings, 60, 114–21.
Smith, L.V. (1982) Microeconomic systems as an experimental science, American Economic
Review, 72, 923–55.
Smith, V.K. (1989) Taking stock of progress with travel cost recreation demand methods:
theory and implementation, Marine Resource Economics, 6, 279–330.
Spash, C.L. (2008) The contingent valuation method: retrospect and prospect, Environment
and Planning C, 26, 34–53.
Stern, N. (2007) Stern review: the economics of climate change, Report, Cabinet Office –
HM Treasury.
Sweet, M. (2014) Traffic congestion’s economic impacts: evidence from US metropolitan
regions, Urban Studies, 51, 2088–110.
Talley, W.K. (2001) Oil spillage and damage costs: US inland water tank barge accidents,
International Journal of Maritime Economics, 2, 217–34.
Texas Transportation Institute (2019) Annual urban mobility report, TTI.
Thomson, J.M. (1977) Great Cities and their Traffic, Gollancz.
Verhoef, E.T. (2005) Speed–flow relations and cost functions for congested traffic: theory
and empirical analysis, Transportation Research Part A, 39, 792–812.
Vickrey, W. (1969) Congestion theory and transport investment, American Economic
Review: Papers and Proceedings, 59, 251–60.
Weisbrod, G., Vary, D., and Treyz, G. (2001) Economic implications of congestion,
NCHRP Report 463, Transportation Research Board.
Whiffen, A.C. and Leonard, D.P. (1971) A survey of traffic induced vibrations, Road
Research Laboratory Report, LR418.
Winston, C. and Langer, A. (2004) The Effect of Government Highway Spending on Road
Users’ Congestion Costs, Brookings Institution.
World Commission on Environment and Development (1987) Our Common Future,
Oxford University Press.

BUTTON 9781786435668 PRINT.indd 235 27/04/2022 11:07

 7 The Pricing of Transport

7.1 The Principles of Pricing

Pricing is a method of resource allocation. There is no such thing as the ‘right’

price, but rather there are optimal pricing strategies that permit specified goals to
be obtained. The optimal price, for example, to achieve profit maximization may
differ from that needed to maximize welfare or ensure the highest sales revenue.
In some cases, there are no attempts to devise a price to maximize or minimize
anything, but rather prices are set that permit lower-level objectives (for example,
acceptable levels of security or market share) to be attained.
Often linked to this is the matter of who is setting the price. Prices may be set
to achieve certain objectives for transport suppliers in terms of their welfare (this
is normally the case with private enterprise transport undertakings) while in other
cases prices may be set to improve the welfare of consumers (as has been the case
with some publicly owned transport undertakings). The distinction here can be
a fine one and many undertakings consider that the employment of the pricing
mechanisms to achieve their objectives is automatically to the benefit of customers.
One of the major problems in discussing pricing policies in practice is to
decide what exactly the objective of pricing is. A good example is port pricing,
where there has been a traditional blurring between the ‘European’ doctrine of
setting prices to facilitate the economic growth of a port’s hinterland and the
‘Anglo-Saxon’ approach that attempts to ensure that ports cover their costs
and, where possible, make a profit with respect to the effects on the wider local
economy (Bennathan and Walters, 1979). In providing local transit services, there
are often conflicts between setting prices that allow use by lower-income groups
and pricing so that the system does not get excessively congested at peak, rush-
hour travel times.
When the private sector sets the price, much depends on the market power
exercised by the supplies of the transport system and those using it, and this in
turn is affected by the prevailing market structure. If the supplier operates within
a highly competitive market, then there is no opportunity for the transport
company to control the price; the market sets it. If, in contrast, and at the other
extreme, there is a single monopoly supplier and numerous customers, then the
transport undertaking can dictate prices to a large extent, but always must be
cognizant of losing business and net revenue, by setting them too high. In inter-
mediate cases, the various actors involved play ‘games’ with each other, jockeying
to gain the most benefit, be it consumer surplus or profit.
The nature of market power and the ways in which it may be exercised
depends to a considerable extent on the nature of the market itself. There are

BUTTON 9781786435668 PRINT.indd 236 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­237

Table 7.1 
Features of various market structures

Feature Perfect competition Perfectly competitive Monopoly

Profit maximization Yes Yes Normally

Barriers to entry/exit No No Yes
Perfect mobility of inputs Yes Yes No
Ubiquitous information Yes Yes No
Large number of firms Yes Maybe No
Homogeneous service Yes Maybe Yes
Firms confronted by
the cost function Yes Yes Yes
U-shaped AC functions Yes Maybe Maybe
Profits Normal Normal Monopoly rent

many different forms of market and it is beyond the scope of this volume to
explain all of them. Table 7.1, however, provides a very brief outline of the main
features of competitive, contestable, and monopoly markets; forms that are
often said to exist in transport. They are not the only ones, but the aim is to give
an indication of the various parameters that determine market behavior rather
than to be comprehensive.
This chapter looks at the appropriate pricing policies to adopt for trans-
port undertakings, considering a variety of objectives and when those involved
are confronted by different market conditions. In simple terms, in the transport
context it is about maximizing the benefits from transport given that there are
resource and other constraints on what can be provided. While the later sections
focus on criteria concerned with maximizing the social benefits of transport that
embrace the interests of both suppliers and uses, this section briefly reviews the
prices likely to exist in situations where transport enterprises are interested in
purely commercial criteria (defined here as the pursuit of their own self-interest,
which often, although not always, involves profits). The chapters that follow
consider, along with other things, the role of pricing of the environment and of
congestion.

7.2 Matching Supply with Demand

Traditional neoclassical economics assumes that profit maximization is the

motivation behind the activities of private enterprise undertakings. The actual
price level in this case depends upon the degree of competition in the market
and the costs of production. Where competition is considerable then no single
supplier has control over price and must charge that determined by the interac-
tion of supply and demand in the market (Adam Smith’s ‘invisible hand’). In
this ­perfectly competitive environment, it is impossible for any supplier to earn
super-normal profits in the long term because of the incentives such profits have
on new suppliers entering the market and increasing aggregate supply. Following

BUTTON 9781786435668 PRINT.indd 237 27/04/2022 11:07

 238 TRANSPORT ECONOMICS, 4TH EDITION

the same logic, economics tells us that in the long run price will be equated with
the marginal (and average) costs of each supplier.
In contrast, a true monopoly supplier has no fear of new entrants increas-
ing the aggregate supply of transport services and has the freedom either to set
the price or to stipulate the level of service they are prepared to offer. The effec-
tive constraint on the monopolist is the countervailing power of those demand-
ing the service that prevents the joint determination of both output and price.
However, given the absence of competition and the degree of freedom enjoyed
by the monopolist, it is almost certain that a profit-maximizing price will result
in charges above marginal and average cost (the only exception being the most
unlikely situation of a perfectly elastic market-demand curve). This is one reason
why governments have tended to regulate the railways, ports, and other transport
undertakings with monopoly characteristics, or to supply the transport services
themselves.
This simple description of textbook situations does, however, hide certain
peculiarities that may arise in some transport markets. Since the actual unit of
supply, the vehicle, is mobile it is possible for the overall transport market to
appear to be essentially competitive, but the individual suppliers to price as if
they were monopolists or, at least, exercise some monopoly power. The unregu-
lated urban taxi-cab market is an example of this. In Figure 7.1, DM is the market
demand for taxi-cab ‘rides’ per hour in a market supplied solely by cruising taxi-
cabs. The cost of taxi cruising activities is almost constant irrespective of whether
a fare is carried or not, and to stay in business the cab operator must charge fares
which permit such costs to be recovered.
In the figure, the iso-profit curve for a single operator indicates combina-
tions of fare and ridership that allow a normal profit to be earned. It is con-
strained to a minimum fare (P2) by the physical impossibility of carrying more
than R2 passengers an hour. Also, it is unlikely that a fare above P2 would ever be
feasible; potential users would simply not accept it. For the overall market, fares
must exceed P1 if taxi-cab services are to be offered, but is not the true long-term
floor level of fares. Because potential customers are seldom positioned exactly

$
‘Useful’ rides offered Rides offered

P2
P0
Iso-profit curve
R3

P1
DM

0 R2 R1 R0 R3 Rides
per hour

Figure 7.1 Taxi-cab fare determination

BUTTON 9781786435668 PRINT.indd 238 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 239

where empty cabs are cruising, there must be an excess of rides offered above
total demand if enough rides are to be supplied; this is indicated by the ‘useful’
rides curve.
This lack of synchronization means, in effect, that the ‘rides-offered curve’
in Figure 7.1 is not a true supply curve since it is dependent upon demand
conditions. At higher fares, the rides offered will increase but even at the inter-
section with the total demand curve (with R3 rides offered) there will still be
unsatisfied demand (Shreiber, 1975), that is, the number of taxi rides taken is
less than the number demanded. This is because the taxis may not be at the same
location as potential customers. Only if cabs were always exactly where they were
wanted would demand always be satisfied. The demand will, in normal circum-
stances, only be fully satisfied at a price above the intersection – say P0 – because
at this and higher prices the ratio between rides demanded and the number
offered will correspond to the rate of occupancy. This is so because the number
demanded is then assumed equal to the number of rides taken, and there are no
frustrated passengers who give up waiting because they are unable to obtain a
ride.
The actual fare level may be set at any point above P1 but below P2, hence
the apparently perfect taxi-cab market does not have a unique price. However,
there are reasons to suspect that the final price will be nearer P1 than P2, thus
permitting the earning of super-normal profit by the cab operators. It also means
that those who still wish to pay the fare and use taxi-cab services will have a
good service provided for them – the rides offered being well in excess of those
demanded – although the short waiting time and abundance of capacity is likely
to be wasteful in resource utilization.
The tendency towards high fares is caused by the monopoly power enjoyed
by any individual taxi at the point of hire. Unlike normal perfect markets, individ-
ual suppliers are not normally confronted with perfectly elastic demand schedules
for their services but when hailed by a potential customer are virtual monopolists
able to charge a high fare for their services. People seldom turn away a cab, upon
hearing the fare, to hail another one – the low probability of a lower cab fare does
not justify it: once fares are at the higher level there is, therefore, no incentive for
individual cabs to cut their fares, because to customers they all appear alike and
no additional business is attracted (that is, revenue for any cab acting differently
will inevitably fall).
Of course, the cab market is somewhat more complicated than the simple
model suggests (there are, for instance, cab ranks, and it may be possible to dif-
ferentiate cabs by color schemes, etc.), but the fear that cabs could exploit local
monopoly power of the type described and keep fares suboptimally high is one
reason why authorities in most major cities control fare levels. While this may be
justified, it is hard to see why at the same time most cities outside of the United
Kingdom regulate the number of taxi-cabs operating within their domain; if fares
are deemed optimal at P0 then rides offered will automatically adjust to R0 and
there is no need for official regulation of capacity which can seriously distort the
market (Beesley, 1973).

BUTTON 9781786435668 PRINT.indd 239 27/04/2022 11:07

 240 TRANSPORT ECONOMICS, 4TH EDITION

While it is possible that the simple picture of perfectly competitive price

determination is often complicated in the transport sector, it is equally true
that the basic model of monopoly also on occasion needs modification. There
are few if any natural monopolies in transport: there are normally competitive
modes even if the one in question tends to be monopolistic in character. Users
of transport services often have the alternative of either changing their method
of production (in the case of freight transport) or pattern of consumption (with
passenger modes) so that transport is itself competitive with different forms of
human activity. In some cases where these countervailing forces are weak or the
introduction of competition would mean wasteful duplication of services, gov-
ernment may institutionalize a monopoly but by controlling price and other com-
mercial aspects of its operations prevent the exploitation of customers. Much of
the regulation initiated in a wide range of countries in the l930s and 1940s, which
is discussed in Chapter 14, was ostensibly introduced for this sort of reason.

Exhibit   The issue of predatory pricing

The US Supreme Court in 1986 defined predatory pricing as ‘pricing below an appropriate
measure of cost for the purpose of eliminating competitors in the short term and reducing
competition in the long run’. The Court’s criteria for proving predation were based on
market power, predatory conduct, and recoupment. (Recoupment being the ability of an
incumbent supplier to recover revenues lost during the period prices are reduced, to force
out a competitor.)
In practice, few cases of predation have been proved in transport markets. Large
incumbents may set prices that offer a reasonable long-term profit but which are not large
enough to stimulate new suppliers who will encounter entry costs: ‘limit pricing’. If there
are no market-entry and -exit costs as in contestable markets, then fear of hit-and-run
competition will automatically exclude incumbents making abnormal profits. But there may
also be other ways of deterring market entry.
The traditional approach to predatory behavior, with its foundation on microeconomic
theory, was questioned by the US Department of Transportation (US DOT) in 1971. In its
place the DOT favored a more structure-conduct-performance-based approach, looking
at the actual behavior of suppliers over networks rather than simply the existence of
monopoly power on any individual route: a wider concept of a ‘domain of power’. It also
recognized in the airline case that carriers can deter competitors through the use of long-
term contracts with suppliers, control over infrastructure (for example, of slots and gates),
and the benefit of greater financial resources.
To examine the different views, Ashutosh Dixit et al. (2006) studied airline concentrations
at the 50 largest US domestic airports between 1991 and 1999, including 17 major hubs
and two slot-controlled airports, and, in addition, looked at the specific behavior of a major
airline.
One of the issues addressed was the response of incumbents to new market entry and
also their re-entry. Fifteen routes of a major airline were selected when there were

BUTTON 9781786435668 PRINT.indd 240 27/04/2022 11:07

 241
THE PRICING OF TRANSPORT ­

market entries by discount carriers, and 15 when there was none. Comparisons of these
groups showed the major airline increasing average fares by 15 percent between 1994 and
1999 on routes where there was no discount airline competition. Of the routes where
there was competitive entry (ten in 1993 and five between 1997 and 1998), fares were
reduced.
Considering the markets where there has been a competitor (or competitors) entering the
market but which have subsequently left it (see figure), the dominant carrier initially cut
its fares but then increased them to the same or higher levels than before the competitive
entry. There are also two cases of entry and re-entry by discount carriers and in both cases
the incumbent responded with fare cuts, subsequent fare rises, and then cuts again when
competitive re-entry occurred. This is all in line with the Supreme Court criteria.

240
Major airlines’ price ($)

220
200
180
160
140
120
100
93 94 95 96 97 98 99
Years

See also: A. Dixit, G.T. Gundlach, and F. Allvine (2006) Aggressive and predatory pricing:
insights and empirical examination in the airline industry, Journal of Public Policy and Marketing,
25, 172–87.

The fear of potential competition, especially in the long term, tends to lead to
the regulation of the activities of quasi-monopoly transport suppliers even when
government intervention is minimal regarding other transport modes. Table 7.2,
for example, offers information on some of the regulations imposed on taxi-cab
services.
The pricing policies pursued by liner conferences, when shipping companies
combine to monopolize scheduled maritime services between major ports, offer
another illustration of this problem. Sturmey (1975) argues that conferences do
not price to maximize immediate profits but rather to maximize the present value
of the flow of revenue from the market. More discussion of detailed pricing of
consignments by conferences is contained in Section 7.5, while here we focus on
the relevant general principles.
The emphasis on revenue reflects the concern with market size, while that on
the present value shows that long-term objectives dominate short-run considera-
tions. If, in Figure 7.2, the intention was to maximize profit in each market, then
price would be set, assuming the conference enjoyed a short-term monopoly posi-
tion, at PM with a monthly output of QM. If sales-revenue maximization (subject

BUTTON 9781786435668 PRINT.indd 241 27/04/2022 11:07

 242 TRANSPORT ECONOMICS, 4TH EDITION

Table 7.2 
Examples of taxi-cab regulation in the early 2000s

Country – city Fare regulation Entry regulation Period Restrictions

Belgium Yes Yes 5–10 years 1 vehicle per 1,000 inhabitants/

personal & non-transferable
Czechia Yes No – Non-transferable
Denmark Yes Yes 10 years Non-transferable
France – Paris Yes Yes – 100 new licenses per year
Germany Yes No 5 years License quota & fee
Hungary Yes – – –
Ireland Yes No (2000) – License quota & fee
Italy Yes Yes – 4.5 per 10,000 inhabitants/1
license per person
Japan Yes No (2002) – –
Korea Yes Yes – –
Netherlands Yes (2004) No (2002) – –
Norway City dependent Yes – Neither tradeable nor
transferable
Sweden – No (1990) – –
Switzerland City dependent Yes 3 years Neither tradeable nor
transferable
US – Seattle No (1979) No (1979) – –
Romania Yes Yes – 4 vehicles per 1,000
inhabitants

Source: Salanovaa (2011).

$
MC
AC
PM

PNVP

PR

AR

MR

0 QM QNVP QR Output
per month

Figure 7.2 Pricing of shipping conference services

to cost recovery) is the objective then price PR is charged (expanding output to

where the MC curve hits the horizontal would violate the break-even constraint).
In practice, however, Sturmey argues that conference rates are found to be
below PM because the high short-term profits would encourage competition to

BUTTON 9781786435668 PRINT.indd 242 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 243

enter the market; they are also unlikely to equal PR because the conference looks
beyond the immediate period although there is no a priori method of telling
whether they will be above or below this level. The conference is likely to base
its pricing policy on a relatively long time horizon, hypothesized by Sturmey to
be the period over which the scale of productive enterprise is unchanged, but
long enough to allow for additional capital equipment, which duplicates exist-
ing equipment, to be installed – although not long enough for all factors to be
considered truly variable. The net revenue over this period, discounted to yield its
current worth, is then seen as the key variable to maximize. The conference rate
is, therefore, likely to be, say, at PNPV in Figure 7.2 at which the maximum present
value (the value of the flow of net revenue as perceived today) is obtained without
attracting new entrants.

7.3 Marginal Cost Pricing

As was pointed out in the previous section, the pricing policy adopted by any
transport undertaking depends upon its basic objectives. The traditional, classi-
cal economic assumption is that firms price so that profits are maximized. More
recent variations on the theory of the firm suggests that many undertakings adopt
prices that maximize sales revenues when in an expansive phase, or simply price to
ensure that certain satisfactory levels of profit, security, market domination, etc.,
are achieved when a defensive stance is adopted – ‘satisficing behavior’, to adopt
Herbert Simon’s term.
The sales-revenue-maximizing ideas of William Baumol (1962) illustrate the
sorts of deviation that this implies from conventional profit-maximizing ideas.
In Figure 7.3 the total cost and revenue curves associated with different levels
of output are depicted. The profit-maximizing business will produce an output
of QΠ, but the sales-revenue maximizer will continue producing to the point
QR, where total revenue is highest subject to cost recovery. Costs in this case
may be viewed as embracing a reasonable return for the owners of the transport
undertaking.
Whatever the underlying operational objective, the theory of the firm
assumes that the supplier is intent on maximizing his or her own welfare, be this
defined in terms of profits or higher-level objectives.
Welfare economics takes a rather wider view of pricing, looking upon it as
a method of resource allocation that maximizes social welfare rather than simply
the welfare of the supplier. In some cases, since the good or service is provided
by a public agency, this may be equated with maximizing the suppliers’ welfare.
In other instances, controls or incentives may be applied to private companies
so that their pricing policy is modified to maximize social rather than private
welfare. This may take the form of restrictions on pricing flexibility, or the taxing
and subsidizing of firms so that their prices are socially optimal.
Social optimality has a wide variety of meanings but in broad terms it means
maximizing the joint net social surplus – that is, the total revenue (TR) plus

BUTTON 9781786435668 PRINT.indd 243 27/04/2022 11:07

 244 TRANSPORT ECONOMICS, 4TH EDITION

$
TC

TR

Maximum

0 Q QR Output

Figure 7.3 Sales-revenue-maximizing output

consumers’ surplus (CS) generated by an undertaking minus the total cost (TC).
We can, therefore, define the objective of public policy as the maximization of:

SW = TR + CS – TC (7.1)

Figure 7.4 takes the example of charging for rail freight capacity as an illustration
of how the optimal price is arrived at. For expositional ease, assume that there
are constant costs and that the railway undertaking is a monopoly. If it seeks
to maximize its profits it will charge PM, which in terms of equation (7.1) will
produce total revenue of PMbQM0, consumer surplus of abPM, and total costs of
PMCeQM0, resulting in a social welfare level of abePMC. While this may yield the
maximum profit to the railway, it is not, however, the price that maximizes social
surplus. That price is the price at which marginal cost is equated with demand. At
this price, the total revenue is PMCdQMC0, consumer surplus is adPMC, and total
cost is PMCdQMC0, which gives a total social welfare exceeding that associated
with the profit-maximizing price by an amount bed.
In other words, social welfare is maximized when price is equated to mar-
ginal cost. What marginal cost pricing does, in effect, is to result in transport
services being provided up to the point where the benefit for the marginal unit is
equated with the costs of providing that unit. The policy is a well-established one
in economic theory, and, indeed, formed the basis for United Kingdom public
enterprise pricing from 1967.
Traditional theory also tells us that such a condition prevails in the long term
when perfect competition exists, even though each firm is attempting to maximize
its own profits. The ability to exercise any degree of monopoly power, however,
permits a firm to price above marginal cost so that it can achieve additional
profit at the expense of reduced output and at costs to the consumers. The price

BUTTON 9781786435668 PRINT.indd 244 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­245

$
a

PM b

e d
PMC MC

MR D = AR

0 QM QMC Tonnes

Figure 7.4 Marginal cost pricing

charged by a profit-maximizing monopolist will force some potential consum-

ers to forgo consumption despite their willingness to pay for the costs of their
activities. Indeed, it was the fear of monopoly exploitation that led to controls
being imposed on railway pricing in the late nineteenth century and has led to
the United States controlling rate fixing by shipping conferences operating from
its ports.
While it has been shown that adopting marginal cost pricing maximizes
social welfare, the exact definition of the appropriate marginal cost has been left
vague. More specifically, there is the question of whether long-run (LRMC) or
short-run marginal cost (SRMC) pricing is the more appropriate (short-run being
when there is a fixed capacity that can be modified only in the long run). SRMC
pricing has the advantage that it ensures existing capacity is used optimally but
does not take account of capital and other fixed cost items. Jack Wiseman (1957)
was concerned with this problem since, he argued, the shorter the time period
under consideration, the lower will the SRMC appear and, ipso facto, the price
charged to users. This concern is misguided, however, because if there is fixed
capacity, as is almost always the case with transport in the short term, a premium
should be added to SRMC as an effective rationing device to contain excessive
demand. Price is, after all, an allocative device.
Figure 7.5 shows the demand for a passenger railway service with capacity
Q1. The marginal cost of carrying each additional passenger is constant until
the capacity of the system is reached, whereupon the SRMC becomes infinite.
If a price of P1 is charged with SRMC1, then demand will exceed capacity by
Q*d – Q1. In these circumstances, where demand will exceed absolute capacity
using SRMC pricing policy, a mark-up to price level P*1 is appropriate to ration
the available seats. The extra revenue thus generated in excess of LRMC provides
an indication that it would be beneficial for the capacity of the railway service

BUTTON 9781786435668 PRINT.indd 245 27/04/2022 11:07

 246 TRANSPORT ECONOMICS, 4TH EDITION

SRMC1 SRMC2

P*1

LRMC

P1

0 Q1 Q2 Q*d Passengers

Figure 7.5 Short-run and long-run marginal cost pricing

to be expanded. The optimal scale of service will, in fact, be offering capacity

Q2, where the price charged to travelers is equated with LRMC (and the upturn
of the SRMC2 curve). We see, therefore, that the long-run optimum is where
P = LRMC = SRMC. In some cases (for example, airports or motorway systems),
indivisibilities may make it impossible to provide exactly the optimal capacity Q2,
and a choice must then be made between a suboptimally small system or a subop-
timally large one. In these circumstances, decisions must be based upon weighing
the full costs and benefits of the alternatives against one another.

7.4 Difficulties of ‘Second-best’ Situations

The preceding analysis contained several implicit, as well as the stated explicit,
assumptions. It assumed that all other prices in the economy are set equal to mar-
ginal cost. A variety of factors – some economic, others political or i­ nstitutional –
mean that all other prices in the economy are not equal to marginal cost. The
problem, again couched in terms of the railway example, then becomes one of
deciding whether marginal cost pricing is, in these circumstances, appropriate in
the railway context.
For simplicity we assume that there is only bus and rail transport in the
economy. Further, the bus sector is under monopoly control and fares are set
above the marginal costs of providing services. The issue is one of whether the
railways should marginal-cost price or adopt some alternative strategy that would
maximize social welfare. In Figure 7.6, we have the production possibility curve
for bus and rail services and, additionally, denote A as the traffic mix which
would result in maximum social welfare; it is tangential to the highest attainable
­indifference curve.

BUTTON 9781786435668 PRINT.indd 246 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­247

Bus
–(PBus /PRail)

B
A

U1

U2

0 Rail

Figure 7.6 Second-best problems

Extending the analysis of Figure 7.4, because this combination maximizes social
welfare, the two modes will be charging marginal costs at this point. Since,
however, we have said that bus fares are above marginal cost (PBus > MCBus)
the attainable position on the production possibility frontier with the actual
price ration if rail adopts marginal cost pricing (PRail = MCRail) is B, which is
on a lower utility curve, namely U2. The question then arises as to whether the
railways, by deviating from marginal cost pricing and adopting a ‘second-best’
pricing strategy, can enhance total social welfare.
The simplest approach to the second-best in these conditions, as estab-
lished by Lipsey and Lancaster (1956/57) is that rail should adopt fares which
deviate by the same proportion from marginal costs as do those of bus. In other
words, it is possible to attain social welfare level U1 by adopting prices that
conform to:
PRail − MCRail PBus − MCBus
= (7.2)
MCRail MCBus

What this effectively does is ensure that the two modes are comparable
in terms of their relative attractiveness. In practice the calculations are more
complex and Nilsson (1992) offers an illustration of how second-best prices could
be determined for Sweden’s freight rail services in conditions where road transport
is not paying its full marginal costs. A general problem is that here we are only
looking at transport, but while the second-best rule will ensure the social optimal
mix of transport use there may be problems with other prices in the economy. If
they all remain set at marginal cost but both rail and bus are priced along second-
best lines above marginal cost, then transport will be relatively expensive when
compared to other possible expenditures. Ideally, all prices should deviate by
appropriate percentages from their marginal costs in these circumstances.

BUTTON 9781786435668 PRINT.indd 247 27/04/2022 11:07

 248 TRANSPORT ECONOMICS, 4TH EDITION

Under some conditions the problem may not be serious and, from a purely
pragmatic stance, it may be more efficient to charge marginal cost prices than to
bear the costs of working out any optimal adjustments. In other cases, deviations
from marginal cost principles elsewhere in the economy may be so remote that
they have minimal influence on the demand for transport. Under such conditions,
and assuming the distortions cannot be removed, Davis and Whinston (1967)
demonstrate that piecemeal optimization within separate sectors of the economy
using marginal cost pricing is optimal. Ed Mishan (1962) suggests that since in
many cases people spend a fixed amount of their income upon transport, there is,
therefore, only a very low cross-elasticity of demand between transport as a whole
and other goods consumed in the economy. This situation means that the issue
can be reduced to optimizing the allocation of traffic between forms of transport,
on a piecemeal basis, rather than having to consider the allocation of expenditure
between a certain form of transport and all other goods. If, in our example, all
competing forms of transport apply marginal cost pricing principles, then these
should also be adopted by the railway service.
While Mishan’s empirical approach has a certain practical common-sense
appeal for some forms of transport – such as inter-urban passenger transport –
it has less applicability in the freight sector or in the context of international
travel. Freight costs have a considerable bearing upon both final prices charged
for products and the location of the manufacturing industry; these are also the
main reasons for the attempts at the macro-macro level to develop a Common
Transport Policy within the European Union and why transportation at borders
has been such an issue in the creation of the North American Free Trade
Agreement (NAFTA).
If all other inputs to industry are priced above marginal cost because, say,
of the monopoly power of suppliers, but transport is priced at marginal cost,
then this could lead to an over-development, from the national efficiency point
of view, of transport-intensive industry (although it is possible that relatively
‘cheap’ transport could break the monopoly power of the suppliers of other
inputs forcing them, in the long run, to price at marginal cost). International air
and sea transport has the complication that, except in certain well-defined areas,
many nations consciously subsidize their ‘flag bearers’, enabling them to charge
rates below LRMC and, on occasions, even below SRMC. Any single operator
charging fares based on marginal cost in this situation would find itself unable to
attract the optimal volume of traffic, and thus some deviation from the marginal
cost principle may be necessary.
The existence of monopoly and other distorting influences in the economy
has been shown to necessitate some variations to marginal cost pricing in certain
transport sectors. The key to the degree to which prices should deviate from mar-
ginal cost is clearly the sign and magnitude of the cross-elasticities of demand
between transport and other goods and services in the economy. The practical dif-
ficulty in many cases is not the derivation of the appropriate theoretical model but
rather our inadequate knowledge of the size of the cross-elasticities. The evidence
that is coming forward tends to be piecemeal. Additionally, most of the evidence

BUTTON 9781786435668 PRINT.indd 248 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­249

is only related to intra-transport cross-elasticities with extremely few estimates of

transport/other goods cross-elasticities.

7.5 Price Differentiation, Price Discrimination, and Yield

Management

There can be quite significant differences in the fares paid by those on a specific
plane or train in most cases, but also for cargo rates on ships. Some of this is
attributable to cost differences – things like superior service. But, in addition to
this, these different prices may be used as a rationing device to allocate out scarce
resources, limited capacity being available at the last minute requiring higher
prices to ration it amongst those seeking a late seat or berth. This is effectively
yield management in its traditional sense. There is also price discrimination,
whereby transport service suppliers seek to gain as much revenue as each indi-
vidual user is willing to pay above costs.
Second-best pricing and the like are forms of price discrimination: differ-
ent users of a transport service pay different prices not entirely related to their
attributed cost. So far we have discussed such discrimination largely in terms of
imperfections in other markets and the need to make adjustments to marginal
cost prices in the transport market of interest to reflect these external distortions.
The adoption of marginal cost pricing can, in certain circumstances, however,
also result in an undertaking making a financial loss even when other markets are
working correctly. It is this type of situation that we now address.

The Basic Idea of Price Discrimination with Decreasing Costs

The classic example of this is the decreasing cost industry where, because of
high initial capital costs, the setting of charges equal to SRMC will result in a
financial deficit. The railways are often cited as an example of an industry where
marginal cost pricing may ensure optimal utilization but leave the undertaking
with a financial deficit. In Figure 7.7 the railways are assumed to be a monopoly
supplier of freight services and, indeed, although it is not shown, if a monopoly
profit-maximizing price were adopted then abnormal profits could be earned.
The adoption of marginal cost pricing (PMC), however, with the downward-
sloping AC and MC curves, at least over the relevant range, will result in a loss
shown by the shaded area. A break-even situation could be attained by average
cost pricing (that is, charging PAC), but this would mean that QMC – QAC poten-
tial rail travelers, willing to pay the additional costs they impose, are priced off
the service.
In these circumstances the adoption of marginal cost pricing is essentially a
welfare decision and if the undertaking does make a financial loss this is attrib-
utable to the pricing policy pursued rather than the incapacity of the service to
be financially viable. The fixed costs of the service may be met in these cases by
subsidy or by operating a ‘club’ system with potential users paying a fixed sum for

BUTTON 9781786435668 PRINT.indd 249 27/04/2022 11:07

 250 TRANSPORT ECONOMICS, 4TH EDITION

the right to travel by rail and a mileage rate (or some other ‘cost’-related variable
fee) to reflect use. It could be argued that the fixed rates of road vehicle taxation
combined with fuel duties reflect a type of club arrangement, but, if so, the system
is extremely imperfect. In the United Kingdom, for example, at the very crudest
level of analysis, there is very considerable variation in the ratio of license fee to
fuel tax revenue that bears no relation to their relative expenditures on investment
and maintenance (Table 7.3). This is a picture repeated in virtually all countries.
Returning to Figure 7.7, if each user was charged a different price so that it
reflects his or her willingness to pay, and output was limited to the point where
MC equals demand (that is, QMC), this would yield a revenue of 0abQMC, which
may be compared with the total cost of providing the service, namely 0PMCbQMC.
Using this first-best discriminatory pricing approach, the costs are fully recovered
and a profit of PMCba is being earned. Of course, price discrimination does not
always guarantee full cost recovery, which depends on the revenue raised vis-à-vis
the costs involved. Further, one should also note that perfect price discrimination,
as described, results in the marginal cost level of output; in fact, it results in social
welfare maximization but with all the benefits being derived by the provider of
the transport service.
Figure 7.8 illustrates a situation, and one that does occur with certain forms
of transport service, where even first-best price discrimination will not produce
full cost recovery. Here, at no level of output does average revenue exceed average
cost. It is impossible in this type of situation for costs to be recovered by charging
a single price to all users even if monopoly-pricing policies are adopted. In this
case even a club arrangement is incapable of preventing the service from being
unprofitable. There may be justifications for keeping the service operating with
the losses financed through subsidies if there are wider benefits to be enjoyed
outside of those generally linked to transport; the service may, for instance, have

$
a

PAC

AC

b
PMC

MR
MC
Demand
0 QAC QMC Tonnes

Figure 7.7 Second-best average cost pricing with degreasing costs

BUTTON 9781786435668 PRINT.indd 250 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 251

Table 7.3 
Road costs and user payments in the United Kingdom in pence per vehicle-
kilometers (1998)

Cost or revenue category Fully allocated Marginal costs

costs
Low High Low High

Costs
Cost of capital for infrastructure 0.78 1.34 n.a. n.a.
Infrastructure operating costs & depreciation 0.75 0.97 0.42 0.54
Vehicle operating costs (PSV) 0.87 0.87 0.87 0.87
Congestion n.a. n.a. 9.71 11.16
Mohring effect (PSV) n.a. n.a –0.16 –0.16
External accident costs 0.06 0.78 0.82 1.40
Air pollution 0.34 1.70 0.34 1.70
Noise 0.24 0.78 0.02 0.78
Climate change 0.15 0.15 0.15 0.15
VAT not paid 0.15 0.15 0.15 0.15
   Subtotal of costs 3.34 7.20 12.32 17.05
Revenues
Fares (PSV) 0.84 0.08 0.84 0.84
Vehicle excise duty 1.10 1.10 0.14 0.14
Fuel duty 4.42 4.42 4.42 4.42
VAT on fuel duty 0.77 0.77 0.77 0.77
   Subtotal of revenues 7.14 7.14 6.17 6.17
Costs–revenues –3.79 0.07 6.15 10.88
Revenues/costs 2.13 0.99 0.50 0.36

Notes: Road sector costs exclude costs attributable to pedestrians, bicyclists, and motorcyclists;
vehicle excise duty at the margin relates to heavy goods vehicles (HGV) and public service vehicles
(PSV) such as buses.

n.a. = not available.

Source: Samson et al. (2001).

a strategic value or it may be deemed a ‘merit good’, a concept first suggested by

Richard Musgrave (1957) in the context of low-cost housing, which society ought
to provide to enable isolated communities to continue in existence.
In summary, in the more conventional case, if subsidies are to be avoided,
two-part tariffs are impractical, and average cost pricing is deemed inappropri-
ate, then price discrimination can still be used to finance the service. Essentially,
price discrimination involves charging ‘what the user will bear’ provided they
cover their marginal costs. In other words, it involves attempting to define the
maximum amount each consumer, or, more realistically, each identifiable group
of customers, is willing to pay for the service. The demand curve reflects ‘willing-
ness of users to pay’ (the AR curve becomes the de facto MR curve) when a trans-
port enterprise price-discriminates and a perfectly discriminating supplier will
charge down this curve. First-best price discrimination is however rare, and more
­pragmatic approaches to price discrimination are pursued in practice.

BUTTON 9781786435668 PRINT.indd 251 27/04/2022 11:07

 252 TRANSPORT ECONOMICS, 4TH EDITION

AC

PMC
Demand
MR MC

0 QMC Tonnes
Figure 7.8 Decreasing costs with demand always inside the average cost curve

Differentiation by Type of Customer/Commodity

In many cases a transport undertaking itself provides several services (for

example, first- and second-class rail services) and it also has a remit to make
a prescribed level of profit, or to break even, rather than to maximize profits.
Given these conditions, and assuming the cross-elasticity of demand for the
different service is negligible, Baumol and Bradford (1970) stated that the
price of any services should be set equal to its SRMC plus a mark-up inversely
­proportionate to the service’s price elasticity of demand (ε). Hence, where the
demand for a service is highly inelastic a substantial addition should be added to
marginal cost. Where the demand is perfectly elastic, SRMC pricing is applied.
In this way revenue SRMC can be obtained to meet the ­financial target without
­distorting the allocation of traffic between ­services.
If cross-elasticities are not zero, which is the common case, then the simple
rule must be modified to ensure that the relative quantities of goods sold respond
to the proportions that would occur if marginal cost pricing were applicable.
The rule, for example, for substitutes is that optimal prices should be derived so
that ‘all output be reduced by the same proportion from the quantities which
be demanded at prices equal to the corresponding marginal costs’. This is often
called ‘Ramsey Pricing’ after earlier work in the United Kingdom by Frank
Ramsey (1927).
Assuming the railways offer two products, express and commuter rail ser-
vices, then they should, following this principle, price such that:

(Pexp − MCexp )ε =
( Pcom − MCcom )
εcom (7.3)
exp
Pexp Pcom

BUTTON 9781786435668 PRINT.indd 252 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 253

It is not always possible, for practical reasons, to discriminate perfectly: the

administrative costs of operating the system may be too high or exact knowledge
of the pertinent demand curves unavailable. These, for example, are the reasons
cited in the 1960s by British Rail for their reluctance to adopt more sophisticated
costing and policies – the diversity of services provided made it impracticable.
In other cases, it may be felt socially undesirable to charge ‘what the user will
bear’ because of distributional consequences. Passengers with a lower elasticity
of demand may be from the poorer sections of the community and unable to
transfer to alternative modes of transport. It may still be justified, even in these
circumstances, to provide services even if not all costs are recovered by the opera-
tor, providing that the potential revenue if discrimination were adopted exceeds
the costs of the service. If this is the case, a subsidy is required to bring actual
revenue up to potential revenue and, to some extent, the British government
accepted this in the 1968 Transport Act when, for a period, it introduced specific
social subsidies to maintain several railway services.
While price discrimination is relatively uncommon outside of the transport
sector it is a familiar feature of pricing policy within it, and indeed has become
more common as information systems have developed and sophisticated elec-
tronic means of levying fares and freights have emerged. Truelove (1992) provides
an apt description of these pricing practices of British Rail in the 1990s before
privatization, illustrating that even then they were widely practiced.
The distinction between first and second class can be further refined by the
exclusion of discount fares from peak weekday and holiday trains, by the limited
availability of old people’s and students’ discounts, and the reduction in supple-
ment for travel in first-class carriages at weekends, when leisure travel is heavy and
business travel almost negligible.
Another well-documented example of third-degree price discrimination
has long been provided by shipping conferences, and now by shipping alliances.
These act as quasi-monopoly suppliers of regular liner services between major
ports, but they also often compete with non-conference liners and tramp ship-
ping that may enter markets when rates rise. Consequently, one can hypothesize
an inverted S-shaped demand curve for individual liner services over any route.
Given the high capital costs associated with shipping, there is adequate evidence
that, prior to the widespread use of containers, without price discrimination most
conference lines would have become unprofitable, and even with containers, and
the apparent homogeneity of the ‘boxes’ being moved, price discrimination is still
a necessary commercial practice.
A detailed look at the Australia–Europe Conference by Zerby and Conlon
(1978) produced a breakdown of rates that clearly shows the high degree of price
discrimination exercised by these maritime cartels. Figure 7.9 reproduces their
average revenues for each type of cargo carried in 1973–74 and when traced out
this corresponds closely to the type of demand curve hypothesized. The low-value
bulk cargoes (that is, ores and metals) are carried at (or sometimes below if ballast
is required) the average incremental costs of loading and unloading. While the
high-value products are carried at considerably higher rates, the tapering off in

BUTTON 9781786435668 PRINT.indd 253 27/04/2022 11:07

 254 TRANSPORT ECONOMICS, 4TH EDITION

$ per tonne

100

50

Sheepskin

Meat
Wool Food Fresh fruit

Ores Metals

0 3 6 9 12 Million tonnes

Figure 7.9 A
 verage revenues from cargoes carried by the Australia–Europe Conference
(1973–74)

rates caused by potential competition from air transport is seen to be effective at

the top of the price range.
Price discrimination not only permits suppliers to recover their costs, but
also helps travelers and consignors in that services can be retained even if, in
some cases, it is necessary to differentiate the quality of service provided as well
as the fare charged. International air travel offers some examples of this type
of situation where differential prices are charged over a route according to the
specifications of the types of service the travelers are willing to pay for. The
gradual breakdown of the International Air Transport Association’s (IATA)
system of regulating airfares across the Atlantic in the l970s was accompanied
by the introduction of cut-price services such as ‘Skytrain’, operated by Laker
Airways (Abe, 1979). No-frill flights were introduced at low fares with seat
allocation dependent upon the willingness of potential passengers to queue for
tickets which could be purchased until six hours prior to take-off. Subsequently
this type of service has been modified and regular scheduled airlines now offer
a variety of ticketing arrangements on their flights. A clear segmentation of the
market had been recognized.
Figure 7.10 provides an illustration of the sort of situation that has
emerged; for example, if there are three different groups of potential travelers
with separate demand curves (D1, D2, and D3), then three separate fares should
be charged (P1, P2, and P3) to maximize the consumers’ surplus enjoyed. P1
is charged for the highest-quality flight, equaling the marginal cost of service,
with lower fares for poorer-quality flights. On the surface it appears that 0N1
passengers will travel first class, reaping a consumer surplus of ABP1, and
N1N2 passengers will pay P2 for the slightly lower quality of service and enjoy
a consumer surplus of HIG. This ignores the possibility that first-class travel-
ers may switch to the poorer-quality but cheaper services (that is, there may be
some ‘revenue dilution’). In fact, travelers N'1N1 could switch and increase their

BUTTON 9781786435668 PRINT.indd 254 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­255

A
C
B
P1
D

E H
J
G
P2
F I D1

L
Q D2
P3
M D3

0 N'1 N1 N'2 N2 N3 Passengers

Figure 7.10 Price discrimination according to service quality

welfare. Similarly, N'2N2 passengers appear to be the probable number of cus-

tomers for the poor-quality service, but again a further N2N3 may be induced to
join them by the lower fare. Whether people do take advantage of the possibility
of switching to cheaper but less convenient forms of service is uncertain, but the
availability of the range of services means the total potential consumer surplus
(the shaded areas) will exceed that generated if only a single price and service
package were available.

Dynamic Yield Management

There is also the practice of what has become known as yield management in the
air transport market, but is also used by other transport modes. Essentially, it is the
ability of management to gain additional revenue above costs from a pre-defined
activity (for example, a scheduled flight or sailing) when there is a need to ration
out capacity. For instance, there may be several people who want to book late for
a flight but the number of remaining seats is small. The airline will then push up
fares to ration out these seats to those willing to pay the most. This is not strictly
the same as price discrimination because the supplier is not trying to get all the cus-
tomers’ consumer surpluses but rather recover costs and allow those who gain most
from a service to have access to it. It is often called ‘dynamic yield management’.
It has become a key element in, for example, the scheduled airline industry
where the chief executive of one major operator, R.L. Crandall, former President
of American Airlines, stated, ‘I believe that yield management is the single most
important technical development in transportation management since we entered
the era of airline deregulation in 1979’. But it is also used elsewhere, including the
maritime sector.

BUTTON 9781786435668 PRINT.indd 255 27/04/2022 11:07

 256 TRANSPORT ECONOMICS, 4TH EDITION

The use of computer reservation systems (CRSs) by airlines enables an

airline to adjust the fares offered on a given flight, the seats available at each fare,
and the preconditions for booking at any fare. The preconditions (combined with
some relatively minor quality differentials inherent in first-class, business-class,
and coach-class areas of an aircraft) introduce product differentiation into the
management equation while the various fares charged for what are, in many ways,
very similar services represent the price differentiation element. The combinations
offered and the adjustments made by an airline as the time of any flight departure
approaches are designed to ensure the yield from the pre-determined decision to
offer that service covers costs and allocates seats efficiently.
The technique is somewhat different from that of the classic economic model
of price discrimination where management increases revenue by charging differ-
ent groups of customers (differentiated by their price elasticities) different prices
for the same product. The cost of serving each customer is, in the classic model,
identical. There is, however, normally some qualitative or other form of product
differentiation involved in yield management which implies cost differentials in
the provision of services to customers. First-class air travel does cost an airline
more than discount travel. Equally, however, the existence of economies of scope
may mean that the supplier of a range of different services can keep unit costs
lower than several suppliers each specializing in the services being offered. In
overall resource terms, therefore, yield management may provide for the develop-
ment a more efficient market structure.
The set of decisions confronting industries practicing successful yield man-
agement is often complex. The interactions and options available are numerous
fare levels, fare structures, constraints, and capacity distribution to be decided
simultaneously. This leads to a further common feature of yield management,
namely sequential decision-making. The form this takes varies between both
sectors and firms within sectors, but a broad indication of the decision sequence
adopted by many airlines when considering a scheduled service is set out in
Figure 7.11. In this case, the nature of the technology does permit a degree of
interaction between some of the stages (for example, discount levels and alloca-
tion of seats to different fare classes) as the time of the flight approaches.
Empirically, separating out yield management, which is a resource allocation
technique, from price discrimination, which is a revenue-maximizing technique,
is difficult in practice and inevitably murky. But some insights into the degrees to
which prices may vary for a service can be gleaned.
An example of what this means in practice can be seen in Figure 7.12 which
looks at fares offered by regularly ‘data scraping’ the web pages of the airlines
involved to see how fares are adjusted over time as departure approaches. The
‘fares-offered’ curves for a monopoly service indicate the degree to which avail-
able prices fluctuate, a reflection of the number of remaining seats, the fares being
offered by the competitor, and the expected demand levels of potential customers
at different times before take-off (last-minute business travelers, for example, tend
to be willing to pay more for a seat and de facto compete for them). The fares
offered rise towards take-off – but part of this may be attributable to rationing out

BUTTON 9781786435668 PRINT.indd 256 27/04/2022 11:07

 257
THE PRICING OF TRANSPORT ­

Determine the capacity of flight

Forecast demand for

full-fare seats

Determine degree of
‘overbooking’ level

Determine the number of

discount seats

Set fare levels and restrictions

Figure 7.11 Stages in arriving at yield-maximizing revenue in scheduled aviation

the increasingly scarce number of unsold seats, and part to the effort of the airline
to distinguish between leisure travelers who know their itinerary well in advance
and are seeking cheap travel and business travelers who often must fly at the last
minute and are willing to pay more.
Another form of differentiation by time of purchase that combines second-
(entailing discounts for bulk buying) and third-degree price discrimination is found
in the air cargo sector (Bowen and Leinbach, 2004). The link between forwarders,
who act on behalf of freight shippers, and airlines is normally initially through

Fare $ Phoenix–Des Moines: 1 August

1050

950

850

750

650

550
America West 08:53–17:20
450

350

250 July Year 2005

Source: Button and Vega (2007).

Figure 7.12 T
 emporal-fares-offered curves for return services from Phoenix to Des Moines
(leaving August 1 and returning August 5, 2005)

BUTTON 9781786435668 PRINT.indd 257 27/04/2022 11:07

 258 TRANSPORT ECONOMICS, 4TH EDITION

their acquisition of allocated or bloc one-way capacity on airlines. This entails

airlines six to 12 months before a flight’s departure offering capacity to freight for-
warders through essentially auctioning arrangements. The freight forwarders offer
‘bids’ (generally in terms of price per kilogram at the point of origin) based on the
anticipated demands of their customers, the shippers. The remaining capacity, the
cargo capacity, is then offered for sale one month before the flight departure.

Differentiation by Length of Journey/Haul

Price differentiation is not only by type of traffic or quality of service but may
also be by length of journey. David Friedman (1979) offers a classic example of
such a policy in the context of long-haul/short-haul differentials on American
railways. The practice of charging short-haul traffic a higher mileage rate on rail-
ways than long-haul, despite attempts to legislate to the contrary, was common
in ­nineteenth-century America. Friedman’s justification for this practice dem-
onstrates that without it there may arise quite serious distortions in transport
infrastructure provision.
As an example, suppose there is a railway link between three towns, A, B, and
C, where B is located between the other two towns. There is also river transport
(priced at marginal cost) available between A and C offering an identical service
to the railway but offering no communication for town B. The fixed, sunk cost of
the rail link is such that:

CAC = CAB + CBC (7.4)

that is, the sunk cost of the line from A to C is the sum of the two component
sublinks. Also, on the same basis, the variable cost is:

VAC = VAB + VBC (7.5)

Figure 7.13 shows the respective demand schedules for transport between
the different pairs of towns. The railways may maximize their profits by charging
down the demand curves DAB and DBC, where there is no competition from river-
borne transport, to the point where marginal (that is, variable) cost is reached.
Where competition does exist over the long route between A and C, the railways,
to attract customers, will want to charge at most the rate offered by the river
transport (call this RAC). The railways will, however, accept traffic at rates below
RAC but above the variable cost. The areas WAVAB, XDVBC, and RACEFVAC
show the revenues enjoyed by the railways on different links. If RAC is lower than
either the highest position of DAB or DBC, then there exists long-haul/short-haul
discrimination in addition to discrimination within each type of traffic. In other
words, identical goods with identical demand for transport schedules would be
charged more per mile for a short haul than for a long haul.
The sum of the revenues in the diagram offers a measure of the social value
of building and operating the ABC railway line. The aggregate producer surplus

BUTTON 9781786435668 PRINT.indd 258 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 259

$ $ $
X Y
W
E
RAC

F
VAC
A D
VAB VBC
DAB

DBC
DAC
0 Route A => B QAB Route B => C QBC 0 Route A => C QAC

Figure 7.13 Revenues from discriminating pricing according to length of haul

generated should be set against the fixed costs of provision, which, together with
calculations for AB and AC separately, indicate the long-term desirability of
keeping the entire line open or only segments of it. Without long-haul/short-haul
discrimination the railway would either have to give up some of its long-haul
business or fail to capture some of the consumer surplus generated on short haul.
Whatever the case, the railway’s incentive to invest would be distorted and some
economically desirable lines would not be operated.

Countervailing Power

The extent to which a supplier can engage in price differentiation depends on the
degree to which the customer ‘allows’ it to price down the curve and the competi-
tion that is encountered in the market. For example, while the airline sector, where
markets have been allowed to operate, has practiced extensive price discrimina-
tion, the operating margins have been well below other industries (Figure 7.14)
which were of the order of 5.6 percent in the 1990s and early 2000s. This is mainly
due to competition, but in other sectors, such as shipping, the power of customers
is important. This ability to essentially resist the monopoly power of a large trans-
port supplier is reflected, to a large extent, in the ‘countervailing power’ enjoyed
by the potential transport user.
The development of the idea of countervailing power is usually credited
to Kenneth Galbraith and his book, American Capitalism. This very ‘unGal-
braithian’ volume essentially argues that large-scale enterprises in oligopolistic
or monopoly markets often have much less flexibility than conventional theory
postulates to exploit their power because of the existence of monopoly power at
other points in what we would now call, as we saw in Chapter 1, the ‘value chain’.
The interest was initially stimulated by the observation that large buyers often
enjoyed significant discounts when purchasing from suppliers. While there may be
cost reasons for this, such as economies of density in shipping and handling, the
outcome may also be the result of the strong bargaining power of the purchaser.
Essentially, what is seen is a situation where, rather than having original
competition in a market, the competition is between the various buyers and
sellers up and down the value chain – there are, according to this approach,

BUTTON 9781786435668 PRINT.indd 259 27/04/2022 11:07

 260 TRANSPORT ECONOMICS, 4TH EDITION

6
Europe
US
4
Global

1990

1991

1992

1993

1994

2001

2002

2003
0
1988

1989

1995

1996

1997

1998

1999

2000

2005

2006
–2

2004
–4

–6

–8

–10

Notes: (i) A lack of a bar indicates a missing observation and not a zero operating margin.
(ii) Memberships of the various reporting bodies vary over time and thus the reported margins
reflect the associated carriers at the time of reporting.

Sources: Boeing Commercial Airplane; Association of European Airlines; Air Transport

Association of America; International Air Transport Association.

Figure 7.14 Airline operating margins: global, European, and United States

bi-lateral monopolies or oligopolies at various points in the chain (a bi-lateral

monopoly being a situation where there is a single buyer and seller of a given
product in a market). The level of concentration in the sale of the product results
in a mutual interdependence between the seller and buyer in price determination.
The notions of price-taking inherent in more standard neoclassical economics
are not upheld.
Countervailing power, Galbraith argues, prevents a large business from fully
exploiting customers: ‘In a typical modern market of a few sellers, the active
restraint is provided by competition but from the other side of the market by
strong buyers … . At the end of virtually every channel by which consumers
goods reach the public there is, in practice, a layer of powerful buyers’. Basically,
market power on one side of a market ‘create(s) an incentive to the organization
of another position of power that neutralizes it’.
It is perhaps useful to put Galbraith’s ideas, together with other forms of
market forces, into a wider context. Figure 7.15 offers a simple illustration of
the various types of direct and countervailing powers that may affect the pricing
decisions of an undertaking providing transportation services. The Figure is rela-
tively comprehensive, embracing ideas such as countervailing power, regulations,
the fear of public disquiet and complaint over excessive prices (the idea of ‘voice’
put forward by Albert Hirschman in his 1970 book, Exit, Voice, and Loyalty:
Responses to Decline in Firms, Organizations, and States), contestability, and so

BUTTON 9781786435668 PRINT.indd 260 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 261

Buyers Suppliers
(countervailing power) (countervailing power)

Voice
Shipping Potential competition
company (contestability)

Regulators

Other similar suppliers

Same industry Different industry (original competition)

Other types of supplier

(imperfect competition)

Figure 7.15 Forces affecting a firm’s pricing decisions

on, as well as original competition (the traditional name for competition within
the market).
Not all have agreed with this outcome, and the idea of countervailing power
from the time of Galbraith initiating his ideas has been the subject of some
skepticism. Academics such as George Stigler (1954), for example, questioned
in a variety of ways whether the ability of large purchases of factors of produc-
tion from monopolists in the value chain really had the incentive to pass on any
savings to final customers; why not keep the rent themselves? Stigler was strong in
his criticism of the concept:

[I]t simply is romantic to believe that a competitive solution will emerge, not merely in
a few cases, but in the general run of industries where two small groups of firms deal
with one another suddenly all the long-run advantages of monopolistic behavior have
been lost sight of in a welter of irrational competitive moves.

The theory is messy in the sense that it does not, except in very stylized
situations, produce a neat-equilibrium, optimal outcome, but rather results in
prices and outputs determined by games played by the various parties involved
and influenced by their power in the market. These may be short-term equilib-
riums but are not optimal. There is no canonical model. The degree to which
the outcome approaches either monopoly or competitive outcomes depends on
the strength and strategic aptitudes of the various parties involved. A variety of
things can influence these factors.
One indicator of the extent of countervailing power is the ability of, say, an
airport to price-discriminate between airlines that may want to operate from it.
The ability to first-degree-discriminate, involving charging different prices
for each take-off or landing depending on willingness to pay, is rare and unlikely

BUTTON 9781786435668 PRINT.indd 261 27/04/2022 11:07

 262 TRANSPORT ECONOMICS, 4TH EDITION

$ $ $
MCM

PLC

PLCC

MRLCC ARLCC MRLC ARLC MRM ARM

Low-cost carrier flights Legacy carrier flights Flights

Figure 7.16 Third-degree price discrimination between low-cost and legacy carriers

to be a practical proposition for an airport even if it does enjoy considerable

monopoly power and there are numerous airlines seeking slots. The information
and administrative requirements are too large. A more viable proposition to con-
sider is third-degree price discrimination whereby the market for airport services
can be segmented and where the segments have different elasticities of demand.
At the most basic level we can think in terms of segmentation between low-cost
and traditional, full-service carriers.
We can take the simple case of a profit-maximizing airport that is serving
both a low-cost carrier and a legacy carrier. In Figure 7.16 the airport decides
its output by equating its overall marginal costs (MCM) with its overall marginal
revenue (MRM). The price levied on the airlines will then be at the point on the
average revenue curve directly above the flights attracted (ACM) not illustrated in
the figure. However, landing fees are normally a larger part of a low-cost airline’s
costs than for a traditional, legacy carrier. This makes the former more sensitive
to the level of such fees. To increase its revenues, therefore, the airport can try to
charge separate fees to its two potential customer groups; for example, it may use
slot auctions (Button, 2008). It would effectively do this by drawing a horizontal
line through the MCM = MRM point and carry this over to the separate markets
for low-cost and legacy carriers. This intersection gives the combined profit-
maximizing output for the airport. Taking the resultant marginal cost across
to the separate marginal revenue curves of the different categories of carrier
optimally shares out slots between them. Prices for the low-cost carriers (PLCC)
and the legacy carriers (PLC) are then determined by the average revenue curves
corresponding to these slot numbers.
The types of issues associated with these conditions were explored by
Francis Edgeworth in the nineteenth century, and have subsequently attracted
considerable attention within the industrial organization fraternity with an inter-
est in game theory. Figure 7.17 provides a simple, and extreme, example with a
monopsony buyer (say a large airline) negotiating landing slot charges with a
monopoly airport.
Initially at one extreme, we treat the airport as a monopolist with numerous
potential airlines wanting to use its slots. In this case we get the standard text-
book situation where the airport will set its output where marginal cost equals
marginal revenue (Y1) and set price to clear the market (that is, price will be P1).

BUTTON 9781786435668 PRINT.indd 262 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­263

Slot prices ($)

Marginal slot cost

P1 Slot supply/
marginal cost

P2
Marginal revenue
product of slots/demand

Marginal revenue

0 Y1Y2 Slots

Figure 7.17 Price and output under a pure bi-lateral monopoly

In contrast, if an airline (or a strong alliance of airlines) has monopsony power,

and the airport is confronted with competition from alternative facilities, then the
price of a slot will be P2. This is because the market-supply curve for slots will be
positively sloped with a market-slot curve above it. The marginal revenue product
curve of slots – the additional revenue the airline obtains by buying an additional
slot – is downward-sloping, reflecting diminishing marginal productivity. To
maximize its revenue, the airline will buy slots up to the point where the marginal
cost of a slot is equated to the marginal revenue it will generate (that is, Y2 slots),
for which the market clearing price is P2.
The situation where there is bi-lateral monopoly generally leaves the
outcome indeterminate with a market price somewhere between P1 and P2. The
resultant price and output combination will depend on the bargaining skills of
the two parties and that, of course, will be a function of their game-playing skills.
While one can relate to this in the case of a large hub-and-spoke airline negoti-
ating slot rates at its main hub airport, there are other points in the value chain
where these types of bi-lateral monopoly or oligopoly may emerge. Of relevance,
here, is the monopoly power of labor unions in their dealing with airports and the
airframe manufacturers with the airlines.
What is clear, irrespective of the power of airports and airlines in the
example, is that the final consumer – passenger or shipper – does not benefit
greatly from a bi-lateral monopoly situation further up the value chain. The
sharing-out of economic rents seen in Figure 7.15, between the two providers
of intermediate services, is an effective extraction of consumer surplus from
the final users of air transportation services. The situation may also pertain
in other cases where there are imperfections up the value chain that result in
non-­competitive prices being pushed down the chain and being imposed on
­passengers and freight consignors.

BUTTON 9781786435668 PRINT.indd 263 27/04/2022 11:07

 264 TRANSPORT ECONOMICS, 4TH EDITION

Testing for the existence of countervailing power in any market is challeng-

ing. It is difficult to discern, for example, whether the fact that a large-scale buyer
of a product from a large producer gets a discount because of the scale economies
of such transactions or because the buyer can drive a hard bargain by virtue of
its purchasing power. Much of the general empirical work trying to ascertain
whether countervailing power has any significant impacts on markets uses a
structure–conduct–performance framework, looking at links between supplier
profits, or mark-ups, and on supplier concentrations with the additional consid-
eration of buyer concentrations. Limited work of this type, however, has been
performed on the aviation sector. There has been some work trying to benchmark
different airports’ efficiency, generally using data envelopment methods, but
normalizing for the characteristics of airports is difficult, especially when using
programming techniques.
One of the problems in exploring countervailing effects is that the market
power on either side of the market varies considerably by context. The situation
depicted in Figure 7.15 of a monopoly confronting a monopsony is very seldom
seen, and more common are oligopolistic bi-lateral structures where the market
powers on both sides are defused by continual games being played between the
buyers and suppliers themselves. Turning specifically to aviation, in some cases
groups of clustered airports are under common ownership, as with the main
London airports, giving them considerable regional monopoly power, whereas
in other cases there may be alternative terminals within proximity; in the United
States a two-hour drive between airports is generally seen as providing competi-
tion. Additionally, many airports have their rates regulated – for example, in the
United Kingdom the British Airports Holdings (BAH) is price-capped by the
Civil Aviation Authority (CAA) – making econometric analysis of market power
difficult. Air services are also not homogeneous, and nor are the airlines providing
them, making generalizations difficult.
The degree to which airports can segment in this way, however, depends
on the extent to which the airlines can dilute the revenue being collected by the
airport – essentially their ability to prevent differential fees. Many airports are
used by a small number of airlines, and in some cases even large airports rely
on a single carrier for 80 percent or more of their business. The extent to which
airports practice third-degree discrimination has not been rigorously examined,
although a study by Jacobs Consultancy (2007) did find differential passenger
charges for low-cost terminals at Budapest, Kuala Lumpur, Marseille, and
Singapore airports.
While airports may exercise some forms of price differentiation by carrier
type, it becomes more difficult to separate out the effects of cost variations from
charging policies based upon willingness to pay. Even in the cases cited, low-cost
carriers are often based in separate terminals that offer passengers more limited,
less grandiose, and generally less well located facilities. For example, Richard de
Neufville (2006) cites Geneva, Berlin, Kuala Lumpur, Paris Charles de Gaulle,
Singapore, and Warsaw airports as situations where specific facilities for low-cost

BUTTON 9781786435668 PRINT.indd 264 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 265

carriers have been provided. In effect, different fees may largely be reflecting dif-
ferent costs and not strict price discrimination by demand.
The fact that only a few cases of clear third-degree price discrimination
seem to exist could be suggestive of airlines being able to exercise a consider-
able degree of countervailing power. This would seem to be illusory because of
capacity problems at major airports in many parts of the world, and particularly
Europe, and regulatory regimes that force airports to set charges based upon
narrow accountancy costing principles. The form exists when, given the capacity
of an airport, the nature of the demand curves for potential low cost and legacy
carrier users are such that profit maximization leads to a high price that only the
traditional, legacy airlines are willing to pay.

7.6 Pricing with Stochastic Demand

Our discussion of price discrimination as a method of recovering costs has to

date made the rather heroic assumption that the supplier of transport – be it ship
owner, railway manager, airline operator, or whatever – has perfect knowledge
of the demand situation being confronted. Most transport managers have, from
past data and employing ‘managerial intuition’, some notion of the average level
of demand for their services and some idea of how this demand fluctuates. With
yield management, regularly posting and changing the fares offered can often
give a pretty good idea of short-term demand conditions. In practice, though, the
assumption of complete or near-complete information is often unrealistic. This is
particularly so over longer periods. The specific problem of systematic and regular
known short-term peaks in demand is considered in the following section; here we
concentrate on irregular fluctuations in demand of a stochastic nature. A trans-
port undertaking may, for instance, know that fluctuations are about 20 percent
of the average daily demand, but have no way of telling whether tomorrow’s
demand will be, say, 7 percent above the average or 14 percent below it.
The introduction of this notion of ‘stochastic’ demand requires a slight modi-
fication to the marginal cost pricing approach. The conventional arguments for
discriminate pricing revolve around the idea that simple marginal cost pricing will,
because of declining average cost, result in the supplier incurring a financial loss.
Ralph Turvey’s (1975) widely accepted position formulated nearly five decades
ago is that the problem should be expressed in terms of the difficulties associated
with matching the services supplied (usually vehicle journeys) with those that are
demanded (usually passenger journeys) when a fixed timetable is operative.
Approached in this way, the problem is one of defining a price structure
which will cover all relevant costs but which will, at the same time, ensure reason-
able utilization of the transport capacity available. Two broad approaches emerge.
The first of these depends upon having fairly reliable information about demand
and its range of fluctuation, while the second relies upon good information on all
relevant costs. Here, we concentrate on the first approach and, especially, the load
factor for a service.

BUTTON 9781786435668 PRINT.indd 265 27/04/2022 11:07

 266 TRANSPORT ECONOMICS, 4TH EDITION

If fares are fixed so high that the number of people wishing to use the service
never exceeds the available capacity, then, as a result of the fluctuating demand
condition, there will frequently be substantial numbers of empty seats and
resources will be wasted. Alternatively, if the fares are set so low that capacity is
always fully utilized, then many people, who often have spent time queuing for
the service, will, again because of demand fluctuations, find themselves unable
to obtain a seat. Clearly, common sense suggests a compromise between these
extremes will meet the requirements of both supplier and potential traveler.
Turvey’s pragmatic solution is that operators should structure their fares so that
on average a certain percentage of seats will remain empty.
To some extent this is the situation that has developed with yield manage-
ment on the scheduled airlines where passengers have even greater flexibility by
being able, via booking early, to ensure themselves a seat if they wish rather than
risk disappointment, but at the cost of forfeiting the fare if they do not make the
flight. The situation was also evident on the streets of central London in the past
where parking meter fees were fixed so that on average 15 percent of spaces were
vacant although, of course, from experience we know that at times it is impossible
to find a vacant parking space while at others they are in abundance. One might
also point to the ‘standby capacity’ kept by British Rail until the 1968 Transport
Act, which it was claimed acted to cope with long-term fluctuations in demand
for railway services.
To cover costs in this type of situation without recourse to either direct or
cross-subsidization it is likely that price discrimination is necessary. Any of four
standard types of discrimination (that is, by type of passenger, degree of comfort,
regularity of use, and/or seat availability) could be used for this.
Where knowledge of demand fluctuations is less precise, then Turvey’s
second and rather more pragmatic approach may be applicable. Here fares can
be determined by simply dividing available costs of the service by the passengers
carried and the service only runs if such fares broadly correspond to those on the
remainder of the transport system. Additional revenue may then be gained on an
ad hoc basis by raising fares for those groups where willingness to pay exceeds the
cost-based fare. The actual avoidable costs can be estimated, where there is uncer-
tainty about initial traffic levels, using the following formula that, for simplicity, is
couched in terms of a passenger railway service:
[(Probability that marginal passenger will necessitate as extra
(7.6)
carriage) × (Cost of extra carriage)]
+ [(Probability that marginal pasenger will necessitate an extra
train) × (Cost of extra train)]
Because this probability LRMC curve represents costs as an increasing func-
tion of the number of passengers and since this is itself a decreasing function of
the fare charged, there is likely to be a fare structure in which such marginal costs
are recovered.
Whether the fare is optimal, however, depends upon timetable flexibility;
so far we have implicitly assumed a given timetable. The overall fare is set at the

BUTTON 9781786435668 PRINT.indd 266 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 267

level of the marginal social cost of an extra passenger, in other words equal to
the frustration and inconvenience he/she causes to other potential but disap-
pointed travelers by occupying a scarce seat. The combination of timetable and
fare that equates the marginal cost, so defined, with the marginal financial cost is
thus an overall optimum. In practice, of course, imperfect knowledge of demand
situations, plus the need to make timetabling and pricing decisions simultane-
ously, makes it unlikely that such an overall optimum will be attained except by
chance.

7.7 The Problem of the Peak

Most forms of transport, both freight and passenger, experience regular peaks in
demand for their services. Urban public transport (upon which our attention will
be focused later) experiences peaks in demand during ‘rush hours’ each weekday
morning and evening. While there is considerable variation between cities, typi-
cally rush hour lasts from about 6 to 10 am and from 4 to 7 pm local time, with
some people traveling places during their lunchtime causing a mini rush hour
from noon until 2 pm. Urban freight transport also has peaks in demand to
match the seasonal needs and operating practices of customers. In London, for
example, most deliveries are made between 7 am and 1 pm. Over a year, air, bus,
and rail services meet peaks in demand from holiday traffic during the summer
months and over public holidays, while within a week there are marked differ-
ences between weekend and weekday demand levels. Over an even longer period,
shipping is subjected to cyclical movements in demand as the world economy
moves between booms and slumps.
The difficulty in all these situations is to determine a pattern of prices that
ensures that transport infrastructure is used optimally, provides a guide to future
investment policy, and ensures that all relevant costs are recovered. Unlike the
previous section, we are concerned here with problems arising from systematic
variations in demand, frequently over a relatively short time period during which
adjustments cannot be made in capital equipment to ensure that price is always
equated with LRMC. The problem is essentially one of indivisibility in the time
dimension of supply relative to demand and is, therefore, a form of the joint pro-
duction problem. Problems of this kind do occur in other sectors of the economy,
but transport, like electricity and some other forms of energy, cannot be easily
stored to reconcile systematic changes in demand with smooth, even production.
Reconciliation can only be effectively achieved through price adjustments or
shortages and excesses in the market.
Before proceeding to look at the peak-load pricing problem it is worth noting
that there exists a parallel spatial/directional problem of joint costs in transport
which can be treated in an identical way to that of the peak. This involves the
question of deriving appropriate rates for front-hauls and back-hauls – a situ-
ation often found in the provision of unscheduled road haulage, or freight and
shipping services. Basically, there is a high demand for a service in one direction

BUTTON 9781786435668 PRINT.indd 267 27/04/2022 11:07

 268 TRANSPORT ECONOMICS, 4TH EDITION

(the front-haul) but a lower one for a return service (the back-haul); that is,
demand is uni-directional in nature whereas supply consists of round-trip jour-
neys. This situation is directly analogous to the peak-load situation, with the
front-haul being the spatially directional equivalent of the peak, and simple sub-
stitution of word ‘yields’ the appropriate analysis.
Perhaps the most widely discussed peaking problem involves urban public
transport and bus services. The size of most urban bus fleets is determined by
the demand for public transport services during the morning and evening com-
muter rush hours. Typically, over half the passengers carried during a day travel
during the main peak periods. In Manchester, for example, 1,090 buses were
required to meet rush-hour demands in 1966 while only 400 were used during
the midday period. Comparable figures for Birmingham Corporation Transport
Department in 1969 were 1,500 and 327 vehicles respectively. Bus road crews
may also be considered as a joint cost since numbers are determined by peak
demand. It is seldom possible to cover both daily peaks with one shift, hence
either two shifts are required, or else split-shifts must be introduced usually
involving inconvenience payments (often equal to the standard wage) being
paid for time between peaks. The total wage bill for road staff, which amounts
to about 50 percent of the total cost of most British bus operators, is, therefore,
almost invariant with demand and may be treated as a joint cost of providing
peak and non-peak services.
To determine optimal prices let us assume that during a 12-hour period a
bus operator is confronted by two different demand situations, each of six hours’
duration. In Figure 7.18, D1 is the low, off-peak demand situation and D2 the
peak-level demand curve. The SRMCs of operation (fuel- and mileage-dependent
depreciation) are assumed constant at level 0a until the capacity of the bus fleet,
which initially is assumed as fixed, is filled, whereupon they become infinite. In
the short term, with capacity fixed, the objective is to maximize social welfare by

$
SRMC

Dcycle

2a + A LRMCcycle

a+A LRMC
P'2
P2
P'1
P1 = a

D2

D1
0 QP0 Q0 QN0 Passengers

Figure 7.18 The optimum peak price

BUTTON 9781786435668 PRINT.indd 268 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 269

making optimal use of the fleet. In this case the off-peak and the peak demands
should be priced at their respective SRMCs. The fares should, therefore, be
P1 and P2 at the off-peak and peak respectively, with corresponding passenger
numbers QP0 and QN0.
Changes in capacity brought about by varying the fleet size do not influ-
ence SRMCs except to the extent that the capacity constraint is pushed further
to the right if vehicles are added and to the left if they are withdrawn. LRMCs
are treated as constant at a level A. Since the capacity is joint to both subpe-
riods, changes in capacity should be determined by the combined demand of
peak and off-peak periods, that is, the full cycle of activities (Hirschleifer, 1958).
Consequently, it is Dcycle that is the relevant demand curve (because this repre-
sents the vertical summation of D1 and D2). Further, it is LRMCcycle that is the
relevant long-term cost curve, because this represents the combination of the
short-run costs in the two periods plus fixed costs; that is, 2a + A. The situation is
analogous to that of a collective good.
The optimum long-run capacity in the case illustrated, and assuming there
are problems of indivisibilities, involves a contraction of capacity to 0Q0. The
non-peak travelers will then pay P'1 and peak travelers P'2. Given the way the
Dcycle and LRMCcycle curves are derived, the combined revenues from off-peak
and peak fares will exactly equal the costs involved. Also, the off-peak fare now
exceeds SRMC and capacity is fully utilized around the clock, the pricing differ-
ences reflecting differing strengths of demand. In summary, changes in capac-
ity, because they are joint to both periods, now depend upon the sum of the
differences between price and the capacity’s operating costs per period relative
to the cost of providing new capacity for the entire cycle. That is, investment is
justified if:

[(P1 – a) + (P2 – a)] > B (7.7)

It is relatively simple to extend the analysis to any number of subperiods,

which may be of unequal duration, by weighting the different periods according
to their fractional importance in the entire cycle.
In practice, the situation is often somewhat more complicated than that
depicted, because passengers are not rigidly confined to either period but may
switch trips between them if differential fares are levied. Unfortunately, little
empirical work has been conducted on this, but it is quite possible that our implicit
assumption that D1 and D2 are independent is often unrealistic and allowance
needs to be made both for their own and for cross-elasticities. Also, of course, it
may be necessary to take cognizance of interdependencies between the demand for
urban bus travel as a whole and the possibilities of switching to private transport.
While there is a firm economic basis for peak-load pricing, its implementa-
tion has been piecemeal. An early example was the pricing policy pursued in the
Manchester–Salford area in the United Kingdom between the summers of 1970
and 1975 (Tyson, 1975) and there are differential metro fares in American cities,
such as Washington DC. Some American airports have attempted to introduce

BUTTON 9781786435668 PRINT.indd 269 27/04/2022 11:07

 270 TRANSPORT ECONOMICS, 4TH EDITION

Table 7.4 Pricing structure at New York, LaGuardia Airport (2002)

Type of charge Minimum charge

Operation during peak (8 am to 9 pm), non-scheduled carriers $100

Operation off-peak, non-scheduled carriers $25
Operation during peak (8 am to 9 pm), scheduled carriers $50
Operation off-peak, scheduled carriers $20

Source: Port Authority of New York and New Jersey, Aviation Department, Schedule of charges
for air terminals, 2003; Port Authority of New York and New Jersey, Guidelines for constructing an
economically defensible peak-hour flight fee.

peak pricing but, as with facilities such as Boston Logan, it has met with legal
problems, and at other places, such as LaGuardia (Table 7.4), it hardly reflects the
magnitude of demand variations (Schank, 2005). It is also employed at the major
airports in the United Kingdom, although the primary objective has historically
been to raise revenue for investment rather than as a strict short-term capacity-
rationing device. First attempts to introduce peak-load pricing date back to 1972
when the BAA established a runway movement charge at the busiest time of day
at Heathrow. Passenger peak charges were implemented in 1976. Both the differ-
ential between peak and off-peak charges and the definition of peak periods have
changed considerably over the years. The ratio of peak to off-peak fees increased
over the years until the mid-1980s. Several differentiations for peak landing fees
have been tried but these were eventually succeeded by a uniform peak-hour fee.
The charter airlines in Europe in the 1990s – the mainstay of the tourist indus-
try before the emergence of low-cost carriers – also practiced peak-load pricing
when setting their fares. In their case, this reflected seasonal variations in tourism
demand patterns.

7.8 Transport Subsidies, Operational Objectives, and Pricing

Many sectors of transport enjoy quite substantial levels of either central or local
government subsidy. (Table 7.5 shows the percentage of transit costs recovered in
some of the world’s major cities, and Figure 7.19 provides some detail about the
situation regarding recent time tends in London’s transport subsidies.) Because
they represent payments for transport services by non-users, subsidies complicate
the pricing problem.
To some extent the type of problem created depends upon the form of
subsidy given. If a central or local government provides the subsidy for a specific
service, then it may be seen as representing that government’s demand for that
service and treated alongside the demand of other customers; the service subsi-
dies given by local authorities to specified bus services in the United Kingdom
under the 1985 Transport Act may be categorized in this way, as may many of
the subsidies to franchised services in other European countries and in some
American cities (Morlock, 1987). From a pricing/operational point of view such
subsidies are relatively easily assimilated into standard economic models. This is

BUTTON 9781786435668 PRINT.indd 270 27/04/2022 11:07

 271
THE PRICING OF TRANSPORT ­

Table 7.5 Levels of operational transit subsidies in various cities

City Percent subsidy City Percent subsidy

Perth 72 Brussels 27
Adelaide 60 Munich 54
Brisbane 46 Stockholm 33
Melbourne 76 Vienna 41
Sydney 45 Hamburg 38
Phoenix 72 Copenhagen 34
Denver 81 London 7
Boston 76 Paris 39
Houston 72 Singapore −15
Washington 50 Tokyo −5
San Francisco 65 Hong Kong −36
Detroit 77 Kuala Lumpur −35
Chicago 54 Surabaya −27
Los Angeles 57 Jakarta −1
New York 53 Bangkok 7
Toronto 39 Seoul 3
Frankfurt 55 Beijing 80
Amsterdam 60 Manila −12
Zurich 40

Source: Newman and Kenworthy (1999).

so in the United Kingdom case where the routes to be subsidized are put up for
competitive tender, a process that is designed to ensure the minimum subsidy is
paid for the designated service. Further, the evidence from London, and from
similar systems in the United States, is that there are cost efficiency gains (after
allowing for administrative and monitoring expenses) of about 20 percent over
more general subsidy arrangements.
When lump-sum subsidies are given to transport undertakings for general
revenue purposes, however, problems arise in deciding upon the best methods of
using the subsidy and the appropriate charge to levy on customers. It is difficult
to devise pricing and operational objectives which ensure that management uses
the fixed subsidies efficiently to attain the welfare objectives for which they are
intended. A few possible objectives have been cited as offering a way around this
problem.
It has been argued that commercial criteria (with profit-maximizing pricing)
in this situation would lead to monopoly exploitation and be counterproductive
in terms of the social objectives justifying the subsidy, while a social-welfare-max-
imizing criterion (with marginal cost pricing) would break the link between costs,
prices, and output, and lead to probable X-inefficiency: the provision of services
at excessive costs. To circumvent these problems, and to provide clear pragmatic
guidelines for lower-level management, London Transport attempted in the late
1970s to maximize passenger mileage subject to a budget constraint (that is, that
costs are recovered after the fixed-sum subsidy has been taken into account).
Operationally, when the criterion is applied at the margin, this means:

BUTTON 9781786435668 PRINT.indd 271 27/04/2022 11:07

 272 TRANSPORT ECONOMICS, 4TH EDITION

35.0 Net Public Transport Support

Concessionary Travel spend
30.0 Bus Service Operators Grant

25.0

20.0

15.0

10.0

5.0

0.0
5

4
/0

/0

/0

/0

/0

/1

/1

/1

/1

/1
04

05

06

07

08

09

10

11

12

13
20

20

20

20

20

20

20

20

20

20
Note: Public transport support is monies spent by authorities to provide unprofitable ‘socially
necessary’ services generally by tendering them out to private operators. Concessionary fare
reimbursements are monies paid to compensate bus operators for carrying certain types of
passengers, such as the aged or physically impaired, at low or zero fares.

Figure 7.19  et government support per passenger journey for London bus travel
N
(current £)

Reduce price as long as the increase in passenger mileage resulting exceeds the loss of
revenue multiplied by the shadow price of public funds; increase bus mileage as long
as the increase in passenger mileage resulting is greater than the net addition to the
financial loss multiplied by the shadow price of public funds. (Nash, 1978)

The criterion, therefore, permits adjustments to the system operated so

that costs are met (after allowing for the subsidy) and relatively junior manage-
ment can assess the desirability of alternative courses of action. The criterion
is, however, demonstrably inferior in theoretical terms to a pure marginal cost
pricing strategy: price discrimination and cross-subsidization may result which
push fares on inelastic services above marginal operating costs so that revenue
is available to finance services which exhibit a relatively high elasticity (and thus
easily provide an increase in passenger miles). It is quite possible, therefore, for
a service to be operated for which the level of demand is never high enough to
permit price to cover marginal costs.
It is possible, at the expense of complicating the criterion, to devise weight-
ing schemes which can be attached to passenger miles on various routes to reflect
their importance to the decision-maker, and schemes do exist which yield the
same results as marginal cost pricing. Whether it is justified to adopt such weights
must be an empirical question, depending on the loss of welfare that may accom-
pany the simple unweighted passenger-miles-maximization approach relative to

BUTTON 9781786435668 PRINT.indd 272 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 273

the administrative costs of implementation. Dieter Bos (1978) has also pointed
to the distributional implications of the criterion and suggests that positive dis-
tributional effects may justify a certain level of welfare loss although, again, the
exact distribution effect cannot be determined by a priori argument. (Empirical
evidence in London suggests, however, that the London Transport scheme did
have desirable distributional implications.)

7.9 Market Instability, Suboptimal Supply, and the Empty Core

To recap, economic theory tells us that, when there are no fixed costs, bargaining
between suppliers and customers will ensure that prices are kept to a minimal
level that allows suppliers to recover all costs over the long term, and the marginal
cost of meeting customer demand represents the entire cost of production. The
problems come when there are fixed costs, or indivisibilities in the cost function.
The traditional view of fixed costs was developed when the bricks, steel, and
mortar of industrial plants had to be paid for. It was largely seen in relation to
manufacturing, or in the context of a services industry, in terms of the immobile
hardware involved: rail track and bridges. This seemed logical at the time because
fixed costs are defined as invariant with the amount of production and the physi-
cal plant and infrastructure of the time fitted this description. Fixity is, however,
a relative concept, and while, as we have seen in Chapter 5, a rail track may be
seen as fixed over a long period, a railway locomotive may be seen as fixed over a
shorter time until it needs replacement. In other words, some costs of production
are fixed over the given decision period, while the train still operates, but become
variable when replacement is needed. While the locomotive keeps functioning,
the use made of it will be influenced by the marginal costs of maintenance, crew,
fuel, and so on.
But with service industries, and especially those involving scheduled services,
the fixed costs are somewhat different in detail, if not in their strict definition.
While airlines and conference shipping lines, for example, do use expensive hard-
ware, this is not their underlying fixed cost problem. Indeed, the largest costs of
airlines have traditionally been their labor. These in the traditional sense are gen-
erally, but not entirely, seen as variable costs. Even aircraft are now seldom owned
by the carriers but are leased, sometimes on a wet-lease that includes crew. The
result is that airlines are increasingly becoming ‘virtual carriers’ that act to bring
together packages of services owned by others and thus are encumbered with few
fixed costs themselves in the traditional economic sense.
Fixed costs in a modern service industry, including airlines, however, can
take a different form. An airline is committed to a scheduled service some six
months or so before the flight: it is committed to have a plane, crew, fuel, gates,
landing and take-off slots, etc., available at a given time. This does have an advan-
tage that fares are often collected before the airline has to provide the service, but
in a highly competitive market this is generally more than off-set by the limited
amount of revenue that is ultimately collected. The commitment to a scheduled

BUTTON 9781786435668 PRINT.indd 273 27/04/2022 11:07

 274 TRANSPORT ECONOMICS, 4TH EDITION

$ MC1

MC1 + MC2

P1

AC

D
P2

MR

0 Passengers

Figure 7.20 Indivisibilities in supply

take-off time poses the traditional cost recovery problem associated with any
decreasing cost situation: costs cannot be recovered by charging a single marginal
price.
The problem can be seen in Figure 7.20. Here we have a situation where there
is a single air (or ship or bus) scheduled service that has a marginal cost curve of
MC1, an average cost curve of AC, and is confronted with a market demand of D.
Being a monopoly provider, it will set its price at P1 and enjoy monopoly profits
of the shaded area above AC curve. A second carrier, attracted by these profits,
can enter the market with identical equipment and will push out the marginal cost
curve to MC1 + MC2. The result in this competitive situation, where price is set
equal to marginal cost, is that fares will decline to P2, resulting in a combined loss
for the two suppliers equal to the shaded area below the AC curve. There must,
therefore, be some mark-up above marginal prices to sustain the two carriers. As
we have seen, airlines and other suppliers in modern deregulated markets largely
engage in second- and third-degree price discrimination and charge passengers
and shippers different prices to try to extract as much revenue as possible. In
general, this is temporal price differentiation.
While this approach can allow revenues above marginal cost to be generated
when a service has some degree of monopoly power (for example, no competing
flight to the destination within a reasonable time frame), the problem is that with
a fixed schedule in a competitive market, the various airlines set take-off times for
each destination at about the same time. This leads to intense competition to fill
seats and forces fares down to levels that do not allow all the costs of individual
services to be met. It is worth filling a seat once offered with anyone willing to pay
for its marginal cost.
Empirical analysis supporting the logical basis of the temporal-fares-
offered curve has been well established in studies of European and American

BUTTON 9781786435668 PRINT.indd 274 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 275

Fare $ Phoenix–Kansas City: 1 August

450
Southwest 05:45–16:20
America West 09:07–17:35
400

350

300

250

200
July Year 2005
150

Source: Button and Vega (2007).

Figure 7.21 T
 emporal-fares-offered curves for return services from Phoenix to Kansas City
(out August 1 and returning August 5, 2005)

air transport markets. Figure 7.12 above, just as an example, looks at a United
States monopoly market and illustrates the consistent rise in fares as the time of
departure approaches. This pattern is also found on the numerous other routes
where this technique has been applied, and holds irrespective of whether the
monopoly airline is a low-cost carrier or pursues the traditional, full-service
business model.
Figure 7.21 shows the sort of temporal-fares-offered curve that emerges
when two carriers offer nearly identical services, differentiated largely by a small
difference in departure times. Again, the pattern is consistent and can be exam-
ined in detail in other works. This is also from the United States and shows the
volatility that arises, and the lack of a significant and consistent rise in fares
towards take-off is clear. Other studies looking at services where there are more
than two competitors show a further flattening-out of the temporal-fares-offered
curve, with near commonality of fares offered by all up to take-off, as one would
expect in the large numbers case.
It also leads to suboptimal levels of investment despite excess capacity during
peaks in the cycle. The fact that full costs are not recovered, and that ultimately
an airline will withdraw a service or go out of business, is known as the ‘empty
core problem’. It is neither a new concept – it was developed in the 1880s by a
largely forgotten Oxford economist, Francis Edgeworth – nor is it one that has
limited application. In the long term, as potential investors become aware of this
problem, they will reduce or cease to put new capital into the industry.
Overall, there are several conditions in which there may be no core and,
hence, a market may not be sustainable. These occur when there are relatively

BUTTON 9781786435668 PRINT.indd 275 27/04/2022 11:07

 276 TRANSPORT ECONOMICS, 4TH EDITION

large fixed costs, avoidable (set-up) costs, indivisibility, network effects, or severe
fluctuations in demand. An unsustainable market may also exist when some
suppliers enjoy a degree of institutional or financial protection, and when there
are significant variations in the costs of suppliers. In practice, many public
utilities, transport industries, and some manufacturing industries meet these
­criteria.
The lack of a core can broadly result in two outcomes depending on the
reactions of the players in the market. If the transport suppliers think they can
beat out competitors in the market, then there will be instability as suppliers keep
entering and leaving the market. Alternatively, if potential suppliers are more
rational and realize that their market entry will result in excess capacity and an
inability to recover full costs, then they will not enter. The result is that there will
be a suboptimally low capacity provided. Looked at in another way, there are
indivisibilities in the aircraft or ships scheduled and the overall capacity will be
lower than would occur if they were perfectly divisible.
The fact that this economic conundrum applies to the scheduled airlines
(Button, 2003) and the scheduled shipping industry (Pirrong, 1992) has been
appreciated for some time, but largely ignored in policy formulation. The com-
plexity of the underlying economic model has hindered the communication of
the issue to decision-makers. This situation also runs counter to some traditional,
often ideological, views of competition policy which hold that there can ‘never be
too much competition’.

7.10 Indirect Pricing

In some cases, there are excessive transactions costs involved in directly pricing
a user of a transport service, or if it is not possible for technical reasons. The
outcome in many cases is an adjustment to the price of a complementary good;
in the case of a publicly owned transport facility, this means de facto a tax. An
example of this in the United States and many other countries has been the
pricing of road use from the 1930s when the road system began to be improved
and largely paved. Tolls paid on turnpikes had been a traditional way of pricing
many major roads, but with the advent of the widespread adoption of the motor
vehicle, the excessive congestion created around manual toll booths saw a fuel tax
come in as a replacement.
There are important trade-offs to consider when using indirect pricing. It
may reduce the transaction costs of collection, which may involve savings for
the supplier of transport services, other users, or third parties, but it is not an
efficient price. It does not directly reflect the use of the transport system and
thus does not optimize that use. A fuel tax, for example, is poorly correlated to
the costs of individuals’ road use; payments vary by engine size and where the
vehicle is used, which may not be where the costs are incurred. There may be
jurisdictional issues when, for example, fuel is purchased in one location but the
driving is largely done in another. There is also the danger that because of the

BUTTON 9781786435668 PRINT.indd 276 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 277

lack of direct accountability, the authorities may manipulate the indirect charge
to raise revenues not directly related to the costs of the facility. (This is distinct
from a pure sales or purchase tax that is a clear form or government revenue
collection and is levied on many items.) Finally, the charge may end up serving
several purposes.
There have been efforts made to see to what extent these surrogate prices do
reflect actual costs of use. But this is not easy. In addition to the micro issues of
how to allocate costs to categories of road user, trucks, cars, and so on, there is
the issue of what items to include (for example, is air pollution a road cost in the
conventional sense?), and how to value them. Even when it comes to items such
as capital costs there are problems. Road investments are lumpy and outlays vary
considerably year by year. The ‘pay-as-you-go’ method simply takes the annual
expenditure, which may be considered a rather simplistic approach given the
volatile nature of investment. The ‘public enterprise’ approach looks at a road as
an investment and considers how much interest would have to be paid to borrow
the money required to make that investment. While there are some sound eco-
nomic grounds for this methodology, it does involve somewhat strong assump-
tions in calculating the capital tied up in the system and the interest that would be
required to finance this capital.
There are also issues about how costs should be allocated between users of
common infrastructure. Trucks, for example, cause more pavement damage than
do cars but because they go more slowly, design standards for bends, for example,
could be more flexible. If the pavement depth were designed optimally for trucks
then there would be less maintenance required and the allocation of operating
costs of the track would be less for larger vehicles. Simply looking at taxes and
other non-user-based charges and treating them as prices is not helpful. In one
sense, the methods of pricing much transport infrastructure is a little like charg-
ing a premium on butter to cover the costs of bread, but allowing consumers to
take as much bread as they wanted. The charges do little to make consumption of
bread efficient. Likewise, most surrogate charges for transport infrastructure do
little to ensure it is used optimally or to offer guidance as to when capacity needs
to be adjusted.
Despite these problems, there may be practical considerations that make
using indirect pricing a tractable and reasonable second-best approach. This
may be so in some countries where there are complex regulatory systems and
there is a need for a significant flow of funds for investment. The distinction
between the transport service and the complementary good may also be some-
what blurred.
An example of this is often found at airports that provide direct services to
airlines but also provide a variety of services on the ground to passengers. This
is often referred to as the ‘two-till’ approach to financing an airport’s activities.
Figure 7.22 provides an illustration of the sort of circumstances in which this
may arise. In more mature countries, where there is a significant traffic flow but
growth in traffic is relatively slow, there is more incentive to extract extra revenue
for passengers. This situation is compounded when the income of that country is

BUTTON 9781786435668 PRINT.indd 277 27/04/2022 11:07

 278 TRANSPORT ECONOMICS, 4TH EDITION

Traffic growth

Developing countries
Simple • Increased capacity of the airport system
economic • Large share of revenues from airside
regulation charges

Developed countries
• Maximum revenue base with limited
passenger growth
Complex
• Large share of revenue
economic
regulation

Share of commercial revenues

Source: Juan (1995).

Figure 7.22 Financing airports from two ‘tills’

high and travelers are willing to pay for additional airport services. When incomes
lower and airports are subjected to less regulation, the tendency is to rely much
more on direct income from airlines; it is also easier to collect than through a
multiplicity of retail concessions and the like. Whether the use of commercial
revenues leads to more efficient use of resources than charging higher rates to
airlines is, however, debatable. While it may be relatively easy to isolate the costs
of the air services charges from commercial services offered on the ground, a dif-
ferential amount of monopoly power may lead to distortions in investments and
in pricing. How to handle this issue has, for example, been a major policy issue at
London’s airports (Starkie, 2001).

References

Abe, M.A. (1979) ‘Skytrain’: competitive pricing, quality of service and the deregulation
of the airline industry, International Journal of Transport Economics, 6, 41–7.
Baumol, W.J. (1962) On the theory of the expansion of the firm, American Economic
Review, 52, 1078–87.
Baumol, W.J. and Bradford, D.F. (1970) Optimal departures from marginal cost pricing,
American Economic Review, 60, 265–83.
Beesley, M.E. (1973) Regulation of taxis, Economic Journal, 83, 150–72.
Bennathan, E. and Walters, A. (1979) Port Pricing and Investment Policy for Developing
Countries, Oxford University Press.
Bos, D. (1978) Distribution effects of maximisation of passenger rules, Journal of
Transport Economics and Policy, 12, 322–9.
Bowen, J. and Leinbach, T. (2004) Market concentration in the air freight forwarding
industry, Tijdschrift voor economische en sociale geografie, 95, 174–88.

BUTTON 9781786435668 PRINT.indd 278 27/04/2022 11:07

 THE PRICING OF TRANSPORT ­ 279

Button, K.J. (2003) Does the theory of the ‘core’ explain why airlines fail to cover their
long-run costs of capital? Journal of Air Transport Management, 9, 5–14.
Button, K.J. (2008) Auctions: what can we learn from auction theory for slot alloca-
tion? in A. Czerny, P. Forsyth, and H.-M. Niemeier (eds), Airport Slots International
Experiences and Options for Reform, Ashgate.
Button, K.J. and Vega, H. (2007) The temporal-fares-offered curves in air transportation,
Journal of the Transportation Research Forum, 46, 83–100.
Davis, O.A. and Whinston, A.B. (1967) Piecemeal policy in the theory of second-best,
Review of Economic Studies, 34, 323–31.
De Neufville, R. (2006) Accommodating low-cost airlines at main airports, International
Airport Review, 1, 62–5.
Friedman, D.D. (1979) In defense of the long-haul/short-haul discrimination, Bell Journal
of Economics, 10, 706–708.
Hirschleifer, J. (1958) Peak loads and efficient pricing: comment, Quarterly Journal of
Economics, 72, 451–62.
Jacobs Consultancy (2007) Review of Dedicated Low-cost Airport Passenger Facilities,
Commission for Aviation Regulation.
Juan, E.R. (1995) Airport Infrastructure: The Emergence Role of the Private Sector,
Cofinancing and Financial Advisory Services, World Bank.
Lipsey, R.G. and Lancaster, K. (1956/57) The general theory of second best, Review of
Economic Studies, 24, 11–32.
Mishan, E.J. (1962) Second-thoughts on second best, Oxford Economic Papers, 14, 205–17.
Morlock, B.K. (1987) Privatizing bus transit: cost savings from competitive contracting,
Journal of the Transportation Research Forum, 28, 72–81.
Musgrave, R.A. (1957) A multiple theory of budget determination, Finanzarchiv, 17,
333–43.
Nash, C.A. (1978) Management objectives, fares and service levels in bus transport,
Journal of Transport Economics and Policy, 12, 70–85.
Newman, P. and Kenworthy, J. (1999) Sustainability and Cities: Overcoming Automobile
Dependence, Island.
Nilsson, J.E. (1992) Second-best problems in railway infrastructure pricing and invest-
ment, Journal of Transport Economics and Policy, 26, 245–59.
Pirrong, S.C. (1992) An application of core theory to the analysis of ocean shipping
markets, Journal of Law and Economics, 35, 89–131.
Ramsey, F. (1927) A contribution to the theory of taxation, Economic Journal, 37, 47– 61.
Salanovaa, J.M., Estrada, M., Aifadopoulou, G., and Mitsakis., E. (2011) A review of the
modeling of taxi services, Procedia Social and Behavioral Sciences, 20, 150–61.
Samson, T., Nash, C.A, Mackie, P.J., Shires, J., and Watkiss, P. (2001) Surface transport
costs and charges: Great Britain 1998, Report, UK Department of the Environment,
Transport and the Regions.
Schank, L.L. (2005) Solving airside airport congestion: why peak runway pricing is not
working, Journal of Air Transport Management, 11, 417–25.
Shreiber, C. (1975) The economic reasons for price and entry regulation of taxicabs,
Journal of Transport Economics and Policy, 9, 268–79.
Starkie, D. (2001) Reforming UK airport regulation, Journal of Transport Economics and
Policy, 35, 119–35.
Stigler, G.J. (1954) The economist plays with blocs, American Economic Review, 44, 7–14.
Sturmey, S.G. (1975) Shipping Economics, Macmillan.
Truelove, P. (1992) Decision Making in Transport Planning, Longman.
Turvey, R. (1975) A simple analysis of optimal fares on scheduled transport sources,
Economic Journal, 85, 1–9.
Tyson, W.J. (1975) A study of the effect of different bus fares in Greater Manchester,
Chartered Institute of Transport Journal, 37, 334–8.

BUTTON 9781786435668 PRINT.indd 279 27/04/2022 11:07

 280 TRANSPORT ECONOMICS, 4TH EDITION

Wiseman, J. (1957) The theory of public utility price: an empty box, Oxford Economic
Papers (New Series), 9, 56–74.
Zerby, I.A. and Conlon, R.M. (1978) An analysis of capacity utilisation in liner shipping,
Journal of Transport Economics and Policy, 12, 27–46.

BUTTON 9781786435668 PRINT.indd 280 27/04/2022 11:07

 8 Containing the Environmental Costs
of Transport

8.1 Introduction

Chapter 6 offered evidence on the magnitude and the diversity of the external
economic costs associated with transport. We now move to consider the various
methods of confronting the problems posed by negative externalities. This
chapter focuses on those issues involving the environmental while Chapter 9 is
concerned with excessive traffic congestion.
It should be emphasized that the public policy focus on the external costs
of transport, while glaring in the developed world, is much less intense in many
less-wealthy countries. The affluence in Western nations has transferred part of
the desire from improving material living standards to that of improving (or
retaining) environmental quality. The marginal utility of additional financial
income, it is often argued, is for most people in the West of less value than a
cleaner, quieter, and safer environment in which to live. This is a comparatively
recent phenomenon when it comes to transport economics, with books on the
subject written in or before about 1960 giving scant attention to environmental
externalities.
Although increasingly concerned with matters such as deforestation, water
pollution, and soil erosion, most Third World countries still retain some of this
comparative indifference to the environmental impacts of transport: their gen-
erally poor living standards and inadequate transport systems necessitate that
effort be directed almost exclusively at improving material output. This is despite
mounting environmental problems, including those stemming from the release
of global climate change gases. What has changed over time is that more larger
countries have moved up the economic ladder – such as China, India, and Brazil –
and are becoming more engaged in environmental dialogues.
We have seen in Figure 6.1 that, ideally, externalities should be contained to
the point where the costs of further reductions exceed the marginal social benefits
(as stated in the early 1970s by the United Kingdom’s Royal Commission on
Environmental Pollution, ‘pollution should be reduced to the point where the
costs of doing so are covered by the benefits from the reduction in pollution’). It
should be re-emphasized that this is unlikely to mean zero pollution or zero con-
gestion but rather optimal levels of external cost; a congestion-free road system
would mean periods of wasteful idleness. To achieve this optimum, several possi-
bilities recommend themselves and the objective of this chapter is to evaluate the
effectiveness of the main ones.

BUTTON 9781786435668 PRINT.indd 281 27/04/2022 11:07

 282 TRANSPORT ECONOMICS, 4TH EDITION

These options are largely assessed in terms of their economic implications,

but in practice there are important political (most notably the lobbying power of
various groups) and wider social concerns (involving such things as ‘environmen-
tal justice’) that come into play when policies are being considered. There is con-
siderable analysis of these broader aspects in the economic literature, but these
are not dealt with in any detail here.

8.2 The Main Economic Approaches

When it comes to the environment, the adverse economic effects of transport result
from a chain reaction (Figure 8.1), starting with market failures such as inappro-
priate pricing and investment policies. Most notably these negative third-party
externalities affect the costs perceived by those in the transport sector who react
to input prices that do not appropriately reflect the external costs. Basically, there
is too much transport and often the wrong sorts of transport being supplied. This
in turn causes excessively damaging environmental effects such as noise, emission
of global warming gases, and water pollution. The ultimate welfare effects are felt
further down the chain in the form of higher environmental damage to health,
sustainable local biosystems, and reduced long-term productivity.
From a policy perspective, the diagram is useful because it shows that eco-
nomic policies designed to reduce, or optimize, the environmental costs associated
with transport may be introduced at various points in the chain. If, for political
reasons or because of transactions costs, it is not possible to tackle the market
failures per se, there may be relatively cost-effective second-best approaches that
can be applied further down the chain. For example, by building sound-walls
along the sides of streets to reduce excessive noise, restricting motor access areas
to noise-sensitive areas, or planting forests to act as carbon sinks. These second-
best approaches, however, may distort other markets and this needs to be borne
in mind when assessing their merits.
At the more specific instrument level, a broad indication of the range of
policies which can be applied to limiting the external costs of the motor vehicle
are set out in Table 8.1. As can be seen, policies can broadly be divided between

Market failures Industry Environment Social

• Property rights • Traffic levels • Noise • Health

• Prices • Vehicle types • Local air pollution • Conservation
• Investment • Schedules • Global pollution • Sustainability
• Regulations • Routes • Safety • Economic productivity
• Access • Soil/water contamination
• Land take
• Resource depletion

Source: Derived from Button (2001).

Figure 8.1 The stages in the environmental cost chain

BUTTON 9781786435668 PRINT.indd 282 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 283

Table 8.1 Taxonomy of policy instruments to control the environmental impacts of

motor vehicles

Market-based incentives Command-and-control instruments

Direct Indirect Direct Indirect

Vehicle Emissions Differential Emissions Compulsory inspection

 fee taxation standards  & maintenance of
Tradeable Tax allowances emissions control
permits for new systems
vehicles Mandatory use of
 low-polluting
vehicles
Compulsory scrappage
of old vehicles
Fuel Differential Fuel Fuel economy
 fuel taxation  composition  standards
High fuel Phasing out of Speed limits
taxes  high-polluting
fuels
Traffic Congestion Physical Restraints on vehicle
 pricing  constraints  use
Parking of traffic Bus lanes & other
charges Designated priorities
Subsidies for areas Information systems
 less polluting
modes

those based on market-oriented approaches and those involving command-and-

control measures. In practice, however, ‘baskets’ of instruments that reinforce one
another are usually employed. The table also reflects the ability of policy-makers
to direct their attention at various aspects of the transport problems with poli-
cies possibly being aimed at the vehicle, the fuel used, or at affecting the level and
composition of traffic.
While it is useful at times, both for illustration and to retain a link between
theory and policy, to refer to actual measures employed by transport authorities,
the emphasis is on the direct economic implications of the alternative approaches
rather than their political or social virtues. Broadly we proceed by initially looking
at the strict market approach to the subject (the ‘Coasian solution’), and then
move on to consider fiscal instruments and the internalizing of external costs by
charging those who generate them (the ‘Pigouvian solution’). Variants of both are
also looked at before considering command-and-control measures.

8.3 Marketable and Tradeable Permits

Much of the work of Nobel Prize-winning economist Ronald Coase (1960)

focused on the effects that complete or imperfect markets for property rights

BUTTON 9781786435668 PRINT.indd 283 27/04/2022 11:07

 284 TRANSPORT ECONOMICS, 4TH EDITION

have on creating excessive social costs. The solution to this problem he saw as
ensuring that individuals have more complete ownership over property rights
and that they can trade these rights to maximize the efficient use of resources.
This generally involves marketable or tradeable permits. The owner, say, of some
rights to pollute a river to some legal extent may sell those rights to another
individual or company. The term ‘tradeable permits’ is sometimes also used
to describe this concept, but on other occasions relates to the situation when
the owner of legal rights to pollute may trade these with another individual
or firm for rights to carry out similar pollution at a different place or time. A
transport example of the latter is the trading of landing and take-off slots at
major airports.
A more concrete example of the use of marketable permits was the Kyoto
round of negations that led to an international protocol on the release of global
warming gases from 2005 regarding global carbon emissions and ipso facto the
release of global warming gases. The system devised gave participating countries
an allocation of carbon and allowed them to trade these amongst themselves. The
countries, or individuals and firms within them, that can make the most effective
use of carbon will then buy additional amounts in a carbon market from those
that have allocations but cannot use them effectively. In this way the allocations
end up in uses that yield the greatest economic gains. Transport largely fell outside
of this particular system but there have been specific markets devised for modes
such as aviation.
The general way marketable permits work is illustrated in Figure 8.2. In this
diagram, which largely replicates Figure 6.1, the MNPB curve, drawn linearly
purely for the sake of simplicity, shows the net benefits that the perpetrator of the
externality (emissions in our example) gain from that activity. There are diminish-
ing returns as more emissions are generated. The MEC curve reflects the marginal
costs of emissions that are assumed to rise as they increase – for example, small

Costs/benefits

MNPB
MEC

0 Q* Q† Emissions

Figure 8.2 The role of property rights

BUTTON 9781786435668 PRINT.indd 284 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 285

emissions of NO2 cause little damage, but as they mount up, increments become
more serious.
If there are no property rights, in other words the perpetrators can use up
as much of the environmental resources as they wish at no cost, then emissions
will rise to point Q† in the diagram irrespective of the costs involved. There are
also no incentives to emit beyond this level: it is an equilibrium. If, on the other
hand, those affected by the pollution have the rights to clean air then there will be
no pollution: again, an equilibrium point, despite the net benefits that would be
generated if a degree of pollution were allowed.
Ideally, when the optimum level of emissions is at Q*, the net benefits are
maximized. To reach this point, and for it to be an equilibrium, Coase argues
that rights to the atmosphere should be created and trade in them permitted.
Assuming that those adversely affected were given these rights, then some of
the individuals would be willing to sell these rights to the emitters who would
have a willingness to pay. Trade would continue as long as there were enough
potential emitters willing to buy rights from the polluted that were not so
severely affected; in other words, until an equilibrium is reached with emission
levels at Q*. Alternatively, the initial allocation of rights could be given to the
emitters. The marginal emitters would be willing to be bought out by the most
severely affected polluted and this trade would again continue until emissions
are at Q*.
It does not matter in this context from a long-run perspective who is initially
granted the property rights to the atmosphere. The ultimate trading leads to an
optimal outcome. The initial allocation could be by an auction (in which case the
government would get a windfall once-and-for-all gain), it could be by lottery, it
could be according to current use, and so on. This is more of a political economy
than an economic matter because it has obvious distributional effects – those with
the initial allocation (or the government in the case of auctions) make a short-
term financial gain – but not long-term resource efficiency effects.
While theoretically the Coasian approach has intellectual merit, it has limi-
tations. To allocate property rights to, in the example, the atmosphere there is a
need to be able to define the units being traded. While this can be proxied in the
case of something like greenhouse gas emissions by using tons of carbon, this is
not so easy with things like noise or visual intrusion. There is also the issue of
policing to ensure contracts agreed upon are upheld, and there are the transac-
tion costs involved in the buying and selling of the rights themselves. Efficient
trade also assumes that markets work perfectly, and there is no monopoly power
present.
Despite these challenges, tradeable permits, or at least policies containing
their main elements such as cap-and-trade which is considered below, have been
used to deal with specific transport-related externalities, in addition to more
generic problems such as global warming gas emissions. They were used, for
example, in the United States to remove lead from gasoline in the 1980s and as
part of the efforts of the European Union’s Emission Trading Scheme to reduce
CO2 emissions from aircraft.

BUTTON 9781786435668 PRINT.indd 285 27/04/2022 11:07

 286 TRANSPORT ECONOMICS, 4TH EDITION

A less strict market approach is that inherent in the corporate average fuel
economy (CAFE) standards affecting the fuel efficiency of new vehicle fleets in
the United States and discussed below. Here there is no monetary trading but
car manufacturers can adjust the types of vehicles being made (intra-company
trading) conditional on average fuel burn standards being met which allows for
flexibility in their production to reflect costs and market demand.

Exhibit   Marketable permits for lead in gasoline

Lead was introduced into gasoline in the 1920s to boost the octane level of the fuel and to
reduce engine ‘knock’. The 1970s saw measures to reduce the lead content of the fuel as
catalytic converters, which were damaged by lead, were phased in to reduce hydrocarbon
(HC), nitrous oxides (NOx), and carbon monoxide (CO) pollution levels, and as it became
accepted that lead in fuel had adverse effects on human health and, in particular, on the
brain development of small children. The 1970 Clean Air Act Amendments gave the US
Environmental Protection Agency (EPA) the power to require new cars to use unleaded
fuel and be fitted with catalytic convertors.
In the light of further medical evidence, from 1979 the EPA accelerated the decrease
in pollution from pre-1975 vehicles by requiring refineries to reduce the average lead
content of gasoline, the agency thus deploying a tradeable permit mechanism. Different
standards were applied to the permissible lead content according to refinery size. Large
refineries with a productive capacity of over 50,000 barrels a day, or refineries that as a
corporate group had a capacity of over 137,000 barrels per day, had to produce gasoline
containing on average less than 0.8 grams of lead per gallon per quarter for the first year
and 0.5 grams for the following two years. Smaller refinery limits were based upon five
standards dependent on scale of production. How the refineries were to meet their
standards was up to them. This gave flexibility in terms of technology and also product
mix because the standards were an average across fuel types. The EPA subsequently
tightened and modified the system in 1982 (when a new standard of 1.1 grams was set)
and in 1983.
Refineries could buy and sell lead rights amongst themselves to come under the maximum
lead content stipulation. Those that could not individually meet the standards could ‘buy’
into a group of over-achievers and, by dint of averaging, all could comply with the 1.1-
gram requirement. The EPA used the term ‘constructive allocation’ to describe it. In 1985
refineries that had more than met the standard also had the option of ‘banking’ (basically
carrying over) any lead rights into the next quarter. This was designed to provide refineries
with more temporal flexibility. The lead content of gasoline was then reduced successively
in 1985 and 1986, with all inter-refinery averaging and banking ceasing in 1988, and all lead
additives banned in 1996.
The figure shows the annual reduction of lead in United States gasoline for the 25 years
after 1970. From 1979 to 1988, the regulations on refineries accounted for about 36
percent of the lead reduction that occurred.

BUTTON 9781786435668 PRINT.indd 286 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 287

100

90

80

70

60
Percentage

50

40

30

20

10

19 6
19 0
19 1
19 2
19 3
19 4
19 5
19 6
19 7
19 8
19 9
19 0
19 1
19 2
19 3
19 4
85

19 7
19 8
19 9
19 0
19 1
19 2
93

19 4
95
8
7
7
7
7
7
7
7
7
7
7
8
8
8
8
8

8
8
8
9
9
9

9
19
19

19
Year

Actual data on the effects of the trading scheme is empty space. The EPA did not collect
any. What evidence there is regarding banking suggests that the pre-banking permit price
for 0.1 grams of lead was $0.01 and after its introduction was $0.02 to $0.05, suggesting
the marketable permit program produced several hundreds of millions in abatement costs.
In a dynamic context there were likely additional savings to refineries associated with their
ability to smooth out transaction costs. These savings probably exceeded the $226 million
the EPA had projected, given the scale of banking that transpired.
See also: R.W. Hahn (1989) Economic prescriptions for environmental problems: how the
patient followed the doctor’s orders, Journal of Economic Perspectives, 3, 95–114; and R.G.
Newell and K. Rogers (2003) The US Experience with the Phasedown of Lead in Gasoline,
Resources for the Future.

8.4 The OECD’s ‘Polluter-pays Principle’

There is an inevitable problem with optimizing externalities, as pointed out by

Coase himself, and this involves the perspective with which one looks at an exter-
nality. Coase argued that externalities could theoretically be removed by allocat-
ing environmental property rights, either to polluters or those adversely affected,
and allowing trade to take place in these rights. There are clear practical problems
with doing this, not the least of which is that of policing the system, but the idea
does pinpoint the symmetrical nature of externalities. Should those who suffer be
protected or should those who benefit be compensated for desisting from their
current transport activities? While from an efficiency perspective there is no clear
answer, international agencies such as the OECD (1975) have taken the line that
morally those who pollute should pay for the excessive damage they cause the
environment – the so-called ‘polluter-pays principle’.

BUTTON 9781786435668 PRINT.indd 287 27/04/2022 11:07

 288 TRANSPORT ECONOMICS, 4TH EDITION

Emissions Charges

The concept of emissions charges is not a new one and can be traced back to the
publication of Arthur Pigou’s work on The Economics of Welfare in 1920. The idea
is that the authorities take responsibility for the environment – in effect they become
the property right holders, and charge users of the environment an appropriate price
(or tax) for that use. Figure 8.3 provides a standard type of visual illustration of how
such a charge would apply to road transport. We see the marginal cost of road users
(MPC) rise. Because road users only take account of the cost of trips they incur
themselves, they will keep making trips until the flow is F1. The traffic, however,
creates (for example) noise that when added to the costs of fuel and time in making
trips pushes up the full marginal cost at each level of flow to MSC. If this were taken
into account, then motorists would reduce their traffic and the flow would fall to F2.
The optimal environmental charge designed to reflect the noise costs on residents is
the difference between the MSC and MPC at this flow as illustrated. The road users,
by having the marginal costs of their trips pushed up from P1 to P2, are discouraged
from making trips that do cover this full cost and thus the flow falls to F2.
Strictly, this diagram could be a little deceiving. It assumes that imposing
a pollution charge will affect traffic flow, which it may, but in fact there may be
other ways in which a road user could deal with an environmental charge. In
terms, of say, a charge on noise, a road user could suppress the noise without
making fewer trips, or at least not significantly fewer; or, with a carbon tax, could
buy smaller-engine vehicles to conserve fuel but not seriously affect travel pat-
terns. In other words, part of the impact of an emissions charge would be to pull
the MSC cost down in Figure 8.3.
The more technically correct diagram would be to have emissions on the
horizontal axis to directly link the charge with the de facto effect on the environ-
mental cost. One can easily trace out that the environmental result would be the

$
MSC
(including
environment)

MPC
Environmental
charge
P2

P1

P*1
Demand

0 F2 F1 Flow

Figure 8.3 The Pigouvian environmental tax

BUTTON 9781786435668 PRINT.indd 288 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 289

same as when emissions are perfectly correlated with traffic flow. It is deliberately
drawn this way, however, to allow the discussion to move to the need to target
policies and to measure their impacts in relation to the externality involved. For
example, high fuel taxes may be a good way of reducing certain emissions, such as
CO2, but the ability to change vehicle types means that traffic flows may be unaf-
fected. For handling congestion problems, as we see in the next chapter, there is a
need for direct measures affecting trip-making.
Pigouvian-style charges, while not common, are increasingly being used in
transport, the best-known example being the differential tax on leaded petrol
that virtually all industrialized countries outside the United States adopted in the
1980s, but some airports also vary landing/take-off fees according to the noise
nuisance aircraft create.
While the basic concept is not difficult to appreciate, it has been refined and
argued over since the 1920s. Perhaps the most important of these debates reflects
the fact that it is the authorities who impose the Pigouvian tax who are the main
beneficiaries of them. Unlike tradeable permits where markets allow for compen-
sation through the pricing mechanism, there is no such automatic structure with
charges because the revenues collected go to government. It is then a political
matter regarding who subsequently enjoys them. Historically, it has often been
debates over the use of revenues that have been the main obstacles to the use of
environmental charging.
There are also problems in the calculation of the optimal pollution charge or
price, since it is necessary to have reliable information about the MPC and MSC
curves. As we have seen in Chapter 6, knowledge in this area is scant and although
the use of, for example, hedonic house price indices may shed some light on the
monetary importance of noise nuisance, they are far from perfect and normally
offer little insight into the shape of the curves involved or the responsiveness of
people to financial incentives.
At a rather more theoretical level, it is possible to question whether the
polluter-pays principle is being correctly applied in Figure 8.3. We have implicitly
assumed that the road users should buy the right to pollute in the area, but this
could be turned on its head, and the proposition presented that road users should
buy the right to relatively clean air, that is, the road users should be paid a subsidy
of an amount equal to the emissions charge (P2P*1) depicted in the diagram to
curtail their activities. Again, as with the Coase case, the question is essentially a
moral–legal one involving property rights; although there may also be practical
considerations involving the administrative costs of introducing either prices or
subsidies, these should also be considered.
As William Baumol and Wallace Oates (1988) have stressed, one of the prob-
lems of charging polluters is that information about the MSC curve is imperfect,
and that, even if some initially arbitrary price is charged, there is no indication of
whether this is too high or too low. The usual ‘trial and error’ method of pricing
used in industry is, therefore, not appropriate.
Since information about the MSC curve is necessary for virtually all
optimal containment of noise and emissions, irrespective of the method used,

BUTTON 9781786435668 PRINT.indd 289 27/04/2022 11:07

 290 TRANSPORT ECONOMICS, 4TH EDITION

Baumol and Oates argue in favor of pricing on the grounds that it will cause
fewer distortions than other policies. Their arguments found favor with the
OECD, which maintained that ‘[t]he costs of these measures (to ensure that the
environment is in an acceptable state) should be reflected in the cost of goods
and services that cause pollution in production and/or consumption’. Figure 8.4
considers two modes of transport, trucking (A) and railways (B), and relate the
marginal net private benefits associated with using each mode to the noise nui-
sance emitted.
These curves are unknown to the authorities but it may be decided that it
would be beneficial to do something about noise pollution rather than leave it at a
high level. In these circumstances one may wish to reduce noise emissions by say
15 percent, and to use polluter charges to achieve this. Baumol and Oates demon-
strate that a uniform charge on both road haulage and railway noise is the appro-
priate ‘second-best’ policy to pursue. In the figure the marginal abatement costs
(MAC) of noise emissions by one unit for each of the two modes are plotted.
These curves are not known with any degree of exactitude to policy-makers. A
mandatory reduction of 15 percent for each mode would result in emissions being
reduced to A and B for trucking and railroads respectively.
While the desired objective has been satisfied, what one sees is that the MAC
costs involved differ as between the modes – they are higher for road than for
rail. It would be more cost-effective to reduce noise by a greater amount on the
railways than on roads quite simply because it is cheaper per unit to cut emissions
in the former. A noise emissions charge of P per unit per decibel would automati-
cally achieve the aimed-for improvement because it would be more of an incentive
to cut pollution where it is cheaper to do so (a level B† for rail) and have smaller
reductions where the costs of abatement are higher (to A† in the case of road
haulage).
The Baumol and Oates argument highlights that if a standard is to be aimed
for, then the most efficient way of attaining it is by using fiscal instruments. One
can see the importance of this in many transport contexts. Because of its very
nature much transport activity is a mobile source of environmental intrusion, but
the domains in which it operates often differ in their sensitivity to its presence.

$ $
MACa

MACb

0 A A† Emissions 0 B† B Emissions

Figure 8.4 The advantage of fiscal instruments when there is uncertainty

BUTTON 9781786435668 PRINT.indd 290 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 291

Different airports, for example, because of their locations and prevailing wind
conditions, impose different noise envelopes on their surrounding populations.
The actual physical noise associated with any aircraft type may, therefore, impose
different costs at different airports.
To set a standard that all aircraft should reduce noise levels by a specified
amount would thus be inefficient. A charge, on the other hand, which would bring
about the same overall noise reduction would give the flexibility to airlines to use
their quieter aircraft at locations where noise is a major nuisance and their older,
noisier ones where the problem is less severe.
The polluter-pays principle tends to be favored by many academics but it is
not without its critics. Clifford Sharp (1979), for example, questions the distribu-
tional implications and argues that in some instances an environmental improve-
ment may be obtained as efficiently by means of progressive taxation without the
possible regressive effects of pollution charges. Essentially, the argument revolves
around the fact that the benefits from any environmental improvement are closely
related to income. A poor person would probably have a preference for no pol-
lution charges (and ipso facto lower final money prices) than a wealthier person
whose marginal utility of income is lower. Hence, from a distributional point of
view, a subsidy of P2P*1 in Figure 8.3 to the airport to suppress aircraft noise,
financed from a progressive taxation system, will have the same environmental
effect but none of the regressive features of the pollution charge.
While hardly common, pollution charges have been used in transport with
some success. One clear illustration of where fiscal incentives (in this example,
coupled with regulation) have proved particularly effective has been in reducing
the levels of lead (Pb) pollution. Many countries have introduced significant
tax differentials between leaded and unleaded gasoline but, equally, many have
also initiated regulations regarding the fuels that can be sold. In particular, the
banning of normal gasoline (providing the tank capacity for garages to stock
unleaded fuel and leaving only the more expensive super) has effectively further
reduced the real choice open to most automobile users in the Netherlands,
Switzerland, the United Kingdom, and Germany. The combined impact of
these measures in the United Kingdom was a rise in vehicles using unleaded
gasoline from 0.1 percent of the car park in March 1988 to 25.9 percent in
October 1989. Similarly, in the pre-unified Federal Republic of Germany the
percentage of automobiles using unleaded rose from 11 percent in 1986 to 28
percent in 1987.
To date, however, fiscal policies have tended to be piecemeal, usually focusing
on modes of transport rather than directly on transport externalities and, gener-
ally, with the explicit objective of reducing the effect of different modes rather
than optimizing them. This suggests that the social objective of government
policy has been one of satisficing rather than optimizing although it has been
argued that the actual effect of some regulations has been excessive.
Schwing et al. (1980), for instance, suggested that the United States Clean
Air Act of 1970 imposed car exhaust emission levels that were far too strin-
gent, with a consequential welfare loss. Table 8.2 presents the results of their

BUTTON 9781786435668 PRINT.indd 291 27/04/2022 11:07

 292 TRANSPORT ECONOMICS, 4TH EDITION

cost–benefit study. While the high benefit estimate suggests some welfare advan-
tage from the Act, the underlying assumptions required to reach this conclusion
are deemed very unrealistic. Optimal levels for toxic exhaust emissions were esti-
mated to be 0.73 percent, 0.31 percent, and 0.82 percent control for NOx, CO,
and HC respectively.
The acceptability of this type of work would be questioned by environmen-
talists in terms of both the items included in the cost–benefit calculations and the
valuations placed upon them. The problem, common to most studies of envi-
ronmental aspects of transport, is the inadequacy of knowledge both about the
actual physical impact of the various external effects generated by transport and
about the values society places upon them. Until some clear understanding of
these matters is obtained it is difficult to see how the external effects of transport
are likely to approach the optimal level.

Cap-and-trade

Cap-and-trade emission control policies are a hybrid between a strict market

approach along Coasian lines and the setting of strict standards. A maximum
amount of emission is set based upon political and scientific considerations and
then this is allocated by the buying and selling of permits within this constraint.
One of the aims of this is to introduce more information into the market, the
ceiling, and to thus add more stability.
An example in the context of transport is the European Union Emissions
Trading System (EU ETS). This started in 2005 and includes over 11,000 energy
and manufacturing plants in Europe with a net heat in excess of 20 MW account-
ing for 41 percent of the Union’s greenhouse gas emissions. From 2012, parts of
the aviation sector were included in it. These covers flights where both take-off and
landing are within the EU. Initially, the Union planned to include all flights even
if the take-off or landing was outside of its geography; this is referred to as ‘full
scope’. But after protests from mainly the United States and China, the system
was modified to a reduced scope, which ‘temporarily stopped the clock’. The
allowances under the EU ETS specific to airlines are European Union Aviation
Allowances (EUAA). If airlines need additional allowances, they can also pur-
chase regular European Union Allowances. EUAA issued for 2013–20 corre-
sponded to 95 percent of the emissions of 2008–12. From 2021 onwards, there is

Table 8.2 Impacts of the United States Clean Air Act of 1970

Benefit % Control level Benefit Cost Net benefit

estimate $(109) $(109) $(109)
HC CO NO

Low 0.98 0.97 0.94   9 65 256

Prime 0.98 0.97 0.94 34 65 31
High 0.98 0.97 0.94 102 65 37

Source: Schwing et al. (1980).

BUTTON 9781786435668 PRINT.indd 292 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 293

to be a linear annual reduction until 2064, when no additional EUAA allowances

will be issued. The last European Union Allowances would be issued in 2057.

Carbon Off-setting

A carbon off-set is a reduction in emissions of carbon dioxide or other greenhouse

gases made in order to compensate for emissions made elsewhere. For example, a
shipping company could off-set the carbon generated in its activities by tree plant-
ing, supporting wind farms and biomass energy, constructing biogas digesters, or
building hydroelectric dams. The off-set, providing it reduces carbon emissions by
the same amount as that generated by transport, meets the polluter-pays principle
irrespective of the form it takes. That, however, inevitably raises questions over
the imposition of other forms of externalities – for example, wind farms are not
seen as aesthetically pleasing to all. The Kyoto Protocol has sanctioned off-sets as
a way for governments and private companies to earn carbon credits that can be
traded on a market-place.
The idea, however, that buying an off-set will instantly erase transport’s
carbon footprint is overly simplistic (Becken and Mackey, 2017). Newly planted
trees take years before becoming significant carbon absorbers, and that is only
when forests stay protected for long enough to reach that point. Such cases where
the carbon off-set is sold before it has been achieved are known as ‘forward
crediting’. Furthermore, many projects entail monoculture forestry with just one
species being planted. This supports very little biodiversity and is more suscepti-
ble to fires and disease.
The accounting process, translating emissions reductions from projects to
carbon off-sets that can be sold, is complex. For example, estimations of emis-
sions from air travel are usually generalized across types of aircraft, fuel use,
routes, and other factors, because it is not feasible to measure the emissions of
specific flight. In forestry projects, the amount of carbon absorbed varies widely
depending on the species of tree, and climate and soil conditions. Implementing a
project can also release emissions, which needs to be accounted for to ensure that
only net reductions are claimed.
Any off-set project should have ‘additionality’. The project would not have
happened without the off-sets. Some emission-reducing activities naturally occur
in compliance with existing regulations, or because they are profitable invest-
ments. Ensuring the off-set plays an indispensable role in whether or not a project
is implemented is problematic; it deals in hypotheticals.
The project’s carbon reductions should be permanent, without causing any
eventual emissions that reverse the original effects. For instance, while planta-
tions store carbon, they run the risk of releasing it if the trees end up being cut
or burned down. Some projects keep a buffer reserve of off-set credits, which are
used to compensate for any credits that are inadvertently voided. They can also
mitigate risks from the outset to reduce chances of a reversal occurring.
Voluntary carbon-set programs have been adopted by a number of major air-
lines, and have been put forward by the International Civil Aviation Organization

BUTTON 9781786435668 PRINT.indd 293 27/04/2022 11:07

 294 TRANSPORT ECONOMICS, 4TH EDITION

(ICAO) as its main instrument for reducing carbon emissions. For example, in
Europe, British Airways has a scheme allowing passengers to buy credits for pro-
tecting threatened forests, and EasyJet is engaged in off-setting projects in Peru
and Ethiopia. In the United States, Delta uses avoided deforestation projects as
an off-set. But there are issues with these schemes.
The ICAO’s Carbon Offsetting and Reduction Scheme for International
Aviation (CORSIA), set in train in 2016, has the goal of carbon neutral growth
from 2020. Being phased in from 2021, aircraft operators should purchase carbon
credits from the carbon market. The scheme is voluntary for all countries until
2027. As emissions below 2020 levels are grandfathered, CORSIA regulates 25
percent of aviation’s international emissions applying to international flights as
representing 60 percent of aviation emissions.
Economic comparisons between aviation off-set programs and cap-and-
trade policies (such as the EU ETS discussed earlier) have provided some ex ante
assessment as to their potential effects. Taking Sweden as a case study, Larsson
et al. (2019) examined the possible effects of these policies on air travel emis-
sions from 2017 to 2030. Their analysis shows that when emissions reductions in
other sectors are attributed to the aviation sector as a result of the EU ETS and
CORSIA, carbon emissions are expected to reduce by –0.8 percent per annum.
However, if non-CO2 emissions are included in the analysis, then emissions will
increase. This is less than what is needed to achieve the 2°C reduction target.
When looking at both domestic and international flights, and at total emissions,
not just the increase, it has been estimated that 12 percent of aviation emissions
will be off-set due to CORSIA in 2030 (Scheelhaase et al., 2018).

8.5 More on Environmental Standards

Figure 8.3 indicated that charging a pollution price might optimize external costs.
It is equally possible, however, that rather than operate through the pricing mech-
anisms the desired, environmentally optimal level of traffic could be obtained by
using command-and-control instruments; that is, the establishment of legal or in
some cases voluntary standards covering pollution, noise, and safety.
The setting of environmental standards is long established, as the oft-cited
night-time banning of chariots in classical Rome illustrates. Their use is now
extensive and covers such wide-ranging things as the establishment of ‘noise
abatement zones’ in the United Kingdom, and the controls embodied in a series
of Road Traffic Acts that have, since 1973, laid down regulations regarding car
silencers and exhausts. The CAFE standards define the fuel efficiency of new
vehicles in the United States (see Section 8.8) and most industrial countries now
insist gasoline-powered cars are fitted with catalytic convertors and use unleaded
fuels. There are controls over where oil-tankers may clean their tanks and when
aircraft may take off. Individuals are required to wear seat belts in many countries
and the use of cell phones while driving is often illegal. The list of standards and
regulations is long and continues to grow.

BUTTON 9781786435668 PRINT.indd 294 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 295

They also change over time, as we see later regarding vehicle fuel efficiency.
As another example, noise standards were introduced in the United Kingdom at
the manufacturing stage for new lorries in 1970 with limits of 9l dB(A) for vehi-
cles with engines over 200 hp and 89 dB(A) for less powerful lorries, while from
March 1983 new vehicles coming into production had to meet more stringent
requirements of 88 dB(A) and 86 dB(A) respectively. Supranational legislation
has gained in importance recently with the European Union setting vehicle noise
limits from the early 1990s. The Civil Aviation Act of 1971 laid down regulations
about night movements over built-up areas in the country, and specifies overfly
patterns for aircraft. The speed limits operative on roads in virtually all countries
are primarily designed to reduce accident risk – with some supplementary posi-
tive effects, on occasions, on fuel economy.
The compulsory wearing of seat belts in many countries is also to reduce
accident costs. Similarly, the periodic testing of vehicles and the licensing of
lorries, aircraft, etc. are to ensure that minimum safety and environmental stand-
ards are achieved. In many overseas countries the regulations are more stringent
(for example, the removal of lead from petrol in the United States, and stricter
annual checks on pollution emissions from internal combustion engines in states
such as New Jersey and Oregon) or take different forms (such as airbags in cars
in the United States), but their intended effect is the same: to reduce the marginal
environmental, including external accident, costs of transport.
While all the above represent physical regulations to contain pollution, they
should strictly be divided between those controls that act directly to contain
the externality (for example, noise emission legislation), the third component
of Figure 8.1, and those that control transport in such a way as to reduce the
external costs (for example, lorry routes and aircraft flight path regulations). The
effects of these alternative broad sets of physical controls are not the same.
Actual emission standards act directly to limit the external effects per-
mitting other characteristics of operations to be adjusted freely. Operational
regulations impose much more stringent controls, severely limiting the alterna-
tive courses of action open to the operator. With noise emissions standards for
aircraft flying over an area, for instance, an airline can either conform and pay
the costs of suppressing noise or avoid the area in question; with operational
controls, the latter option is available. This point should be borne in mind during
the more general discussion of physical controls, which follows. We shall return
to the question of operational restrictions, traffic calming, and vehicle routing in
more detail later.
One justification for adopting command-and-control methods is that they
can have lower administrative costs. For example, the imposition of a zero lead
content requirement for gasoline is relatively easy to understand and to enforce.
It may not be optimal, and the effects across different users may not be fully effi-
cient, but the transaction costs are relatively small. Allowing different individuals
to have different amounts of lead in their fuel according to, say, their willingness
to pay is quite clearly going to be difficult to price and to administer. In more
strict economic terms, the transactions costs of pricing à la Pigou, or marketable

BUTTON 9781786435668 PRINT.indd 295 27/04/2022 11:07

 296 TRANSPORT ECONOMICS, 4TH EDITION

permits along Coasian lines, are just not cost-effective. But there may be other
reasons for adopting standards.
While in the simple case illustrated in Figure 8.3 the effect of an optimal
standard produces an identical level of road transport activity (and, ipso facto,
environmental intrusion) to an optimal pollution charge, it can be argued that,
with more realistic assumptions, the pricing approach offers a superior solution
to the externality problem even where administrative costs are similar.
When information about the exact shape of the MEC curve is poor, the
use of standards is demonstrably less efficient than the Baumol–Oates charging
approach seen in Figure 8.4. If, to achieve the 15 percent reduction in transport
noise used in our example, both road and rail were compelled to cut their noise
emissions (that is, to 0A† and 0B† respectively), then it is clear from the diagram
that the marginal net private benefits generated by the two modes are no longer
equal (at the new emission levels, MACa > MACb). Consequently, social welfare
could be improved by lowering the standard for road haulage and increasing it
for rail. Unfortunately. in the real world, lack of perfect knowledge of the MAC
curves means that the optimal differentiation of standards is likely to be impos-
sible to define. Thus, in this imperfect situation, the polluter-pays principle is
almost certainly going to prove superior to the use of emission standards.
It is also probable that pollution pricing will prove more flexible than
­standards. While transport infrastructure may impose external costs of visual
­intrusion it is normally the mobile unit that generates the greatest external
costs. Given the differing income levels and preference patterns in various parts
of a country, one could re-interpret the MAC curves in Figure 8.4 in terms of
the marginal abatement associated with a single mode, but operating in dif-
ferent parts of the country. In this case the uniform emissions charge would
be both theoretically superior and, in addition, reduce the costs to transport
­undertakings of reducing their noise emissions. The imposition of different
standards for each area means that operators must either ensure that vehicles
moving between areas conform to the most stringent standards, or have specific,
variously suppressed vehicles designed to conform to local regulations. Both
options are likely to be wasteful. With a charging regime, the operator can select
a vehicle mix that ­minimizes their overall costs of operation: vehicles numbers
may be suppressed or pay the emissions price, or they may be subjected to a
combination of the two.
Moving to a more dynamic situation, where technology is variable, Maler
(1974) suggested that pollution prices have important advantages over regula-
tions for the encouragement of a rapid adoption of cleaner technologies. His
argument rests upon the implicit assumption that transport suppliers, when con-
fronted with either a pollution price or emissions standard, assume this price or
standard to be fixed in the medium term irrespective of their individual actions.
Consequently, they will always assess the benefits to themselves of adopting
new operating methods or technologies against existing prices or standards. In
Figure 8.5 we show the marginal private costs of reducing exhaust fumes for a
truck operator confronted with the existing technology (MC1) and with the new

BUTTON 9781786435668 PRINT.indd 296 27/04/2022 11:07

 297
CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­

technology (MC2). The MC2 curve is inside MC1 because it is cheaper to suppress
vehicles with the new technology at all noise levels. On the assumption that the
authorities have full information on MEC and can, therefore, define the optimal
level of traffic noise we see that either a pollution charge of 0P or a standard of
0C will ideally be in force.
If the pricing policy is pursued, the trucker will find it financially worth-
while to quieten their vehicle by CD1 (costing CD1B) and pay 0CBP in charges.
With a standard, they pay no pollution charges, but it costs them CBD1 to
conform to the noise regulation. However, if the new technology is available,
an individual trucker will perceive, ceteris paribus, the benefits of adopting it as
ABD1D2 if there is a charging policy operative, that is, the trucker will reduce
their emissions with the new technology to 0H (costing AHD2) and pay charges
of 0PAH. The incentive to adopt the alternative technology with the emissions
standard is only D1EBD1, that is, the cost of conforming to the standard with the
new technology rather than the old. Thus, the pollution charging policy offers an
incentive of the shaded area, ABE, in excess of an emissions standard to move to
the cleaner technology.
One possible option is a combined environmental tax/standards approach
whereby all vehicles are obliged to meet a set standard and there is a scale of
­emissions-related ‘fines’ for vehicles that exceed this. If the standard were rigorous
and well below the existing level of emissions (that is, consistent with the optimal
level of pollution with the cleaner technology in our example above), then this
would be as effective as the pricing approach and at the same time offer a firm
target for vehicle operators to aim at. Such a tax/standards approach may, however,
be particularly appealing at the vehicle manufacturing stage where new technology
can most easily be injected into the transport sector in a gradual manner.
Cost to motorists of reducing

MC1
exhaust emissions

MC2

A B G F
P

0 H C D2 D1
Exhaust emissions

Figure 8.5 Pollution charges, emissions standards, and technical change

BUTTON 9781786435668 PRINT.indd 297 27/04/2022 11:07

 298 TRANSPORT ECONOMICS, 4TH EDITION

8.6 Transport Subsidies and the Environment

To try to stimulate the greater use of less ‘polluting’ transport, in the broadest
sense, there has traditionally been a widespread use of subsidies. These are politi-
cally attractive, largely because they involve a diffuse contributor base but a focused
receptor base. We exclude here the plethora of subsidies that are designed primarily
to meet social objectives, such as ensuring acceptable levels of mobility, meeting the
needs of the physically disadvantaged, and providing access to remote regions, but
are concerned explicitly with subsidies aimed at changing travel patterns with the
objective of encouraging more environmentally benign forms of transport.
In practice, however, it is important to note that these are not normally
‘Pigouvian subsidies’ that are explicitly paid to individuals to desist in generat-
ing external costs, but, rather, indirect subsidies to encourage the adoption of
other activities that are associated with lower levels of external or other adverse
economic impacts. In terms of surface, personal travel the conventional wisdom
is that an efficient public transport system with adequate load factors is more
energy-efficient than the automobile.
There are general issues regarding subsidies of any kind, such as whether it
is reasonable to use taxes collected from the general public to essentially subsidize
public transport and car users, and whether it is possible to have an efficient subsidy
regime that is not highly X-inefficient and captured by the ­transport-providing
agencies and their employees. But from a pure efficiency perspective, for the types
of subsidies given to stimulate less environmentally intrusive modes of transport,
there needs to be a reasonably high cross-elasticity of demand between modes for
public transport subsidies to be effective. We shall return to this later.
Thus, an alternative to operating directly upon the transport undertaking
generating externalities (either pollution or congestion) is seen to be the offering
of a carrot to transport users to switch to more socially desirable modes. This
line of reasoning has been widely used as a partial justification for the large sub-
sidies given to support the railways and urban transport services. In the United
Kingdom, the Railways Act of 1974, for example, permitted government grants
of up to 50 percent of the costs to be paid to British Rail customers for the instal-
lation of sidings and the provision of rolling stock following an assessment of
the environmental harm of the lorry movements which would be avoided if the
investment concerned went ahead. The 1968 Transport Act initiated a system of
centrally and locally financed public transport operating and capital subsidies
(the latter of which have since been abandoned), with the objective of containing
the growth in private motor traffic in large urban areas.
In the United States, net federal subsidies from 1990 to 2002 meant that for
every thousand passenger-miles, transit got $118 in subsidy, although not all of
this was justified on environmental grounds. In addition, there were extensive
state and local support initiatives for public transport, although their motivations
have not always been clear.
In a perfectly competitive world, there would be no justification for this type
of policy, but in a situation where marginal cost pricing is not universal and where

BUTTON 9781786435668 PRINT.indd 298 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 299

political expedience leans against the introduction of measures such as road

pricing, subsidies may offer a pragmatic second-best approach to the externality
problem.
Where the cross-elasticity of demand between transport and other goods
is negligible and the overall demand for transport is totally inelastic – a situ-
ation not unrealistic in the context of commuter travel in many large urban
areas – the optimal subsidy to a zero externality-generating transport mode will
have the same effect on the use of an externality-generating mode as pollution
charges.
In Figure 8.6, the overall demand for transport, which is supplied by a com-
bination of private cars and a light rapid transit system, is fixed. Cars have associ-
ated with their use external costs that can be expressed as the difference between
the MPCC and the MSCC. The light rapid transit system has no such externali-
ties associated with its use (that is, MSCLRT = MPCLRT) and for simplicity we
assume it is a constant cost form of transport. The free market outcome, where
no cognizance is taken of externalities, will be a division of traffic at point 0Q† in
the diagram. The optimal solution is a split of Q*, which may be brought about
either by charging a Pigouvian pollution charge of the level indicated in the figure
or, alternatively, by subsidizing the public transport mode by an identical amount.
The modal split effects are the same.
If aggregate demand is not perfectly inelastic then the optimal subsidy
is more difficult to define, although it may still offer a second-best solu-
tion to the externality problem. Figure 8.7 shows the cost conditions for the
­externality-generating mode (the car) with a demand curve for its services of DC.
It also shows the cost (CPT) and demand (DPT) for the public transport mode. A
subsidy for the public transit mode will cause DC to shift to the left (say to D'C),
Generalized costs
Generalized costs

MSCC
MPCC
MPCLRT = MSCLRT

Charge Subsidy

0 Q* Q† 0
Automobile Light rapid transit

Figure 8.6 O
 ptimal subsidy to a non-polluting substitute mode with fixed aggregate
demand for travel

BUTTON 9781786435668 PRINT.indd 299 27/04/2022 11:07

 300 TRANSPORT ECONOMICS, 4TH EDITION

reducing the dead-weight welfare loss associated with the initially suboptimally
high level of car usage. Unless, however, the demand for car use is pushed so far
left that it intersects the cost curves in some areas where MPCC = MSCC, a dead-
weight loss will remain.
As with Figure 8.6 the subsidy itself will also result in some loss of
welfare. The fall in the cost of car use as people switch to public transport will
have the effect of pulling back the demand for the latter to D'PT and thus reduce
the amount of funds needed to fund the optimal subsidy. The ultimate cost of the
subsidy is the amount per transit user times QPT, which exceeds the area of net
consumer surplus enjoyed at the subsidized fare level by wxy. If the ­dead-weight
loss saving associated with the reduced level of demand for car use is abcd (that
is, the area between MSCC and MPCC as DC shifts to the left), then the optimal
subsidy can be defined as that which maximizes the difference between the
gain from reducing the dead-weight loss of excessive congestion minus the effi-
ciency cost of the subsidy. In the diagram, this is the subsidy that will maximize
(abcd – wxy).
The practical difficulty with this approach is that the optimal subsidy may be
extremely large and, theoretically, if the cross-elasticity of demand between modes
is low, may even result in negative fares. The use of public transport subsidies in
urban areas has been questioned for this very reason. This is a topic that we
covered in Chapter 4, where it was found that price cross-elasticities between bus
and car are generally low, with service cross-elasticities somewhat higher.
Specifically, early survey work by Michael Kemp (1973) on modal transfers
found that in general the direct fare elasticity for urban public transport was low
(–0.1 to –0.7), suggesting that substantial subsidies are necessary to attract pas-
sengers to public transport irrespective of whether they constitute new travelers
or those diverted from private cars. Baum’s (1973) work looking at the possible

$ $

MSCC
d

w y
MPCC
a MPCPT
Sub
c MPCʹPT
DʹC
b x
DC DPT
DʹPT

0 Q'C QC 0 QPT Q*PT QʹPT

Users of cars Users of public transport
Figure 8.7 Optimal subsidies with elastic aggregate demand for travel

BUTTON 9781786435668 PRINT.indd 300 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 301

effects of offering free urban bus transport was even less optimistic, his analysis
yielding elasticities in the range from –0.1 to –0.4 for Britain, the United States,
and West Germany. Later, after surveying some 50 empirical studies of both long-
and short-term cross-elasticities, however, Goodwin (1992) concluded that, in the
short term, bus demand remains inelastic enough to make revenue-raising by fare
increases an effective policy, but demand increases by fare reductions limited. But
in the longer run the effectiveness of the first policy is reduced, and that of the
second is increased.
Finally, De Witte et al. (2006) compared the introduction of ‘free’ public
transport for students at Brussels’ Flemish-speaking colleges and universities
in the 2003–2004 academic year with French-speaking universities and colleges in
the city. They found an increase in public transport use among students benefit-
ting from the measure, but also that the group of non-benefitting students not
only outnumbered the benefitting students when it came to using public trans-
port, the STIB, but also used it more frequently. By contrast, the train, which is
not free, is used more frequently by Flemish-speaking students. This is somewhat
counterintuitive to the introduction of ‘free’ public transport for the latter. It
appears that there are many other factors affecting the uptake of free goods than
just the zero price.

Exhibit   The ‘Ubernomics’ of app-based ride-hailing

Options available for personal urban mobility increased with the arrival of on-demand,
app-based ride-hailing (often called ride-sourcing) in the 2010s. The computer platforms of
Uber, Lyft, DiDi, etc. allow taxi-like services by connecting consumers to nearby drivers.
Initially, this matched car owners with spare seats to customers who needed to travel to
places poorly served by public transportation or conventional taxis. Subsequently, individuals
have acquired vehicles specifically to be drivers and to meet the demands of app-based
vehicle-hailing markets.
Ride-sourcing drivers are largely attracted to the occupation because of the flexibility it
offers, the level of compensation, and because they can adjust their work hours to smooth
flows of income. They are also more similar in terms of their age and education to the
general workforce than to taxi drivers and often had full- or part-time employment before
joining. Drivers and riders also often operate in two ‘homes’ by using several ride-sourcing
platforms. This can minimize the ‘request to pick-up time’, thus offering gains to both
parties.
A customer is paired with the nearest driver and can actively track a vehicle once hired.
Payment is automatically made through the app. The transportation network companies
(TNCs), such as Uber, do not own vehicles, provide training, pay driver expenses,
provide insurance, or accept liability. They market themselves as digital platforms that link
independent contractors with customers.

BUTTON 9781786435668 PRINT.indd 301 27/04/2022 11:07

 302 TRANSPORT ECONOMICS, 4TH EDITION

Ride-sourcing services are often seen as competitors to taxi-cabs that have a history of
being highly regulated with respect to driver qualifications, fares, etc. These regulations
meant that much of the time cabs were idle as fares were sought. There was a divergence
between the rides being offered and useful rides being made available at locations where
potential riders were seeking service. App-based ride-hailing services, being a part of the
‘gig economy’ often involving part-time work, added greater flexibility to the supply of rides,
reducing mismatches between the positioning of drivers and ride seekers.
Increased revenue comes from product differentiation (for example, premium, shared,
and food delivery services). This allows a wider range of vehicles to enter the market and
more users with diverse requirements to be served. With some drivers having vehicles that
can provide a range of services, this generates economies of market presence by speeding
up the matching of rides to riders. TNCs also make use of ‘surge pricing’ to attract more
drivers at peak periods and to ration demand.
Considerable economic/legal debate has surrounded the role of drivers in Uber types of
markets regarding whether they are self-employed or work for Uber. In one sense the
drivers act independently, as is the case in much of the taxi-cab industry, but there are often
different legal responsibilities. TNCs have sought to disclaim employee status, depriving
drivers of social insurance among other benefits, and to deny liability to third-party victims
for damages due to auto accidents or sexual assaults arising out of their business.
There were initial thoughts that ride-sourcing would reduce the demand for driving. The
evidence, however, is that, whatever efficient gains there are in terms of vehicle utilization,
the increased overall demand for shared rides generally results in more congestion and this
is focused in times and space. Traffic peaks are now often in the evening in areas of social
activity rather than daytime jobs or shopping. The traffic system has not traditionally been
designed to cope with this.
There are safety and security issues associated with ride-sourcing. It was anticipated that the
probable uptake of ride-sourcing by individuals who would previously drink and drive would
decrease accidents. Comparing accident trends in areas of California with and without a
TNC presence, one study found ride-hailing availability significantly reduced traffic-related
fatalities for basic services. A wider American study, however, found the staggered arrival
of ride-sourcing across cities to be associated with a 2 to 4 percent increase in fatalities,
although this study made no allowance for a parallel fall in fuel prices on traffic demand. A
study of Austin, Texas revealed a fall of 17 to 40 percent in accidents after Uber had been
operating for four years.
See also: K.J. Button (2020) The ‘Ubernomics’ of ridesourcing: the myths and the reality,
Transport Reviews, 40, 76–94; and J.D. Angrist, S. Caldwell, and J.V. Hall (2021) Uber versus
taxi: a driver’s eye view, American Economic Journal: Applied Economics, 13, 272–308.

BUTTON 9781786435668 PRINT.indd 302 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 303

8.7 Protecting the Sufferers

The strategies examined so far have relied upon either forcing the generator of
externalities to change their production process or encouraging the adoption of a
different method of operation. We have only touched upon the idea of insulating
the public from environmental intrusion (that is, in the context of aircraft landing
path controls and lorry routes). Insulation in the short term may be achieved
either by directing traffic away from sensitive areas or by physically protecting
people and property (for example, with double-glazing for sound insulation),
while in the longer term new investment permits a more efficient separation of
transport from those sensitive to its wider impact (the main reason for the official
rejection of the Roskill Commission’s recommendations on the siting of a Third
London Airport in the 1970s was that an inland site would be excessively damag-
ing to the country’s environment – see Chapter 11).
Fairly typical of the approaches favoring command-and-control approaches
was the Armitage Report produced 40 years ago in the United Kingdom, which
went as far as to recommend the establishment of ‘Lorry Action Areas’ to
protect residents living in a limited number of areas but who suffer from the
worst environmental effects of road freight transport. Specifically, such areas
would involve:

• the installation of double-glazing in houses, which would reduce consider-

ably the major problem of noise in homes;
• grants for repairs to houses physically damaged by lorries;
• maintaining road surfaces to high standards, which would reduce vibrations;
• minor road improvements to reduce accidents and to reduce noise by, for
example, the use of noise-absorbing road surfaces;
• the building up of pavements or the erection of bollards to reduce the prob-
lems of vehicles cutting corners and of damage to buildings through physical
contact; and
• in the worst cases of intense local nuisance by a specific generator of lorry
traffic, compensation for discontinuance action taken by a planning author-
ity in respect of a site with planning permission or existing use rights.

The difficulty with protective options, both long- and short-term, is that their
effects often extend beyond simply protecting sensitive groups in the community
and their overall cost may be considerable. Limiting the flight paths of aircraft
can both increase the risk of accident (by forcing the adoption of less safe climb-
ing and turning patterns) and increase the cost of operations (especially energy
costs). Similarly, lorry routing both necessitates higher infrastructure costs and
often leads to longer trip distances.
In the long term it should, theoretically, be easier to design the spatial
economy so that transport’s effect on the environment is significantly lower.
Options include:

BUTTON 9781786435668 PRINT.indd 303 27/04/2022 11:07

 304 TRANSPORT ECONOMICS, 4TH EDITION

• sterilization of land between nuisance and dwellings;

• use of non-sensitive buildings (for example, light industry) as barriers
between nuisance and sensitive areas;
• design dwellings so that little-used rooms face the nuisance rather than living
rooms or bedrooms; and
• make use of self-protecting developments – for example patio-style housing –
to reduce intrusion.

Such designs obviously generate additional costs and only provide a partial solu-
tion to the environmental problem. Like most of the shorter-term protective
measures they only ameliorate those aspects of environmental costs inflated while
people are at home. Land-use planning may offer some limited protection at other
times especially in the reduction of accident risk – but it is unlikely to separate
transport completely from the non-traveler.
One should also perhaps include under protective measures the notion
of traffic calming. This entails the use of such things as road humps, speed
tables, raised junctions, reduced carriageway widths, and ‘changed road sur-
faces’ both to slow traffic flows and to encourage the use of particular ‘suit-
able’ links in the network or alternative modes. Essentially the idea is to make
streets more attractive and liveable. In Europe, traffic calming has tended
to come about as part of wider packages and has often been tied to legal
speed limits of 30 kph or, in the Netherlands, to a walking pace limit. The latter
is part of the country’s Woonerven – where all road users have equal rights to
road space. In the United Kingdom, traffic calming schemes have come about
more as part of a policy to reduce urban traffic speed for safety reasons – about
70 percent of schemes have this as a primary objective. Evidence from Germany
suggests that serious casualties fell by up to 50 percent in areas where it has been
introduced, for example by 44 percent in Heidelberg after traffic calming was
initiated.
There are also some positive environmental externalities associated with
some types of urban design. Obesity and many forms of illness are associated
with a lack of physical activity. The way that cities are built and the way individu-
als move about in them influence the amount of physical activity that citizens
‘enjoy’, and lack of facilities within walking or cycling distance necessitates the
use of mechanized transport with minimal physical activity. The design of urban
areas can thus be perceived as a way of helping to counteract some of the adverse
effects on personal well-being by fostering a healthier life-style.

8.8 Energy Use

We now spend some time looking in detail at some economic aspects of trans-
port energy policy. In many cases the instruments used are from the generic
toolkit outlined above, but the roles they play are of importance because of the
very high correlation between energy use and form and different components of

BUTTON 9781786435668 PRINT.indd 304 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 305

atmospheric pollution. Ensuring that there is no excessive use of a scarce, finite

resource such as oil is also of importance.
There is theoretically a wide range of policy tools that can be deployed to
affect the energy use of transport. Each has its characteristics and usefulness
depending on the background assumptions that are adopted. Here we are selec-
tive and focus on some of the more important efforts that have been made to
influence fuel consumption in transport. Longer-term policies involving land use
and such policies as ‘compact-city’ design are explicitly omitted. These are large
and multi-dimensional topics in their own right and take us beyond the bounda-
ries of a chapter such as this.
Although theoretically there are numerous ways to influence energy use, a
wide range of practical and political factors determine the policies that have been
initiated to influence the energy consumption in transport. In some circumstances
the costs of introducing, monitoring, and policing some policies, as with some
specific environmental policies, simply make them impractical, at least in the
purest forms. In other situations, there may also be trade-offs between improving
energy efficiency and meeting other objectives, such as removing pollutants from
the atmosphere or ensuring an acceptable level of traffic safety. An example of
the former has been the removal of lead from gasoline in many countries, which
reduces the fuel efficiency of internal combustion engines.
The policy tools that are in place, or have been used, to influence the type
of fuel used in transport, as well as the aggregate consumption, are nevertheless
quite extensive. The following is not intended to be comprehensive in its coverage
but rather should be viewed as illustrative of the types of measures that have been
put in place.

Leaving Things to the Market

One policy option that is often forgotten is to leave things to the market. After
all, while there are market failures, there are also government intervention failures
that may either worsen an original market failure or cause serious and unexpected
distortions elsewhere in the system (OECD, 1992).
In practice, the market has been a significant influence on the types and
amount of energy used by transport. Historically, for example, changes in prices
have demonstrable medium- and long-term impacts on overall energy consump-
tion in transport, most of which we have only appreciated in retrospect. Not all of
these, however, have been directly related to the price of fuel. A simple transmis-
sion mechanism illustrates the difficulty of policy-makers trying to foresee energy
changes and plan for the development of new technologies.
At the beginning of the twentieth century, automobiles were expensive and
canals were limited in the their geographical coverage leaving coal-powered (either
directly or after transformation into electricity) railway systems to dominate
surface transport. The energy effect was brought about by the introduction of
mass production of cars, initially by Fiat in Italy, but on a larger scale in 1913 by
Henry Ford in the United States, aimed at gaining a market share in the very highly

BUTTON 9781786435668 PRINT.indd 305 27/04/2022 11:07

 306 TRANSPORT ECONOMICS, 4TH EDITION

competitive automobile industry of the time. This brought down the costs of car
production and subsequently the price of cars (from $910 for a touring Model-T
in 1910 to $367 in 1925). In turn, this led to more use of cars and trucks (sales of
the touring models were 16,890 units in 1910 rising to 691,212 in 1925), with a
resultant switch in transport away from coal as the primary energy source to oil.
In the East German economy of the 1960s, market forces were largely
ignored when policy moves towards greater car ownership at administered prices
were initiated. The resultant centrally planned outcomes were the Wartburg and
Trabant cars, and, by the time the Berlin Wall came down, there was a waiting
time of nearly ten years to receive a not very comfortable, reliable, or efficient
vehicle. The complexity of planning the design and production of cars proved too
much for even the highly skilled planners of East Germany.
Fuel prices themselves are also powerful influences on consumption. Where
there have been shortages of some forms of energy, either because of physical
factors or institutional ones, markets can bring about changes. In the past there
have been shortages of oil for political reasons. While there are short-term adjust-
ment issues, the long-term effect of a fuel shortage and price rise is that fuel is
used more efficiently.
An example is the impact on the fuel efficiency of the United States ‘car park’
after the oil crises of 1973 and 1979 (Crandall et al., 1986). The average energy
efficiency of vehicles (and possibly the way they are driven) rose following both
crises, albeit with a lag as the adjustment took place. A subsequent survey bring-
ing together work on long-term gasoline fuel price elasticities indicates that about
20 to 60 percent appears to be due to changes in the vehicle miles driven, with
40 to 80 percent being due to changes in fleet composition (Parry et al., 2007). A
more general rule of thumb, suggested by Goodwin et al. (2004) after reviewing
numerous empirical studies, is that fuel consumption elasticities are greater than
traffic elasticities, mostly by factors of 1.5 to 2.00.
Energy, because of the relative inelasticity of aggregate demand for its use,
has traditionally been the subject of taxation. In many cases this has been for
purely sumptuary purposes, but in other cases, as with the federally earmarked
gasoline tax in the United States, it has been used as a proxy charge for some
related consumption item; in the United States’ case, to pay for the use of the
road. In other cases, there have been environmental motivations, for example the
differential taxes applied to gasoline and diesel fuels in many countries.

Fuel Taxation

Examples of taxes on the energy used by transport abound, although as men-

tioned earlier the exact motivations underlying them are not always clear. The US
Energy Tax Act of 1979, for instance, was a law passed as part of the National
Energy Act. One element of the act created the ‘gas-guzzler’ tax applying to
sales of vehicles with official estimated gas-mileage below certain levels. In 1980,
the tax was $200 for a fuel efficiency of 14 to 15 miles per gallon (mpg), and
was increased to $1,800 in 1985. In 1980, the tax was $550 for fuel efficiencies

BUTTON 9781786435668 PRINT.indd 306 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 307

of 13 mpg and below, and was changed in 1986 to $3,850 for ratings below
12.5 mpg. The gas-guzzler tax only applied to cars under 6,000 lbs., which made
sports utility vehicles (SUVs) and other large passenger cars exempt.
In terms of using taxation as an instrument for encouraging energy con-
servation, or changes in the energy source used for environmental reasons,
carbon taxes have been adopted in several countries. These are not transport-
specific but are more holistic in their intent of making optimal use of resources
more generally, although their impacts on transport are often large. In 1991,
Sweden, for example, placed a tax of $100 per ton on the use of oil, coal,
natural gas, liquefied petroleum gas, petrol, and aviation fuel used in domestic
travel. Industrial users paid half the rate (between 1993 and 1997, 25 percent of
the rate), and certain high-energy industries such as commercial horticulture,
mining, manufacturing, and the pulp and paper industry were fully exempted
from these new taxes. In 1997 the rate was raised to $150 per ton of CO2
released. Finland, the Netherlands, and Norway also introduced carbon taxes
in the 1990s.
In other cases, however, efforts at introducing such policies have failed. In
2005 New Zealand proposed a carbon tax to take effect from April 2007 and
applied across most economic sectors, but the policy was abandoned in December
2005. Similarly, in 1993, President Bill Clinton proposed a British thermal unit
(BTU) tax that was never adopted.

Vehicle Standards

Rather than directly regulate on the composition of fuels, or use the pricing
mechanism, there have been efforts to influence energy consumption and pollu-
tion by legislating on the design of vehicles. The details adopted vary and here we
highlight just some of the issues by looking at the European and North American
cases.
There have been somewhat different approaches adopted in different coun-
tries. Returning to Figure 8.1, the approach of the Europeans has been largely
on the environmental emissions themselves, by stipulating in agreements with
their own industry and with manufacturers in Japan and Korea the maximum
levels of pollution that new vehicles may emit. Table 8.3 shows the main stand-
ards set for gasoline cars. Other categories of vehicles have their own standards
that must be met.
In contrast to this, the CAFE standards, first introduced by the United
States Congress in 1975, are federal regulations that sought to improve fuel
economy in the wake of the 1973 Arab oil embargo (Table 8.4). In other words,
in terms of Figure 8.1 they impact on the industry. The regulations initially
applied to the sales-weighted average fuel economy, expressed in mpg, of a
manufacturer’s fleet of current model year passenger cars or light trucks with
a gross vehicle weight rating of 8,500 lbs. or less, which were manufactured for
sale in the United States. Light trucks not exceeding 8,500 lbs. gross vehicle
weight rating do not have to comply with CAFE standards. They constituted

BUTTON 9781786435668 PRINT.indd 307 27/04/2022 11:07

 308 TRANSPORT ECONOMICS, 4TH EDITION

Table 8.3 European new gasoline automobile emissions standards (grams per mile)

Date (first registration) CO THC VOC NOX HC+NOX P PN[#/km]

January 1997 2.2 – – – 0.5 – –

January 2001 2.3 0.2 – 0.15 – – –
January 2006 1 0.1 – 0.08 – – –
January 2011 1 0.1 0.068 0.06 – 0.005 –
January 2013 1 0.1 0.068 0.06 – 0.0045 –
September 2015 1 0.1 0.068 0.06 – 0.0045 6×1011
September 2018 1 0.1 0.068 0.06 – 0.0045 6×1011
September 2019 1 0.1 0.068 0.06 – 0.0045 6×1011
January 2021 1 0.1 0.068 0.06 – 0.0045 6×1011

Table 8.4 Canadian company average fuel consumption (CAFC) goals and the United
States’ corporate average fuel economy (CAFE) standard (liters/100 km)

Model year Passenger car Light-duty trucks

CAFC CAFE CAFC CAFE

1978 13.1 13.1 – –

1980 11.8 11.8 – –
1982 9.8 9.8 – 13.4
1984 8.7 8.7 – 11.8
1986 8.6 9.1 – 11.8
1988 8.6 9.1 – 11.7
1990 8.6 8.6 11.8 11.8
1992 8.6 8.6 11.6 11.8
1994 8.6 8.6 11.5 11.7
1996 8.6 8.6 11.4 11.7
1998 8.6 8.6 11.4 11.7
2000 8.6 8.6 11.4 11.7
2002 8.6 8.6 11.4 11.7
2004 8.6 8.6 11.4 11.7
2006 8.6 8.6 11.4 10.9

Source: Adapted from Perl and Dunn (2007).

some half a million vehicles in 1999 and by 2021 their number had reached
320 million. From early 2004, the average new car had to exceed 27.5 mpg and light
trucks exceed 20.7 mpg. Trucks under 8,500 lbs. had to average 22.5 mpg in 2008,
23.1 mpg in 2009, and 23.5 mpg in 2010. After this, new rules set varying targets
based on truck size ‘footprint’.
Whereas the United States’ regime is statutory, the Canadians have a vol-
untary scheme to foster fuel economy: the company average fuel consumption
(CAFC) agreement, which was established between government and auto manu-
facturers in 1978. Details of the joint goals set are seen in Table 8.5.
The US National Highway Traffic Safety Administration (NHTSA) regu-
lates CAFE standards and the US EPA measures vehicle fuel efficiency. Congress

BUTTON 9781786435668 PRINT.indd 308 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 309

Table 8.5 The economic costs of road injuries and morbidity (2015–30, $ 2010)

Economic loss Percentage Per capita

($ billions) of GDP loss

By World Bank region

    East Asia & Pacific 560 0.123 240
    Europe & Central Asia 345 0.082 374
    Latin America & Caribbean 115 0.116 184
    Middle East & North Africa 103 0.166 227
   North America 515 0.149 1,370
   South Asia 121 0.155 64
   Sub-Saharan Africa 38 0.120 33
By World Bank country income group
   Low income 11 0.120 14
   Lower-middle income 202 0.138 64
   Upper-middle income 621 0.144 237
   High income 963 0.106 779
   Global (166 countries) 1,797 0.120 231

Source: Chen et al. (2019).

specifies that CAFE standards must be set at the ‘maximum feasible level’ given
consideration for: technological feasibility; economic practicality; effect of other
standards on fuel economy; and the need of the nation to conserve energy. If
the average fuel economy of a manufacturer’s annual fleet of car and/or truck
production falls below the defined standard, the manufacturer pays a financial
penalty; thus, there is a very crude pricing mechanism involved. Fuel efficiency
is highly correlated to vehicle weight, but weight has been considered by many
safety experts to be correlated with safety, intertwining the issues of fuel economy,
road traffic safety, air pollution, global warming, and greenhouse gases. Hence,
historically, the EPA has encouraged consumers to buy more fuel-efficient vehi-
cles while the NHTSA has expressed concerns that this leads to smaller, less safe
vehicles. More recent studies tend to discount the importance of vehicle weight
on traffic safety, concentrating instead on the quality of the engineering design
of vehicles.
While there have been changes in the standards over time, Tables 8.4 and
8.5 both show that these have tended to be infrequent and that by, for example,
European standards, the American fleet is relatively fuel-inefficient. Part of the
problem seems to be difficulties in building political alliances strong enough to
carry though measures that tighten the prevailing standards.
A further problem with the CAFE standard approach is that it can lead
to an increase in highway externalities. While the CAFE standards, disregard-
ing any undesirable stimulating effects they have on the sales of light trucks
and SUVs (which are much less rigorously controlled), will increase the fuel
efficiency of vehicles, they also make them cheaper to drive per mile. This can
add to congestion and local environmental damage, including noise nuisance
(Kleit, 2004).

BUTTON 9781786435668 PRINT.indd 309 27/04/2022 11:07

 310 TRANSPORT ECONOMICS, 4TH EDITION

Speed Limits

Engines of all types perform differently at different speeds and each has an
optimal fuel performance speed. Given the operational cycle of any transport
activity, as a generalization more energy is expended at the beginning of a move-
ment, and in some cases at the end, than during cruise, although there is normally
an optimal cruising speed. It is possible, therefore, to influence the energy effi-
ciency of a transport system by regulating the speeds at which individual units
operate over it.
Privately supplied transport operations, such as shipping and airlines, have
financial incentives to conserve energy and, other things being equal, route ships
and planes accordingly and set fuel-efficient schedules. The public authorities,
cognizant of the wider impacts of transport, often off-set these energy goals to
attain other objectives. The most obvious cases are the take-off and landing pat-
terns at airports, which seldom are energy-efficient but take cognizance of noise
nuisance envelopes.
While speed limits are usually imposed for reasons of improving traffic flows
and for safety, there are examples of explicit, speed-based energy policies in trans-
port. As an emergency response to the 1973 oil crisis, the United States Congress
effectively imposed a national 55 mph maximum speed limit in 1974 under the
Emergency Highway Energy Conservation Act by requiring the limit as a condi-
tion of each state receiving highway funds. The limit was unpopular, especially
in western states that have long distances between cities or points of interest.
Subsequent analysis by (Forester et al. 1984) was somewhat unclear on the impli-
cations of the measure on energy consumption. Congress lifted all federal speed
limit controls in the November 28, 1995 National Highway Designation Act, fully
delegating speed limit authority to the states.

Fostering Alternative Technologies

Taxation, vehicle design standards, and other measures, in addition to market

forces, can, and have in many cases, affected the technology of transport and,
ipso facto, energy use and efficiency. The higher fuel prices after the Israeli wars
of the 1970s led to lighter vehicles using alternative materials for bodywork, and
more fuel-efficient engines. In addition, however, there have also been some other
explicit policies aimed at technology shifts with the aim of reducing the use of
oil-based fuels.
There has been a long history, for example, of policies aimed at develop-
ing viable electric cars that can effectively be powered from a variety of energy
sources, including hydro-generated electricity. Historically, in the early days,
electric-driven cars were as numerous as gasoline, steam, or diesel cars, the first
electric car being developed somewhere between 1832 and 1839, but seem to have
gone out of favor because of maintenance issues and the costs of mass produc-
tion. These are sometimes called, as in California, ‘zero emission vehicles’ (ZEVs),
although from the wider geographical perspective, given the primary source of

BUTTON 9781786435668 PRINT.indd 310 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 311

Exhibit   ‘Boris bikes’

In August 2007, London mayor Ken Livingstone announced a cycle hire scheme. This
became operational as Barclays Cycle Hire in July 2010 with 5,000 bicycles and 315 docking
stations distributed across the City area and parts of eight boroughs. Boris Johnson had
become mayor by then, hence the colloquialism ‘Boris bikes’. Initially, the system required
payment of registration and membership fees in exchange for an electronic access key, but
this was changed to allow casual cycle hires by non-members who had a credit card.
The project was expected to cost £140 million for planning and implementation over six
years, and was seen as the only Transport for London system to fully fund its annual cost
of operation, a goal estimated to take two to three years. The ‘on the road’ cost involved
was £28,000 per bicycle, with the docking stations costing around £200,000 each to install.
An expansion in March 2012 involved 2,300 more bikes and 4,800 docking stations. In
December 2013 an extra 2,000 bikes and 150 docking stations in west London were added.
In 2015, sponsorship of the scheme transferred from Barclays to Santander Cycles and by
February 2018 covered 40 square miles of London with over 11,000 bikes and 800 stations.
Ridership however, as seen in the figure, tended to have plateaued by about 2016 at around
10 million rides a year despite the bicycles benefitting from laws banning e-scooters on the
sidewalk or the road, although this began to change in 2020 with trials beginning for the
introduction of e-scooters.
Regular users register on a website for 24 hours or one year. They are given a key to
release a bike from a docking station. From December 2010 casual users could join for
hourly use. The first 30 minutes of each trip is free. Additional usage charges are £2 per 30
minutes or part thereof. Bicycles may be used any number of times within the access period,
each use charged according to its duration.

12

10

8
Million riddes

0
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
Annual ‘Boris bike’ bicycle rides

BUTTON 9781786435668 PRINT.indd 311 27/04/2022 11:07

 312 TRANSPORT ECONOMICS, 4TH EDITION

In the first three months, 95 percent of journeys did not exceed 30 minutes, earning access
but not usage fees. The scheme generated £323,545 in usage revenue in under 100 days,
with 72,700 of the initial 1.4 million journeys earning no revenue. In May 2012, Transport
for London estimated the scheme would cost tax-payers £225 million by 2015–16, almost
five times the maximum due from Barclays. Access fees were doubled in January 2013,
hoping to generate £4–6 million annually. From 2016–17 Transport for London put £3.6
million into the scheme, effectively hiring 10 million bike rides. This amounted to 16.9
percent of the scheme’s operating costs.
The scheme was criticized for allowing riders to have unlimited use by docking the bike
every 30 minutes resulting in a dependence on late fees and penalties to make up revenues.
The system also required the user to find docking stations close to the points of departure
and destination, lacking one of the key advantages of bicycles, dock-less bicycles, and
e-scooters. Coverage was also poor in south-east London, an area with a limited Tube
network, and in outer London, where the scheme is almost non-existent despite most
Londoners living there. Additionally, redistribution of bikes was hindered by the refusal of
Westminster and of Kensington and Chelsea councils to allow bikes to be repositioned
in their boroughs between 11.00 pm and 8.00 am, creating challenges in meeting morning
peaks.
See also: M. Ricci (2015) Bike sharing: a review of evidence on impacts and processes of
implementation and operation, Research in Transportation Business and Management, 15,
28–38.

energy, this is very seldom the case, and even if solar panels are used on vehicles,
there is still the pollution associated with the production of these panels. National
governments have regularly tried to foster the development of economically feasi-
ble electric car technologies by investing in R&D programs.
At a more local level, the California zero emission vehicle (ZEV) program,
initiated in 1990 and followed in some other states as partial zero emission vehicle
(PZEV) programs, was designed to catalyze the commercialization of advanced-
technology vehicles that would not have any tailpipe or evaporative emissions
(California Air Resources Board, 2005). It initially required that 2 percent of
new vehicles produced for sale in 1998 and 10 percent of new vehicles produced
for sale in 2003 would be ZEVs. After auto-makers argued that they could not
meet the 1998 deadline, full implementation of the program was delayed until
2003 with interim measures to encourage the use of more PZEVs. In 2002, auto-
makers sued the state over the program and were granted a preliminary injunction
barring its implementation pending a final court ruling. During the ensuing legal
debate, the state decided to go ahead and make revisions to the rule to sidestep
the legal challenge, with the aim of restoring the ZEV program by 2005. Overall,
these types of policies have not been conspicuously successful in bringing about
sea changes in transport technology. For example, of the 4,000 to 5,000 electric
cars built for California’s ZEV mandate in the late 1990s, only about 1,000 remain
on the road.

BUTTON 9781786435668 PRINT.indd 312 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 313

The European Union, through Joint Technology Initiatives in the 7th

Research Framework Programme running from 2007 to 2013, provided increas-
ing levels of funding for research into fuel cells and hydrogen with the intention
of reducing the time needed to market such technologies by between two and five
years.
While fully electric or hydrogen-propelled vehicles have proved elusive to
develop on a commercial scale, the hybrid vehicle such as the Honda Insight and
the Toyota Prius that combines electric propulsion with, generally, a gasoline
engine has proved more successful. It offers, at prevailing prices, fuel efficiency,
although at a higher capital cost, and in many cases is economically justified in the
market-place. Policy has been instrumental by financing part of the R&D costs
of the underlying technology, but in many cases local policies have provided an
added inducement for its uptake. In the United States many local jurisdictions, for
example, allow hybrids to use high-occupancy vehicle (HOV) lanes on highways
even if they do not meet passenger requirements.
There are also initiatives to foster the use of telecommunications as an
alternative to trips that are primarily for information exchange. This can apply
to such things as teleworking (rather than commuting to work in an office), tele-
shopping, telebanking, and distance learning (Salomon and Mokhtarian, 2007).
The ­Covid-19 outbreak in 2020 and the quarantine measures introduced in many
countries led in the short term to far more activities becoming home-based. In
terms of such activities as teleworking, there had already been a sizeable uptake
of remote working (Figure 8.8).
While the debate over whether the advances in telecommunications in the
long term have added to travel because of the complementary nature of the
home-based activities, or reduced it because of its substitutability features, is

60

50

40
Million riddes

30

20

10

0
Po ey

er ly
y
Sl in

Ire a

itz ce

et US

Fi ds
Be nd

nd

D and

k
K
ga

i
an

ar
en

U
Ita

iu
Sw an

n
rk

la

la
rtu

Sp

m
rla
m

nl
lg
ov

er
Tu

Fr

en
he
G

Figure 8.8 Shares of people teleworking (2015–16)

BUTTON 9781786435668 PRINT.indd 313 27/04/2022 11:07

 314 TRANSPORT ECONOMICS, 4TH EDITION

still the subject of much debate, policy-makers have put forward initiatives to
increase its use as a transport energy policy initiative. Some of this has been in
the form of information (for example, the US Office of Personnel Management
and the General Services Administration have established this joint website on
Telework to provide access to guidance issued by both agencies) and facilitation
(for example, under United States law, federal executive agencies must establish
policies under which eligible employees may participate in telecommuting to
the maximum extent possible). Again, in the United States in 1996, the Clean
Air Act, amongst other things, required companies with over 100 employees to
encourage telecommuting. The European Union also reached a framework agree-
ment to encourage more teleworking and to put in place laws that would help
facilitate it across the member countries.
The difficulty is assessing whether telecommuting and similar activities,
while unquestionably enhancing social welfare by giving firms and individuals
more choices, result in less travel. There are inevitable ‘buy-back effects’ as time
formerly used, for example, in physically commuting can be used for other forms
of trip-making. Additionally, telecommunications may simply add to the conven-
tional travel behavior of people and the way firms use their employees’ time. This
type of situation is illustrated in Figure 8.9.
There are assumed to be physical limits to the amount of travel an individ-
ual can do and the number of physical interactions possible. Improved transport
and communications have undoubtedly improved the situation, but logic would
suggest that there is some asymptote. While one may pull down this asymptote,
or modify the growth path in physical inter-personal communications by foster-
ing telecommunications after time T, there is also the possibility that the new
technology will lead to a higher number of inter-personal communications,
pushing up the overall communications asymptote without reducing the growth

Time
Physical interactions asymptote

T Telecommunications gap

0 Inter-personal contacts

Figure 8.9 The ‘telecommuncations gap’

BUTTON 9781786435668 PRINT.indd 314 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 315

in personal movements, the telecommunications gap reflecting this. If so, then

the impact on the environmental damage associated with transport is left unaf-
fected. Indeed, if there is a degree of substitution within the physical travel
aggregate, then this may be more damaging to the environment if it involves trips
through more sensitive areas of cities or at times of day when citizens are more
sensitive to noise.

8.9 Safety and Accidents

Safety is a major concern in modern transport. Any form of movements of indi-

viduals and goods involves a degree of danger, but that has not stopped the devel-
opment of modern transport. Indeed, while the opening of the Liverpool and
Manchester Railway in the United Kingdom, the world’s first inter-city railway,
in September 1830, saw William Huskisson, the popular Member of Parliament
for Liverpool, run over by the engine, the Rocket, and killed, this heralded the
beginning of a period of ‘railway mania’.
Today, traffic injuries are among the ten leading causes of death in the
world, and they are the leading cause of death among young adults aged 15–29
years. Traffic accidents also produce 20–50 million non-fatal injuries with many
of those involved incurring a disability. Table 8.5 extends this to the global situ-
ations. Table 8.5 offers estimates of the global economic costs of road accidents
and sets them in the context of the world’s GDP. The latter is important because
the costs of accidents involve evaluation of the economic value of victims.
Higher-income countries, therefore, have higher absolute costs, although differ-
entials are not so great when set in the context of a country’s overall economic
condition.
There are numerous causes of accidents, but many are the result of irre-
sponsible behavior by vehicle operators. On the roads, young and elderly drivers
are more prone to accidents, but there is a strong correlation between drinking
and driving. There are also some people who are quite literally, for no apparent
reason, ‘accident prone’. Although there are many factors contributing to the fre-
quency and severity of road crashes, drunk driving is among the principal reasons
for crashes in many countries; for example, 40 percent of fatal crashes in the
United States in 2003 were alcohol-related, a percentage that has remained rela-
tively stable over the past few years. Consistent with this, the National Highway
Traffic Safety Administration, McCarthy and Tay’s (2005) study of 2000 United
States crash data, estimated that alcohol-related crashes imposed a $114.3 billion
cost on society.
Safety is partly an externality and partly it is not. One policy approach is
to internalize more of the costs of safety. This is done by such measures as legal
responsibility for the costs of accidents in the form of compensation, a measure
that is often accompanied by compulsory insurance. Such actions can, however,
in some contexts because of moral hazard, prove counterproductive. Moral
hazard is the prospect that a party insulated from risk (by, in this case, insurance)

BUTTON 9781786435668 PRINT.indd 315 27/04/2022 11:07

 316 TRANSPORT ECONOMICS, 4TH EDITION

may behave differently from the way it would behave if it were fully exposed to
the risk. It can arise because an individual does not bear the full consequences of
their actions, and therefore has a tendency to act less carefully than would oth-
erwise be the case, leaving another party to bear some responsibility for the con-
sequences of those actions. With insurance, it is the other premium payers who
bear the burden. For example, an individual with insurance against automobile
theft may be less vigilant about locking a car, because the negative consequences
of automobile theft are partially borne by insurance company.
One way of reducing the moral hazard problem is to change the insurance
market, and one modification, the pay-as-you-drive (PAYD) insurance, that
embraces actual miles driven for premiums rather than a flat-period sum, has
been suggested. The aim would be to limit trips, rather than encourage them as
the more traditional approaches do. Aaron Edlin (2003) has estimated that this
would reduce the externalities associated with driving, including accidents.
If such internalization is not possible, or thought to be ineffective or imprac-
tical, other policies are commonly used to handle safety concerns. Focusing on
road transport to keep the material manageable, although the experiences extend
in general across all modes, policy initiatives often revolve around keeping some
groups off the road entirely – for example, a minimum driving age – and deterrent
measures such as driving bans if caught with alcohol in the system above a ‘safe’
level, fines for speeding or dangerous maneuvers, and imprisonment for extreme
offenses. Training and testing of drivers are also common practice, and educa-
tional initiatives to discourage dangerous driving are common, although it is not
clear that they always meet their objectives (Tay, 2004).
The reunification of East and West Germany in 1990 provided a natural
economic experiment when one important change for East German motorists
was a relaxation of the legal blood alcohol concentration (BAC) limit (Vollrath
et al., 2005). There was no change in the legal limit for West German motorists.
This offered the opportunity to analyse the immediate, short-term, and longer-
term effects of raising the BAC limit in East Germany between 1992 and 1994. A
major finding was that the relaxed concentration limits led to an increase in blood
alcohol levels among East German drivers but generally did not increase the fre-
quency of drinking and driving. The exception was younger drivers who not only
increased their alcohol consumption after the relaxation in the legal limit but also
increased their frequency of drinking and driving.
Also focusing upon drinking and driving policy, Mathijssen (2005) conducted
a retrospective analysis of anti-drunk driving campaigns that the Netherlands
implemented between 1970 and 2000. The study found significant deterrent
effects following the introduction of a statutory alcohol limit, random breath
testing and evidential breath testing, and changes in the enforcement level; but
mixed effects for publicity and educational campaigns; and little effect for changes
in penalties and driver improvement programs.
Speed is generally highly correlated with accidents, and speed limits are
common for that reason. In April 1987, the United States federal government
enacted the Federal Highway Bill that, among its other provisions, permitted

BUTTON 9781786435668 PRINT.indd 316 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 317

states to increase speed limits on rural sections of their Interstate highways. This
provided the basis for several natural experiments. Forty states responded to
the enabling legislation by raising their speed limits, fueling an ongoing debate
regarding the highway safety effects of higher speed limits. The results of annual
cross-section–time series covering 1981 to 1989 for California, which raised
speed limits in May 1987, found the higher speed limits to have no system-wide
effects on fatal, injury-related, or property damage accidents (McCarthy, 1994).
Redistribution effects, however, were identified in that counties with 65 mph
highways experienced a significant increase in each accident category. Theodore
Keeler (1994) also found, when looking at data for 1970 and 1980, that lower rural
speed limits did little to reduce fatal automobile accidents.
Vehicle design and ‘fittings’ may also affect safety. These types of measures
include such things as seat belts, air bags, and crash helmets on motorcycles, as
much as the quality of brakes and suspension, the types of wind-shield glass and
headlamp lighting, and so on. There is an economic problem forcing regulatory
measures to ensure that the safest engineering technology, or at least a minimum
stand, is adopted. People react to the new technical requirements. Lave and
Weber (1970), for instance, suggested that mandated safety devices (seat belts,
better bumpers, collapsible steering wheels) might lead to faster driving that
could off-set the safety gains. More generally, Samuel Peltzman (1975) found
that while technological studies in the mid-1970s implied that annual highway
deaths would be 20 percent greater without legally mandated installation of
various safety devices on automobiles, the effects of non-regulatory demand
for safety and drivers’ behavioral responses to the devices virtually off-set these.
Sobel and Nesbit’s (2007) study of changes in NASCAR safety rules offers
support for this.
Saas and Zimmerman (2000), for example, studied the impact of United
States state laws mandating helmet use by motorcyclists over a 22-year period and
found helmet laws to be associated with an average 29 to 33 percent decrease in
per capita motorcyclist fatalities. However, since voluntary helmet wearing rates
are higher in harsher climates, the efficacy of helmet laws varied directly with the
warmth of a state’s climate. Repeal of helmet laws in the 1970s and subsequent
re-adoption in the late 1980s and early 1990s had had roughly symmetrical effects
on fatalities. It was also found that alcohol consumption and the number of police
available to enforce traffic laws significantly affected motorcyclist fatalities.
In other words, safety regulations do not necessarily in themselves decrease
accidents and deaths because of the off-setting effects on driver behavior and
enforcement policies.
The effects of piecemeal approaches and of policy conflicts involving pack-
ages of measures were studied by Crandall et al. (1986). They were concerned
about the conflicts which arose in the United States between the setting of safety
standards (by Highway Traffic Safety Administration), emissions standards (by
Congress and administered by the EPA), and fuel economy measures (set by
Congress and administered by the EPA). While static conflicts emerged as techni-
cally unavoidable (for example, the safety regulations initiated in the United States

BUTTON 9781786435668 PRINT.indd 317 27/04/2022 11:07

 318 TRANSPORT ECONOMICS, 4TH EDITION

between 1968 and 1982 pushed average car weights up by 136 lbs. and increased
fuel consumption by 3.5 percent, which conflicted with measures aimed at fuel
efficiency and emissions controls), the dynamic effects of uncoordinated policies,
for instance, slowing down replacement of older, less socially desirable cars was a
serious administrative failure.
An important point is that the level of safety expected from transport
systems does tend to vary between countries. In part this is due to income differ-
entials. As seen earlier in the book, the value society puts on reducing the chance
of a fatal accident depends to some extent on the income of the people involved.
There is ample evidence that when it comes to transport safety, poorer coun-
tries, in part because of their limited resource base, but also because of the high
shadow prices involved, spend less on meeting what in other countries are taken
as minimum safety standards (Jadaan et al., 2018).

References

Baum, H.J. (1973) Free public transport, Journal of Transport Economics and Policy, 7,
3–19.
Baumol, W.J. and Oates, W.E. (1988) The Theory of Environmental Policy, Cambridge
University Press.
Becken, S. and Mackey, B. (2017) What role for offsetting aviation greenhouse gas emis-
sions in a deep-cut carbon world? Journal of Air Transport Management, 63, 71–83.
Button, K.J. (2001) Are current air transport policies consistent with a sustainable environ-
ment? in E. Verhoef and E. Feitelson (eds), Transport and Environment: In Search of
Sustainable Solutions, Edward Elgar Publishing.
California Air Resources Board (2005) California’s air quality history key events, Report,
CARB.
Chen, S., Kuhn, N., Prettner, K., and Bloom, D.E. (2019) The global macroeconomic
burden of road injuries: estimates and projections for 166 countries, Lancet Planetary-
health, September 3, e390–e398.
Coase, R.H. (1960) The problem of social cost, Journal of Law and Economics, 3, 1–44.
Crandall, R.W., Gruenspecht, A.K., Keeler, T.E., and Lave, L.B. (1986) Regulating the
Automobile, Brookings Institution.
De Witte, A., Macharis, C., Lannoy, P., Polain, C., Steenberghen, T., and De Walle, S.V.
(2006) The impact of ‘free’ public transport: the case of Brussels, Transportation
Research Part A, 40, 671–89.
Edlin, A.S. (2003) Per-mile premiums for auto insurance, in R. Arnott, B. Greenwald, R.
Kanbur, and B. Nalebuff (eds), Economics for an Imperfect World: Essays in Honor of
Joseph E. Stiglitz, MIT Press.
Forester, T.H., McNown, R.F., and Singell, L.D. (1984) A cost–benefit analysis of the 55
MPH speed limit, Southern Economic Journal, 50, 631–41.
Goodwin, P.B. (1992) A review of new demand elasticities with special reference to short
and long run effects of price changes, Journal of Transport Economics and Policy, 26,
155–70.
Goodwin, P.B., Dargay, J., and Hanly, M. (2004) Elasticities of road traffic and fuel con-
sumption with respect to price and income: a review, Transport Reviews, 24, 275–92.
Jadaan, K., Al-Braizat, E., Al-Rafayah, S., Gammoh, H., and Abukahlil, Y. (2018) Traffic
safety in developed and developing countries: a comparative analysis, Journal of Traffic
and Logistics Engineering, 6, 1–5.

BUTTON 9781786435668 PRINT.indd 318 27/04/2022 11:07

 CONTAINING THE ENVIRONMENTAL COSTS OF TRANSPORT ­ 319

Keeler, T. (1994) Highway safety, economic behavior, and driving environment, American
Economic Review, 84, 684–93.
Kemp, M.A. (1973) Some evidence on transit elasticities, Transportation, 2, 25–52.
Kleit, A.N. (2004) Impacts of long-range increases in the fuel economy (CAFE) standard,
Economic Inquiry, 42, 279–94.
Larsson, J., Elofsson, A., Sterner, T., and Åkerman, J. (2019) International and national
climate policies for aviation: a review, Climate Policy, 19(6), 787–99.
Lave, L.B. and Weber, W.E. (1970) A benefit–cost analysis of auto safety features, Applied
Economics, 2, 265–76.
Maler, K.G. (1974) Environmental policies and the role of the economist in influenc-
ing public policy, in J.G. Rothenberg and I.G. Heggie (eds), Transport and the Urban
Environment, Macmillan.
Mathijssen, M.P.M. (2005) Drink driving policy and road safety in the Netherlands: a
retrospective analysis, Transportation Research Part E, 41, 395–408.
McCarthy, P. (1994) An empirical analysis of the direct and indirect effects of relaxed inter-
state speed limits on highway safety, Journal of Urban Economics, 36, 353–64.
McCarthy, P. and Tay, R. (2005) Road safety, alcohol and public policy, Transportation
Research Part E, 41, 373–76.
OECD (Organisation for Economic Co-operation and Development) (1975) The Polluter
Pays Principle, OECD.
OECD (Organisation for Economic Co-operation and Development) (1992) Market and
Government Failures in Environmental Policy: The Case of Transport, OECD.
Parry, I.W.H., Walls, M., and Harrington, W. (2007) Automobile externalities and policies,
Journal of Economic Literature, 45, 373–99.
Peltzman, S. (1975) The effects of automobile safety regulation, Journal of Political
Economy, 83, 677–725.
Perl, A. and Dunn, J.A. (2007) Reframing automobile fuel economy policy in North
America: the politics of punctuating a policy equilibrium, Transport Reviews, 27, 1–35.
Pigou, A. (1920) The Economics of Welfare, Macmillan.
Salomon, I. and Mokhtarian, P.L. (2007) Can telecommunications help solve transporta-
tion problems? A decade later: are prospects better, in D.A. Hensher and K.J. Button
(eds), Handbook of Transport Modelling, 2nd edn, Pergamon.
Sass, T.R. and Zimmerman, P.R. (2000) Motorcycle helmet laws and motorcyclist fatali-
ties, Journal of Regulatory Economics, 18, 195–215.
Scheelhaase, J., Maertens, S., Grimme, W., and Jung, M. (2018) EU ETS versus CORSIA:
a critical assessment of two approaches to limit air transport’s CO2 emissions by mar-
ket-based measures, Journal of Air Transport Management, 67, 55–62.
Schwing, R.C., Southworth, B.W., Von Buseck, C.R., and Jackson, C. (1980) Benefit–cost
analysis of automobile emission reductions, Journal of Environmental Economics and
Management, 7, 44–64.
Sharp, C.H. (1979) The environmental cost of transport and the public interest, Journal of
Transport Economics and Policy, 13, 88–191.
Sobel, R. and Nesbit, T.M. (2007) Automobile safety regulation and the incentive to drive
recklessly: evidence from NASCAR, Southern Economic Journal, 74, 71–84.
Tay, R. (2004) The relationship between public education and enforcement campaigns and
their effectiveness in reducing speed related serious crashes, International Journal of
Transport Economics, 31, 251–5.
Vollrath, M., Krüger, H.-P., and Löbmann, R. (2005) Driving under the influence of
alcohol in Germany and the effect of relaxing the BAC law, Transportation Research
Part E, 41, 377–93.

BUTTON 9781786435668 PRINT.indd 319 27/04/2022 11:07

 9 Optimizing Traffic Congestion

9.1 Economics and Optimal Traffic

As we saw in Chapter 6, excessive congestion leads to the inefficient use of

transport infrastructure. Throughout the world, there has been a pot pourri of
initiatives over the years to optimize congestion, not only on city streets but also
on inter-urban roads and at seaports and airports. These initiatives have not
proved conspicuously successful as we can observe in most cities and at many
ports. Many of the policies that have been pursued, however, have been devoid
of serious economic content. The aim of this chapter is to examine the types of
approaches that economists have advocated for reducing excessive levels of traffic
congestion – it is not about eliminating it – and what the outcomes have been
when they have been adopted.
It is not only in the context of pollution that economists have advocated
externality pricing. One idea for optimizing the level of congestion is to use
the price mechanism to make travelers more fully aware of the impedance they
impose upon one another. The suggestion is that motorists should pay for the
additional congestion they create when entering a congested road or that air-
craft should pay a premium to land at busy times of the day. In the case of road
traffic, ideally, as with pollution charges, those causing congestion should pay the
other road users adversely affected, but practically this is quite clearly impossible.
Hence, the idea is that the relevant road authority or agency should be responsible
for collecting the charges.
We spend some time looking at the theoretical and practical issues involved
in doing this, and assess some of the congestion charging schemes that have been
introduced into cities like London, Gothenburg, and Stockholm, and on inter-
city highways.
Other measures, of a less direct and second-best kind, such as parking
charges, public transit subsidies, the deployment of intelligent transport systems
(ITSs), and investment in additional infrastructure, are also considered. In techni-
cal terms, while congestion charging or road pricing seek to equate the demand
for transport use with the full costs of use, including the costs on other transport
users (environmental costs are a separate matter), alternatives such as capacity
expansion and the wider use of ITSs seek to push the costs of transport use down,
whereas measures such as transit subsidies seek to decrease the demand for the
congestion-causing mode through substitution effects.
While much of our attention is on the congestion seen on roads, congestion
is a problem found in many other transport activities, as discussed in Chapter 6.

BUTTON 9781786435668 PRINT.indd 320 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 321

While the general economic approach to developing and assessing remedial poli-
cies is common across the various forms of transport, there are various nuances,
and we shall consider some of these later in the chapter.
The aim of this chapter is not to provide an encyclopedic coverage of all the
literature that has been generated on road pricing and other forms of congestion
charging, nor to catalogue all the efforts that have been made – largely failures
with a few successes – to adopt road pricing. It is not the intention to go into the
finer points of obtuse microeconomic theory. Rather, it is to consider some of the
challenges that confront the adoption of efficient fiscal policy instruments within
the transportation system using road pricing as an illustration. In doing this,
examples and case studies will be used so that readers unfamiliar with the idea of
adopting user charges as a means of improving traffic management will not find
it difficult to understand the basic concepts involved.

9.2 ‘Road Pricing’

Charging for road use has traditionally taken one of two forms. There have long
been tolled roads of the type initiated in Britain in the seventeenth century and in
the United States from 1795, when the Philadelphia and Lancaster Turnpike was
completed, where users pay a fee for use. These tolls – and the same is true where
they are used today – are intended to recover the cost of constructing and physi-
cally maintaining the road. Such tolls are not designed to allocate the road space
or to optimize congestion. When tolls vary it is normally related to the physical
damage done by a vehicle to the pavement and not to the impedance that such a
vehicle may impose on other road users.
More widely, road users are charged even less directly for their use of infra-
structure through a variety of taxes that may (as in the case of the United States
federal gas tax) be hypothecated to be spent on the road network (often through a
road fund mechanism) or flow into the general coffers of the Treasury to be spent
as the government wishes. In general, these taxation systems have little to do with
making good use of road assets.
The role of an economic price, however, is threefold: to allocate what is
available, to indicate where that capacity needs to be changed, and to provide the
resources for financing that change. The acronym AIR sums this up. Traditional
tolls may serve the last of these purposes by recovering investment costs, but
they seldom meet the other two. Table 9.1 provides some very general guidance
as to the tasks that various forms of road charging are set to perform. From the
perspective of optimizing congestion, weight–distance truck tolls and general dis-
tance tolls for example provide limited support, although they are relatively easy
to enforce and meet many budgetary demands.
The adverse effects of inappropriate charges for the use of one asset, in this
case a road, are felt in complementary and competitive sectors. They occur when
authorities seek to contain the primary problem with a second-best approach
involving either restricting the use of complementary services (for example,

BUTTON 9781786435668 PRINT.indd 321 27/04/2022 11:07

BUTTON 9781786435668 PRINT.indd 322
Table 9.1 Technical tasks for various forms of road charging

Road charging Entering/ Presence Position Distance Time/ Vehicle Charges Data Data Payment Enforce-
scheme exiting in area on road traveled congestion class/ owed communi- storage billing ment
facilities network level weight cation

Facility d d​ s d d d d d
congestion tolls
Cordon d d s d d d d d
congestion tolls
Weight–distance s s s d s d d d d d
truck tolls
General d s d s s d d d d d
distance tolls

Notes: d Required for the task; s Optimal but not required for the task.

Source: Adapted from Sorensen and Taylor (2006).

27/04/2022 11:07
 OPTIMIZING TRAFFIC CONGESTION ­ 323

controls over parking) or stimulating the use of alternatives (such as public

transit) – see Button (2006b). This seldom works.
On the expenditure side, road investments are made for a variety of reasons,
often with quasi-economic justifications added as a veneer. Cost–benefit analysis,
which is discussed in detail in the next chapter, is widely used in one of its variants
to provide a social assessment, but the technique is far from perfect and subject
to political manipulation.
More powerful in the age of the car has been the engineering approach of
providing capacity to meet demand. While there may be some justification for
this in some situations, the fact that there is no direct economic price per trip paid
by road users for their activities means that ‘demand’ in this context is nothing
like the notion of effective demand used by economists. The outcome is likely to
be excess capacity and a geographical maldistribution of roads. The reluctance
of citizens to accept the environmental effects of more large-scale investments
in roads, coupled with the increasing costs of construction as successive projects
move down the marginal-returns-on-investment curves, have to some extent
stymied this build-to-meet-demand philosophy.
The distortions that result from inappropriate charges for the use of one
asset, in this case a road, have also been felt in complementary and competitive
sectors as the authorities have sought to initiate second-best policies to contain
congestion. Attempts to discourage car trips terminating in a congested area by
means of parking restrictions and fees leads to higher levels of through-traffic
as well as additional congestion as terminating drivers seek the limited parking
capacity that is available (Shoup, 2005). A significant amount of car movements
in many cities involves drivers looking for somewhere to park. Subsidies to public
transport reduce the incentive for it to be provided efficiently and at the lowest
cost – the X-inefficiency problem.
To reduce urban road traffic congestion, national and local authorities have
gradually been turning to policies with a degree of economic rationale under-
pinning them, rather than simply trying to build their way out of problems or
providing ever-increasing amounts of subsidies to public transportation. There
have been moves to use road pricing as a tool for rationing scarce road space to
those who gain most from its use. This focus on roads, and notably on the pricing
of automobiles, is not simply a reflection of the large number of people who
commute by car, it is also linked to the size and geography of cities. Table 9.2
provides some United States data on the share of automobiles in commuting.
While New York has a relatively small share of auto-commuters, it has some of
the worst because of its compact geography and sheer size. But all the cities listed
suffer from acute road traffic congestion issues.
The importance of appropriate pricing to make the best use of transporta-
tion infrastructure is certainly not new: the principles can be found in the work
of French engineering economists of the 1840s. Unfortunately, for a variety of
reasons, both practical and due to a lack of understanding by many of basic eco-
nomic principles, road space is seldom priced in a way that optimizes congestion
and, as a result, social welfare is not maximized. The modern principles of road

BUTTON 9781786435668 PRINT.indd 323 27/04/2022 11:07

 324 TRANSPORT ECONOMICS, 4TH EDITION

Table 9.2 Mode split (percentage) for travel to work in United States cities with more
than 30,000 commuters (2018)

Single driver Carpool Transit Bike Walk Other Home

New York 22 4 56 1 10 2 5
Los Angeles 70 9 9 1 3 2 6
Chicago 49 8 28 2 6 2 5
Houston 78 10 4 0 1 3 4
Phoenix 74 13 3 1 1 2 6
San Diego 74 9 4 1 4 1 7
San Antonio 78 12 2 0 2 3 3
Philadelphia 50 7 26 2 9 2 4
Dallas 77 11 4 0.20 2 1 5
Austin 75 8 3 1 3 2 8
San Jose 76 12 4 1 2 1 4
San Francisco 30 9 34 4 13 4 6
Columbus 78 10 3 1 3 1 4
Charlotte 75 10 3 0 2 2 8
Seattle 44 7 23 4 12 2 8

pricing go back over 80 years to the work of Arthur Pigou, and were expanded
upon by the Nobel Prize-winning economists James Buchanan and William
Vickrey, and by the former British Prime Minister Margaret Thatcher’s main
economic advisor Alan Walters (Button, 2020).
The idea is simple: since roads are not privately owned and access to them is
not determined by the market, there is a need, if roads are to be used efficiently,
for the authorities to set a user charge that, for any given capacity, ensures that
socially optimal flows are attained – the road price. There is often confusion
about what this means. Road pricing is solely concerned with making the best use
of roads from the narrow perspective of users and is not concerned with third-
party effects such as pollution that should be treated separately. Congestion
may or may not be related to environmental damage; 20,000 solar-powered cars
per hour on a road may congest it but cause minimal environmental damage,
whereas 500 old, badly maintained diesel vehicles may cause no congestion but
a lot of pollution.
Second, as another Nobel Prize-winning economist, Ronald Coase (1960),
demonstrated, policies such as road pricing are not market solutions but the most
effective way of achieving an externally determined target traffic flow. This flow
is independently determined and is not the result of any market-based process.
In the congestion case, where economic pricing has been adopted, this pragmatic
objective is often there to constrain traffic to the legal speed limit.
The standard economic diagram illustrating the basic principles of road
pricing is that originally found in the works of Walters and Vickrey, and although
this is now seen as a gross simplification, it serves well to illustrate the essential
aims of congestion charging. In Figure 9.1 we take the basic case of a straight,
single-lane road with a single entry and single egress point, and with homogeneous

BUTTON 9781786435668 PRINT.indd 324 27/04/2022 11:07

 325
OPTIMIZING TRAFFIC CONGESTION ­

traffic entering the road at regular intervals, although the intensity of entry can
change. The average cost curve shows the generalized costs (a composite of money
and time costs) of using the road for existing traffic, and is the cost observed by
a potential additional user. The demand curve represents the utility that potential
road users enjoy by joining the traffic flow at different levels of generalized cost.
Individuals will join the traffic flow, provided they feel that the benefits to be
enjoyed exceed the costs. In the diagram, this leads to a traffic flow of F1.
The problem, however, is that by undertaking the very act of joining the
traffic stream a vehicle slows all succeeding vehicles by some small amount unless
there is excess capacity on the road. This may only be a few seconds for each
vehicle, but when the flow is large this mounts up. This combination of the cost
borne by the vehicle driver and the cost imposed on other vehicles is depicted as
the marginal cost curve in the diagram. If motorists take these combined costs
into account then the flow would fall to F2. The idea of the road price is to make
them cognizant of the congestion cost element by imposing a charge of C2 – C1
on each road user, this being the additional congestion costs associated with
the marginal user at the optimal traffic flow. The result of this is a welfare gain
amounting to HJI – traffic flow F1F2 only generated benefits to road users equal
to the areas F1F2JI, but at a cost of F1F2JH.
The optimal road price is relatively simple to calculate, and may seem to
require little information. Taking the average cost of making a trip to the vehicle
user as:

AC = a + (b/v) (9.1)

where a is the cost per vehicle in $ per mile, b is the value of travel time in
$ per hour, and v is traffic volume, then the total (TC) cost of a traffic flow of

Generalized
costs Marginal cost
H

Average cost
J
C2

I
C1
K

Demand

0 F2 F1 Traffic flow
Figure 9.1 Simplified diagram of the effects of road pricing

BUTTON 9781786435668 PRINT.indd 325 27/04/2022 11:07

 326 TRANSPORT ECONOMICS, 4TH EDITION

q vehicles is equal to ACq. An additional vehicle joining the flow increases the
overall cost to road users by:

dTC dAC
MSC =
= AC + (9.2)
dq dq

The congestion externality is given by q(dAC/dq).

It takes little imagination to realize that over a complex network with numer-
ous junctions, various road types, different vehicles mixes, and so on, the case
in the diagram and equations is simple in the extreme and that calculating the
optimal charge would be challenging. Indeed, this is sometimes used as a reason
for not charging. The pricing of a computer, a car, or any other sophisticated
product or service is far from easy but it is done, albeit far from ideally, and
companies like Microsoft and Apple seem to be held in much higher esteem than
many local politicians and traffic engineers. Private sector companies in Western-
style economies make an educated guess at the appropriate direct price for their
products and then adjust them in the light of experience; if there is a large
demand for their product, then they raise prices and use the revenue to invest in
additional capacity that will ultimately reduce costs and pries. They do not charge
a low price and then let queues allocate the output.
Figure 9.1, despite its simplicity, also provides a way of looking at alterna-
tives to road pricing as congestion containment policies. Expanding the capacity
of a road by adding physical capacity or initiating better traffic management
through intelligent transportation initiatives will shift the cost curves down and
out and thus reduce the divergence between them at the point where demand
equates with the average cost. Such actions are clearly not costless and still leave
a divergence between the new actual traffic flow and the optimal flow given the
added capacity. Effectively traffic increases to fill the additional road capacity
available. Subsidizing public transportation will pull the demand curve for road
use to the right, but again will still leave, albeit smaller, a traffic flow in excess of
the optimum. Limiting or charging for parking will have similar impacts. We look
at these effects in more detail in Section 9.9.

9.3 Applications of Urban Road Pricing

What is so surprising is that the principles of using pricing to allocate scarce

resources, indicate where more are needed, and provide revenues to fund
­expansions – very simple economic ideas that are central in other parts of market
economies – have taken so long to be applied to roads. Basically, if there is a
shortage of something, prices rise to allocate it to those who benefit most from
its use. The higher prices indicate that more capacity is needed, and generate the
revenue to finance it.
Although it seems unlikely that many of the policy-makers who have moved
towards road pricing as a means of reducing congestion are aware of the finer

BUTTON 9781786435668 PRINT.indd 326 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­327

Table 9.3 Characteristics of eight major road pricing schemes

City Electronic system Entry charge for Toll ring Average daily Annual revenue
starting date a small vehicle area (km2) crossings (millions)

Trondheim 1991 $2.40 50.0 74,900 $25.00

Oslo 1991 $2.40 64.0 248,900 $196.00
Bergen 2004 $2.40 18.0 84,900 $36.00
Stockholm 2006 $1.33 to $2.66 29.5 550,000 n.a.
Singapore 1998 $0.33 to $2.00 7.0 235,000 $80.00
Rome 2001 $3.75 4.6 75,000 $12.30
London 2003 $15.0 22.0 110,000 $320.00
Santiago 2004 $6.42 n.a. 250,000 n.a.

Note: n.a. = not available.

points of economic theory, a combination of desperation at the failure of traffic

engineers to provide solutions to what is an economic issue, combined with a
tightening of public funding, concerns about the environmental implications of
continued massive infrastructure expansion, and the emergence of new technolo-
gies for fee collection, have created settings conducive to congestion pricing. We
review some of these here, and Table 9.3 offers a few more details; Lehe (2019)
goes into more depth.
In 1975 Singapore implemented a cordon-based variable pricing scheme
with the aim of reducing congestion in the city’s central business district (CBD).
In 1999, the scheme was extended and now charges are applied to users on certain
expressways and outer ring roads as well. Charges take effect on weekdays, from
7.30 am to 7 pm in the business districts, and from 7.30 am to 9.30 am on outer
roads. Automated electronic road pricing was implemented in 1998.
Cordon tolling has been a very widely used strategy, with some initia-
tives that were first intended to finance additional engineering works evolving
toward congestion charges. The deployment of cordon tolling, for example, was
popularized by its use in a number of Norwegian cities primarily for revenue-
generating purposes. In 1986, Bergen established the first urban toll ring as a
supplementary funding mechanism for new roads, public transport, parking
space, pedestrianization, and an environmentally improved city center. The
cordon-based scheme only requires payment on entering the city center. There
are six toll points around the city, with fees collected between 6 am and 10 pm
on weekdays. Approximately 13 percent of the revenue is used to cover operating
costs. Enforcement is dependent on digital video control. Because of the level
and structure of the charges, there has been little impact on traffic levels and
the system is not really a demand management tool in its current form. The toll
ring was expected to cease operating in 2011 but was retained and a second ring
added.
A toll ring scheme was initiated in Oslo in 1990 with the objective of financ-
ing additional roads capacity. Charges are in effect seven days a week at all times
of the day. Electronic collection started in 1991. The current toll ring is expected

BUTTON 9781786435668 PRINT.indd 327 27/04/2022 11:07

 328 TRANSPORT ECONOMICS, 4TH EDITION

to cease operations in 2007 but there is interest in having the Oslo system evolve
into a genuine congestion pricing scheme. Trondheim introduced an area-wide
variable pricing scheme in 1991, again to finance road infrastructure and public
transportation. Since its inception, Trondheim has opted for differential rates
being charged for the morning and evening rush hours and a lower rate between
10 am and 6 pm. An innovation to this system, which was originally cordon-based
only, was the introduction of inter-zone charging in 1998.
In 1998 Rome implemented a cordon-based pricing scheme with the aim of
preserving its historical areas. Payment is required to enter the city center and
only residents and employees working in the area with secured parking spaces are
permitted to enter the city center. Authorized non-residents, numbering about
35,000, are charged a flat rate. Charges are in effect six days a week, from 6.30 am
to 6 pm on weekdays, and from 2 pm to 6 pm on Saturdays. Electronic pricing
started in 2001 with 22 toll points around the city.
Santiago de Chile’s scheme consists of a network of toll roads that cross the
city from north to south and also surround it. In 2004, the first of four major toll
road concessions involved the implementation of a pricing mechanism, and in
2006 two additional projects were completed. The main purpose has been reduc-
ing air pollution in the city, with the alleviation of excess congestion a secondary
concern. There are three levels of charge depending on traffic speeds, the lowest
being effective when the speeds are above 70 kph and the highest when speeds
fall below 50 kph. Payment is required at all times when using any of the road
concessions.
Central London has been the subject of an area pricing regime since 2003
(Leape, 2006). The charges apply to vehicles using roads in the city’s core area
between 7 am and 10 pm on weekdays, with a 90 percent discount for those living
in the area, buses, and some other groups, and free entry for ultra-low-emission
vehicles. Payment, £15 per day in 2020, is made through a variety of channels and
there is the opportunity for retrospective payment.
During the first seven months of 2006, a full-scale cordon-based road pricing
trial was implemented in Stockholm with the objective of reducing congestion
and improving environmental quality. This was followed by a referendum in
September 2006 to test views on making the scheme permanent, a proposal that
received over 50 percent of the vote. Road users were obliged to install a free-of-
charge transponder on their wind-shields and a smartcard could be bought at
different locations, recharged, or linked to a bank account to make payments.
Charges were in effect on weekdays from 6.30 am to 6 pm, and on Saturdays from
2 pm to 6 pm. A limit of $8.30 was set as the maximum a user could be charged
per day. As with other schemes, users were able to make a payment after they used
the roads. Violations were considered tax evasion as the scheme was classified
as a governmental tax (Börjesson et al., 2012). A similar scheme was adopted in
Gothenburg in 2013 (Börjesson et al., 2016).
Table 9.3 focuses on urban pricing schemes but some semi-urban schemes
have involved a somewhat different approach that sets a level of service on a road
and then adjusts the price to attain this. This has been deployed in the Interstate

BUTTON 9781786435668 PRINT.indd 328 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 329

15 scheme in California. In this case, the toll varies with traffic to maintain a
target traffic speed with the road users being given the option of a ‘free’ road
alongside the priced one. Information on average tolls for various times of the day
is made available to help in route choice decision-making. Tolls can change every
15 minutes from 5.45 am to 7 pm as well as by day of the week. The minimum
toll was 50 cents and under normal conditions it could go up to $4.00, although
when road sensors detected extremely heavy traffic it could be as high as $8.00.
The project was a three-year demonstration starting in 1996 that allowed single-
occupant vehicles to use the existing I-15 high-occupancy-vehicle express lanes
for a fee, with the aim of maximizing the use of the existing I-15 express lanes, to
fund new transit and road improvements in the I-15 corridor, and to use a market-
based approach to set tolls. The pricing project met its primary objectives, and by
2020 variable toll lanes, in a variety of forms, were to be found across the state.

9.4 Some Difficulties with Road Pricing

Theoretical economics, with its love of abstraction and mathematical exactitude,

has long been drifting away from more traditional ideas of political economy. It is
easy to show, as we have seen above, that with a few equations or diagrams, there
are benefits to be gained by manipulating road user charges to contain excessive
congestion. Implementation of such strategies, however, has been confronted by
a plethora of practical and political challenges (Gu et al., 2018). In addition, these
have not always been independent of each other, with opponents to road pricing
using technical imperfections of any scheme as a justification for attacking the
concept in its entirety.
Thus, despite the adoption of more economically based road charging
systems in recent years, and indeed more economically based congestion pricing
systems in other contexts, as we see later, road pricing is not the norm. There are
many reasons why this is the case, some of which have been explored in detail by
academics, but others remain speculative. From a review of public service atti-
tudes, Peter Jones (1998), for example, listed the following reasons for opposition
to road pricing:

• Drivers find it difficult to accept the idea of being charged for something
they wish to avoid (congestion), and feel that congestion is not their fault but
rather something that is being imposed upon them by others.
• Road pricing is not needed, either because congestion is not bad enough or
because other measures are superior.
• Pricing will not get people out of their cars.
• The technology will not work.
• Privacy concerns.
• Diversion of traffic outside the charged area.
• Road pricing is just another form of taxation.
• Perceived unfairness.

BUTTON 9781786435668 PRINT.indd 329 27/04/2022 11:07

 330 TRANSPORT ECONOMICS, 4TH EDITION

We have touched lightly upon some of the issues, such as those associated
with distributional impacts and the expenditure of revenues, when discussing the
rationale for the systems that are now being used. The topic is a large one, however,
and some more scratching of the surface is attempted here regarding several of the
other factors that have proved impediments to improving the way roads are used.
Additionally, it should be remembered, there is no single, ideal way of implement-
ing road pricing, and the rejection of some schemes that have been put forward
in the past may also have related more to their details than to the acceptability of
the concept itself. While the basic theory of road pricing is, thus, comparatively
straightforward, its detailed implementation has been subject to debate.

Political Ideology

Reluctance to adopt economic pricing principles may in part be due to a feeling

that individuals have some form of ‘inalienable right’ to mobility and that pricing
would restrict this. Matters of inalienable rights are complex, and this is not the
place to discuss them in any detail. What is perhaps germane, however, is that
rights are not free goods; there is an opportunity cost involved in acquiring and
retaining them. The costs of providing ubiquitous mobility are high, and, while
some may see ubiquitous mobility as desirable, the opportunity costs of achiev-
ing it are opaque in a world where there are inappropriate signals of the wider
economic and social costs involved. Nevertheless, notions such as the ‘freedom of
the King’s Highway’ persist, and many are reluctant to pay for the use of roads
irrespective of the economic consequences of current and future generations of
excessive congestion.
In some ways, related to this notion of a right to mobility is the position
taken in some countries that once a road has been built using public money, nor-
mally financed by taxation, then the public have the right to freely use the facility.
This, for example, is a common argument voiced in the United States, and one
of the reasons why road pricing in California has been limited to new facilities
offering a better quality of service to the existing public roads; hence the charging
regime is called ‘value pricing’. The argument that, once paid for, a facility should
be open to all runs against the basic ideas of economic resource allocation but has
emotive appeal. Unless there is optimal capacity whereby user charges should just
reflect upkeep costs, then there is a need to somehow ration the scarce road space
so that its benefits are not diluted, and pricing is the way this is normally done in
market economies and seems to work well.

Calculating the Road Price and its Impact

The simple analysis of the sort seen in Figure 9.1 assumes a lot, and especially
that the demand for road space may be represented by a continuous function. In
practice there may be kinks or discontinuities and indeed, theoretically, there may
be no optimal road price because the demand situation is such that pricing would
result in either too much traffic or too little, depending upon the levy charged.

BUTTON 9781786435668 PRINT.indd 330 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 331

This is a theoretical possibility. After all, the smooth curves used in basic analy-
sis are only expositional aids and the actual form of the demand function is an
empirical question that can only be resolved in the light of practical experience.
It is clearly difficult, perhaps impossible, however, to calculate the optimal
road price, especially because it is often context-specific. Indeed, Table 9.4 pro-
vides a very wide range of calculations from some of the main studies. To begin
with, the road price is simply the efficient way of attaining a traffic-related objec-
tive, and especially those pertaining to flow, speed, or traffic density. This objec-
tive is set by the authorities, often based on the views of traffic engineers who take
into account such things as safety and road wear in their estimates. It is largely a
subjective indicator of the type of performance that is wanted from the road. The
road price is then set to limit traffic to attain the objective.
What this charge will need to be will depend on the costs perceived by
motorists using the road; and this is largely a combination of direct costs and
the implicit money costs of travel time. In practice neither is easy to calculate.
The problem is that people’s demand elasticities depend on their perceptions and
not on objective measures. Hence, while the money costs of a trip will embrace
such things as fuel costs and wear and tear on the vehicle, because these involve
infrequent outlays, the road user tends to ignore most of them when deciding on
a journey. Travel time poses a different sort of problem because, although it is
recognized that ‘time is money’, to adopt the old saying, exactly how much money
a minute of travel time saved is worth is open to some debate. As we have seen,

Table 9.4 Early estimates of optimal road prices

Study Place Road price at peak

times (current prices)

Walters (1961) Generic US urban expressway $0.10–0.15/auto-mile

UK Ministry of Urban areas in Great Britain £0.375/auto-mile
Transport (1963)
Greater London Central London £0.60/auto-day
Council (1974)
Elliott (1975) Los Angeles $0.03–0.15/auto-mile
Kraus et al. (1976) Twin-cities expressways 1970 $0.03–0.15/auto-mile
Keller & Small (1977) Bay Area expressways 1972 $0.027–0.343/auto-mile
Dewees (1978) Toronto 1973 $0.04–0.38/auto-mile
Cheslow (1978) Berkeley 1977 $2.00/auto-trip
Spielberg (1978) Madison 1977 $1.00/auto-trip
Mohring (1979) Twin-cities $0.66/auto-trip
Gómez-Ibáñez & Boston $0.5–1.0/auto-trip
Fauth (1980)
Viton (1980) Bay Area bridges 1972 $0.154/auto-mile
Starrs & Starkie Adelaide arterial roads 1982 A$0.025–0.22/auto-km
(1986)
Cameron (1991) Los Angeles expressways $0.15/auto-mile

Sources: Button (1984); Morrison (1986). These papers contain the full references to the studies
cited.

BUTTON 9781786435668 PRINT.indd 331 27/04/2022 11:07

 332 TRANSPORT ECONOMICS, 4TH EDITION

there are a variety of techniques available for putting a valuation on time savings,
but issues arise about such things as the appropriateness of the methods used
(generically between stated and revealed preference techniques) and whether one
can sum the value of many small savings to get the value of a large saving.

Difficulty in Devising a Practical Method of Collection

Congestion varies across road networks but it is quite clearly impossible to make
a separate charge for each segment of a road network. The pioneering Singapore
area licensing scheme was, for example, simple and just involved vehicles having
to display a card to show that the driver had paid to enter the core area during
designated times. From an economic perspective, this procedure had low trans-
action costs – the production and sale of permits was cheap and enforcement at
entry points to the city became part of normal policing – and provided drivers
with prior knowledge of potential costs upon which to base decisions regarding
entering the city. It did not, however, provide any direct link to the congestion
caused by a vehicle once it was in the city, and its application elsewhere in urban
areas with far more entry and exit points would make enforcement more difficult
and costly.
A general attempt at reflecting congestion more accurately is found in the
Stockholm and Gothenburg schemes, where there are differential charges, but
difficulties arise if this is taken too far because motorists are only retrospectively
made aware of the congestion costs of their trips. Such systems are also likely to
be more expensive to install and administrate, although recent advances in ‘smart-
card’ technology have reduced this problem (Button and Vega, 2007).
Crude area licensing, even if enforced electronically as in London, is cheaper
and, since permits must be purchased before entering specified urban zones, the
full cost of a journey is made known to motorists before they enter congested
streets – if they are still prepared to do so. The disadvantage of this simpler
system is its insensitivity to changes in traffic conditions throughout the day. It is
a trade-off between the price accurately reflecting the full economic costs of a trip
and the ability to relay this to the potential road user.
Manual charging systems also inevitably lead to ‘step tolls’ in the sense that
you either pay or you do not pay, and even when there is payment the price goes
up in discrete jumps. The academic evidence is that there are considerable effi-
ciency losses when such crude charges are imposed (Arnott et al., 1993).
The United Kingdom’s Smeed Report on road pricing recognized as early as
1964 the limitation of this type of simple fee collection nearly 60 years ago and
discussed various electronic options. The technology of the day was, however,
limited and expensive. Things have changed and now a variety of alternatives are
available and many have been adopted.
There are two broad ways of implementing more sophisticated forms of
electronic road pricing. The first uses automatic vehicle identification (AVI),
which records centrally the congestion costs of individual trips for each vehicle.
The second does not identify individual vehicles but deducts the cost of using

BUTTON 9781786435668 PRINT.indd 332 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 333

congested roads from a stored value medium where the proprietor of the system
is not able to establish who is using the facility. This latter approach, which might
usefully be called non-smartcard technology, can be extended to the use of smart-
cards, similar to credit cards, which automatically debit the costs of trips directly
from bank accounts or charge them to a credit card account (such as Visa or
Mastercard).
There was early experimentation with equipment and operational practices
in Hong Kong. On the operational side, the two-year experiment during the mid-
1980s involved fitting over 2,500 vehicles with AVI together with the setting of
electronic loops in the road surface at the edge of charge zones. When a vehicle
crossed a boundary, a power loop energized its AVI that sent a message, via induc-
tive receiver loops, to a roadside recorder. The technical and economic feasibility
of the system used was found to have achieved 99 percent effectiveness and reli-
ability against the criteria set for it.
There are, however, practical difficulties with the system. An early concern
that emerged involved confidentiality of the information gathered. The issue is
still a sensitive and possibly inevitable one in some countries. The Hong Kong
experiment, while successful in showing that the technology used was reliable and
that real-time adjustments to charges were possible, also raised concerns that the
information collected could infringe on personal liberties. The electronic conges-
tion charging schemes in Stockholm, London, Gothenburg, and elsewhere that
have been initiated since that time have taken care to minimize the possibility of
them being used for ‘tracking’, with powerful legal protections being built into the
institutional structures in which they operate. The widespread use of electronic
payment across a huge number of sectors also means a social acceptance of the
mechanism has developed more generally.
The London system makes use of off-the-shelf camera-based automated
number plate recognition technology to enforce speed limits. Video recognition
closed circuit television cameras target optical characters on the number plates of
vehicles at a rate of one per second even if the vehicle is traveling up to 100 mph.
Charging is automated through computerized systems that deduct funds from the
smartcards or through billing.
The systems now deployed in Stockholm, Norway, and Santiago use AVI
involving the installation of a transponder or a smartcard that are detected by
beacons installed at toll booths and other check points. The exchanges of infor-
mation between the tag and radar are possible using simple radio frequency iden-
tification technology. Regular users have their vehicles fitted with an electronic tag
while visitors can make cash payments using specific lanes. Violators are identi-
fied through camera-based recognition of the vehicles’ license plates and charged
a retrospective fee.
Singapore’s second-generation scheme does not require a centralized com-
puter system to keep track of vehicle movements since all charges are deducted
from an inserted smartcard at the point of use, with records of transactions kept
in the memory chip of the card belonging to the individual. As a further step
to assure the public of privacy, all records of transactions required to secure

BUTTON 9781786435668 PRINT.indd 333 27/04/2022 11:07

 334 TRANSPORT ECONOMICS, 4TH EDITION

payments from the banks are erased from the central computer system once this
is done – typically within 24 hours.
In a sense, these types of devices are often just more efficient ways of cordon
charging. While it is possible to develop these types of technologies to provide
real-time charging that reflects the actual levels of congestion when a vehicle is
using a part of the road network, they still suffer the major economic drawback
that consumers should know the price of something before coming to the pur-
chase decision. The problem is that the drivers’ own actions affect the price that
should be paid and this makes real-time charging extremely difficult in practice.
Efforts at circumventing this problem have been made. One approach is to
provide advanced information on local traffic conditions so that a driver has more
insight as to the likely road price to pay. Modern technology facilitates the provi-
sion of real-time traffic information on relevant parts of the network.

Distributional Effects

Technically, in economic terms, road pricing leads to a Hicks­–Kaldor societal

welfare improvement (a subject we shall return to when looking at investment
criteria in Chapter 11). What this means is that, overall, while some people will be
worse off as a result of the price (if they were not it would be a Pareto improve-
ment), those who benefit will be able to compensate them and still retain some
gains from the scheme. In the case of road pricing, it is the authority that imposes
the charge that benefits from the revenues collected (represented as C2JKC1 in
Figure 9.1), with the average road user being worse off. This leads to two related
issues: first, defining exactly which road-using groups lose and by how much; and,
second, whether and by how much the authorities should provide compensation.
The first of these issues is dealt with here, and the second in the subsection that
follows.
Under a road pricing regime, road use depends on the capabilities of poten-
tial users to pay the congestion charge. Whether this would result in undesirable
regressive effects on social welfare is an empirical question. It is likely that surface
public transport, as we see below when discussing London, would move more
freely and would provide a better service for the lower-income groups that tend to
patronize it. Also, the wealthy are likely to benefit from being able to ‘buy’ uncon-
gested road space, a situation they value because of the importance they attach to
time savings. In contrast, middle-income groups could be forced to switch from
private to public transport, a mode they consider inferior.
Inter-personal welfare comparisons of this type are difficult to make, but it is
possible, for example, that if some of the revenue from road pricing were directed
towards further improvements in public transport (which may or may not mean
heavily subsidized fares), then the adverse effects on the middle-income group
could be substantially reduced. Whether this is socially acceptable or desirable is
a political question.
All forms of pricing inevitably lead to differential consumption patterns
that are to some extent influenced by income levels. Western-style economies

BUTTON 9781786435668 PRINT.indd 334 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 335

accept this, partly because it prioritizes what people want to consume but also
because income is itself seen as reflecting the endeavors of individuals. If there
are concerns about the income distribution, this is treated as a normative issue
that should be handled through redistribution as part of a political process. Road
pricing is no exception to normal pricing principles. Much of the debate about
the distributional effects of congestion charging centers on the ability of poorer
individuals to be able to buy road space, but the poor are unable to buy many
things. Thus, the problem is really one of income distribution per se, rather than
of the price mechanism.
There is a second area of distributional interest, namely how the impacts
of road pricing are spread over geographical space. In some instances, there is
a clear link between this and the income distribution issue because of the cor-
relation between land-use patterns and household incomes. In another context,
however, there are concerns that road pricing will penalize businesses in areas
where road users pay for their congestion effects as opposed to those that do not.
The evidence on this is limited, but a study of the London road charging scheme
by Quddus et al. (2007) found that although individual stores were adversely
impacted by the road price, it did not affect overall retail sales in central London.
Isolating the implications of the congestion charge is, however, not easy because
of a failure of the Underground’s Central Line during the initial period of its
introduction, the heightened fear of terrorism at the time, and the on-set of a
general macroeconomic downturn.

Expenditure of Revenue

Linked to the distributional issue, Clifford Sharp (1966) pointed out that the rev-
enues from road pricing need to be re-allocated with a degree of circumspection.
A distinctive outcome of the road price is that the revenue does not normally go
to a private company that then makes commercially oriented decisions on how to
spend it, but rather to a public agency that has less clear-cut criteria regarding the
use of the revenue flow. This poses short-term practical as well as intellectual chal-
lenges. Sharp assessed some alternatives for using the revenues, including reducing
other road user charges. This has an appeal if the funds are not being used to fulfil
other explicit policy objectives, such as reducing CO2 emissions from carbon fuels
or acting as a sumptuary tax to finance other, socially approved expenditures.
In strict efficiency terms, the revenues should be used for purposes that yield
the highest social return. Since in most cases large revenues from optimal conges-
tion charges themselves indicate inadequate capacity, a strong case can be made
that some should go into providing additional road capacity (Button, 2006a).
A related but slightly different argument is that the F1 – F2 motorists priced
off the roads in Figure 9.1 should receive some compensation. Direct transfers
back to former motorists, however, pose the problem that they are likely to use
at least part of the money to ‘buy back’ road space. An alternative method of
compensating motorists adversely affected may be to use the road pricing rev-
enues to construct more roads, but there are issues here. Daniel Graham and

BUTTON 9781786435668 PRINT.indd 335 27/04/2022 11:07

 336 TRANSPORT ECONOMICS, 4TH EDITION

Stephen Glaister (2006), for example, take a wide geographic perspective of road
pricing across the entire United Kingdom’s road network and find that while its
introduction in urban areas would increase the revenues enjoyed by the relevant
road authorities, revenues would fall on other parts of the network that are far
less used.
Given that investment decisions should be based upon a much wider range
of criteria than simply the revenues raised from road charges (see Chapter 11), an
alternative approach would be to treat the revenues as a pure tax income and to
use them as part of general public expenditure, and in this way wider problems
of efficiency and distribution may be tackled. In this vein, from a pragmatic,
political-economy perspective of forming coalitions of interests that would
­
endorse road pricing, Phil Goodwin (1989) and Kenneth Small (1992) argued for
spreading the expenditure of revenues across different interest groups, includ-
ing road users (in the form of investment in additional capacity and intelligent
transportation technology to better manage existing capacity), public transport
authorities (in the form of capital and operating subsidies), and the general
public (in the form of reduced taxation or increases in public-endorsed non-­
transportation expenditures). The trick, of course, would be to find the appropri-
ate balance of this redistribution between the various groups that would carry
local political opinion. This has proved, however, difficult in many cases.
Several efforts have been made to ask those affected how they would like
to see the revenues from road pricing spent. Erik Verhoef et al. (1997) did an
extensive survey in the Netherlands, and found that, in decreasing order, the
preferences were for investment in more road capacity, reduction in vehicle taxa-
tion, reduction in fuel taxation, investment in public transportation, subsidies for
public transportation, investment in carpooling facilities, general taxation cuts,
and expansion of other forms of public expenditure.
In practice, most road pricing schemes do involve a high degree of earmark-
ing of revenues, either because the local authority has needed to do this to get
policies through or because central government has mandated it as part of its
wider policy agenda. In the case of the toll roads in Norway, although not strictly
a road pricing scheme, the revenues were specifically earmarked for local trans-
port expenditures on roads and buses. In the United Kingdom, the revenues from
London’s congestion charging must, by national law, be spent on transport in
London, but much of it goes to subsidize bus services.

Road Pricing is a First-best Solution in a Second-best World

Marginal cost pricing of road space is only strictly optimal if all other goods in
the economy are also marginal cost priced; it forms a benchmark. To deal with
situations where marginal cost pricing is not universal, there is a need to adjust
the road price to reflect these distortions elsewhere in the system. For example,
if the costs of public transport are set above their costs inclusive of congestion,
then the charges imposed on roads should also be set above the simple calculation
of the road price.

BUTTON 9781786435668 PRINT.indd 336 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 337

While the concept is relatively simply to follow, operationalization can be

difficult, as shown by Verhoef (2002). The nature of pricing imperfections can be
complex and extensive; they may also be positive, in the sense of reinforcing the
road price, or they may conflict with it. They may also be within the road system;
for example, introducing congestion pricing on some roads but not others affects
the relative attractiveness of various routes and the economic activities located
about them. At the micro level, charging for only some routes may deflect some
traffic onto unpriced ‘rat-runs’, which that not only has adverse effects for con-
gestion, but may also have adverse environmental implications. There is a danger
that one region or city may not introduce road pricing, hoping to attract business
from adjacent areas that do. Similar problems arise when it comes to airport and
seaport pricing that are discussed later.
These types of problems can easily be seen in Figure 9.2, which shows a
situation where there are two roads linking A and B, one free and the other with
an optimal road price in place. (One could easily interpret the unpriced route as
an alternative mode and the arguments would remain valid.) The roads are pure
substitutes for each other in the sense of the generalized costs, using the sum of
the monetized travel cost, c, plus the toll r where it is levied. The costs will be
equal for both roads (following the so-called ‘Wardropian’ equilibrium principle),
with people trading off lower monetized costs of a non-priced trip against the
higher-priced but faster facility. Drivers are nearly identical in their preferences
(for example, their values of time savings are the same), except when demand is
not perfectly elastic.
Given the different prices, the challenge is to set a second-best charge for
the priced road that maximizes the efficiency with which both roads are used.
In general, this will be below the optimal road price when only one link is con-
sidered. This is because the charge on the priced road will improve the situation
over conditions when there is no access fee. But this charge will divert some traffic
to the other, unpriced road, adding to the level of congestion there. Thus, there
is a trade-off in terms of improving travel conditions on the facility that is road
priced, but worsening them on the route that remains free. Ideally both routes
should have a congestion charge, but we are assuming that is not possible. It is the
establishment of the optimal trade-off in this constrained case that necessitates
the second-best pricing.

Priced road

A B

Free road
Figure 9.2 The classic two-route second-best case

BUTTON 9781786435668 PRINT.indd 337 27/04/2022 11:07

 338 TRANSPORT ECONOMICS, 4TH EDITION

9.5 Impacts of Road Pricing

There are numerous academic and official studies that have examined the poten-
tial effects of using road user charges of one form or another to reduce urban
traffic congestion. Rather than rehash these, or try to produce some meta-analysis
of their statistical conclusions, here we look briefly at what has transpired when
road pricing has been adopted. One thing is clear, the evidence from the various
road pricing schemes that have been initiated and heavily studied to date shows
that drivers do respond to prices and that traffic congestion can be controlled
through direct user charges. Besides the direct traffic effects on congestion, the
initiatives have exerted secondary influences, many of which were generally antici-
pated but not on the scale that they have occurred.
There are inevitably problems in assessing the pure impacts of road pricing.
All the schemes that have been introduced to date have been part of larger pack-
ages, and isolating the effects of any element is difficult. The Singapore area
licensing, for example, was accompanied by additional bus capacity and invest-
ments in park-and-ride facilities as well as exemption for high-occupancy cars.
Also, the charges themselves have inevitably embraced political compromise
that, for example in the London case, resulted in significantly lower charges for
residents and some types of road users such as taxis, trucks, and two-wheeled
vehicles. Added to this can be unforeseen circumstances that affect the system,
such as the closure for a period of the main Underground railway line in London
at the time congestion charging was initiated.
The world is also not static, and traffic growth takes place over time.
Consequently, there is a need for the level of congestion charges to change with
circumstances. However, this makes it hard to estimate anything but very short-
term ex post elasticities because of the problem associated with allowing for these
background growth trends. The outcomes of any scheme at any point in time will
thus be contextual on exactly how recently road pricing has been imposed and the
extent to which it has been modified in the light of experience.
Table 9.5 provides details of the outcomes of some of the major urban road
pricing schemes in terms of the direct effects on automobile traffic, on the traffic
situation more generally, and on public transport. The pattern that emerges is
clear and does not need much in the way of elaboration; road pricing where it has
been employed has reduced the use of cars, improved traffic flows, and led to a
modal shift toward public transport during peak periods.
There have also been secondary benefits, such as reduced traffic-generated
environmental pollution, although this is difficult in practice to quantify and
evaluate because people allocate their time saved from not being held up in traffic
in ways that in themselves can result in adverse environmental consequences. Just
as an example, it has been estimated that the Stockholm scheme reduced CO2 by
10 to 14 percent in the inner-city area and by 2 to 3 percent in the surrounding
area, although there was little impact on noise levels, while the London charging
scheme produced an annual $6 million benefit in terms of reduced CO2 emissions,
and $30 million in lower accident costs.

BUTTON 9781786435668 PRINT.indd 338 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 339

Table 9.5 The effects of major road pricing schemes

City Traffic effects Congestion effects Public transport effects

Singapore, –44%; –31% Average speed Modal shift, from 33% to

1975–98a by 1988 increased from 46% trips to work by city
19 to 36 kph bus, 69% in 1983
Trondheim, 1991 –10 % n.a. +7% city bus patronage
Singapore, 1998b –10 to –15% Optimized road Slight shift to city bus
usage, 20 to 30 kph
roads, 45 to 65 kph
expressways
Rome, 2001 –20 % n.a. +6%
London, 2003 –18% 2003 vs –30%; 1.6 min/km +18% during peak hours
2002; 0% 2004 typical delay 2003, bus patronage 2003,
vs 2003 2004 vs 2002 +12% in 2004
(2.3 min/km)
London, 2005c Small net –22%; 1.8 min/km Bus patronage steady
reductions typical delay
–4% 2005–2006
Stockholm, 2006 –30% 2006 –30 to –50% journey +6%
vs 2004 time

Notes:
a. Although called Area Licensing Scheme, the system was a cordon toll rather than an area license.
b. Electronic fee collection introduced.
c. New rate introduced.
n.a. = not available.

Not all the systems described are strictly road pricing; the Norwegian ones, for
example, were introduced primarily for revenue-raising. In addition, while the
Singapore scheme showed that road pricing can reduce congestion successfully, it
also showed that the effect of any scheme seems rather more difficult to forecast
than some advocates suggested. The actual details of this scheme are obviously
tailored to the geography and political climate of Singapore and have not been
exactly replicated elsewhere. It was also questionable whether the actual prices
charged are truly optimal or whether they are excessively high, acting as a method
of revenue collection for the government as much as an instrument of microeco-
nomic resource allocation. Quite clearly transport may be a legitimate field for
pure indirect taxation but in assessing the effectiveness of a road pricing scheme
it is important to isolate the price efficiency aspect from that of taxation per se.
Of course, not all outcomes have been exactly as predicted. In very many
cases, the authorities have under-estimated the power of pricing and the reduc-
tion in traffic has exceeded forecasts, and the revenue collected has been less than
expected. Equally, public transport has generally been found to be a more popular
substitute than expected once travelers are aware of the congestion costs of using
cars, the park-and-ride facilities introduced in Singapore being a notable excep-
tion to this. This poses some operational challenges in terms of budgeting, sched-
uling, and the provision of public transport more generally, but has not seriously
brought any scheme into difficulties.

BUTTON 9781786435668 PRINT.indd 339 27/04/2022 11:07

 340 TRANSPORT ECONOMICS, 4TH EDITION

Exhibit   The initial London congestion charge scheme

By 1998 traffic speeds across London were down to 10 mph. In the larger London Area
drivers were spending about 30 percent of their time stationary during peak periods. Traffic
management measures and public transportation subsidies had failed to halt the rising level
of congestion. In 2003, however, a £5.00 daily charge (later raised) was imposed for driving
or parking in central London between 7 am and 6.30 pm on weekdays.
The idea was based on the notion of road pricing initiated by Pigou in 1920 and ‘discovered’
in an obscure paper by Milton Friedman and a student by the mayor of London, Ken
Livingstone. The congestion charge was imposed within six months of Livingstone’s election.
The system entails a toll ring with no effort to charge for the marginal costs of individual
trips within the zone.
The £5.00 daily fee was based upon estimates from prior traffic studies, but in subsequent
economic analysis was found to approximate to the marginal cost of driving two miles in
inner London. Transactions costs for road users were minimized by offering a variety of ways
(retail outlets, kiosks, call centers, and the internet) for pre-recording license information and
for paying the charge. Ex post payment was also possible at a small premium within a defined
time period, and a surcharge was added to payments made just prior to travel.
Enforcement was, given the photographic technology used, akin to that for speeding or other
traffic violations. License plate numbers were recorded and non-payers were then sent
penalty notices. There were immediate problems with identification because of the basic
nature of the camera technology involved, and improved camera technology was introduced.
The short-term implications of the scheme are seen in the first table. A clear reduction in
car traffic is seen, together with an increase in taxi and bus traffic. It should be said that
this was before the development of hail-by-app cabs (for example, Uber), which have had a
major subsequent impact on mode spilt. In terms of speed, all-day average network speeds
rose from about 8.9 mph to 10.4 mph in the month after the charges were introduced.

Vehicle type Year Change

2002 2003

Cars 771 (47) 507 (35) –34

Vans 287 (18) 273 (19) –5
Trucks 73 (8) 68 (5) –7
Taxis 256 (16) 312 (21) 22
Buses 54 (3) 65 (5) 21
Motorcycles 129 (8) 137 (9) 6
Bicycles 69 (3) 89 (6) 28
All vehicles 1,640 (100) 1,451 (100) –12

Note: Traffic in thousands of vehicle-kilometers (and percent).

While these sorts of data provide indicators of traffic effects and the implications for modal
split, they are inadequate indicators of economic benefits that reflect, amongst other things,

BUTTON 9781786435668 PRINT.indd 340 27/04/2022 11:07

 341
OPTIMIZING TRAFFIC CONGESTION ­

the effects of charges on person travel and operating costs of the system. The second
table provides a fuller social cost–benefit assessment of the charges completed by London
Transport in 2006.

Annual costs (£ millions in 2005 prices)

Administrative 5
Scheme operating 85
Set-up (opportunity costs/depreciation) 23
Traffic management (e.g., increased bus services) 20
Charge payer compliance 30
  Total 163
Annual benefits (£ millions in 2005 prices)
Time savings & reliability
  Car users 110
   Vans and trucks 35
  Taxis 40
  Buses 42
Deterred drivers –25
Reduced accidents 15
Reduced CO2 emissions 3
Other resource savings 10
  Total 230

The calculations suggest an annual social welfare gain of $67 million (in 2005 prices), but
there are questions concerning some of the assumptions they make. For example, the
demand and cost curves used do not take full account of the differentiated nature of road
usage and the heterogeneous values of travel time. Further, the calculations are based upon
first-best assumptions, namely that all markets are based on marginal cost pricing, but there
are arguments that local labor markets would be affected according to how the revenues
from the charging are spent. Linked to this is the question of efficiency. There may be net
benefits associated with the London congestion charging regime, but this does not mean
that resources are not being wasted. There may be inefficiencies in the infrastructure used
and the ways in which the system is managed.
See also: J. Leape (2006) The London congestion charge, Journal of Economic Perspectives, 20,
157–76.

Enforcement has posed challenges to the authorities responsible for some

schemes, although there seems to have been a learning process. Transport for
London, for example, reported in 2006 that the number of penalty charge notices
fell throughout 2005, with 21 percent fewer charges overall. In Rome, violations
in the control zone resulting in fines fell from an initial 22 percent of the overall
access flow to about 7 percent of those registered in 2005–2006. Reductions in
traffic flows were mainly achieved through this reduction of illegal traffic entering
the limited traffic zone. Better monitoring was made possible by the installation
of electronic gates in 2001; illegal traffic decreased from an estimated 36 percent
before their activation to under 10 percent after a year and a half.

BUTTON 9781786435668 PRINT.indd 341 27/04/2022 11:07

 342 TRANSPORT ECONOMICS, 4TH EDITION

9.6 Parking Policies

In some cases, often for political reasons but also possibly because it is more
cost-effective to operate, a parking charging policy may appear preferable to road
pricing as a means of containing congestion externalities. This is true if much
of the traffic is terminating in the area concerned and there is a lot of circulat-
ing traffic seeking parking spots because of inadequate signage. Again, while
the most extensive coverage of parking is in the context of urban car traffic, the
principles can be equally valid for stands at airports or berths at seaports; for
example, high stand charges for aircraft at air terminals can stimulate changes in
flight patterns and traffic concentrations.
Parking is seldom provided and operated in an economic manner, and
optimal fees are rare (Young, 2001). Parking can either be considered in a first-best
world, where there is appropriate pricing elsewhere, including road pricing, or in
a second-best context as a means of limiting traffic congestion when there are not
appropriate prices charged for other goods and services (Button, 2006b).
The first-best approach is simply to set the parking charge equal to the mar-
ginal cost of parking, including the congestion costs of seeking a parking space.
In practice, their detailed effects can be influenced by the institutional structure
of an area; for example, any attempt to influence the parking market in one
part of a city will influence the scale and nature of the induced externalities in
adjacent boroughs by, depending on the nature of the policy, attracting or dis-
couraging terminating traffic and affecting such things as the land value of office
premises with parking facilities. Institutionally this can lead also to game playing
by employers who argue for favorable parking concessions as a pre-requisite for
them locating in any local administrative unit. As behavioral economics shows,
these longer-term costs are generally more difficult to handle because of hyster-
esis. Once people change their activity patterns it is often difficult to encourage
them to adjust to optimal behavior; there are stranded cost considerations and
uncertainties to contend with.
A simple way to look at the role of parking fees in a second-best context
is set out in Figure 9.3 (Verhoef et al., 1995). Quadrant A has the marginal cost
and average cost curves (where we assume simple interaction congestion to be the
only externality) but added to this is the cost of parking (reflecting the oppor-
tunity costs of a parking space and assuming for simplicity that individuals’
parking durations are identical). This gives the combined costs curve of TMC.
The optimal road price plus parking fee based on traffic density then become RP
in the diagram. We now take the extreme case that for some reason the road price
element of RP cannot be collected.
On the assumption that all vehicles paying appropriate prices which enter
during the study period will find parking, then we can draw the 45° line in
quadrant B of the diagram, which relates traffic density to parking occupancy.
The parking fee for the period, F, required to achieve this is then mapped out
in quadrant C. This demand curve again assumes that willingness to pay the
set fee will guarantee a parking place; if this assumption is not valid because of

BUTTON 9781786435668 PRINT.indd 342 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 343

Cost
A
TMC

MC
AC

RP
Demand
FO
Parking fee DO Da Density

PO

Pa
Demand

C B

Parking occupancy

Figure 9.3 A simple analysis of parking charges

suboptimal capacity, the analysis requires modification. The demand curve for
parking places is derived from the difference between Demand and the AC in
quadrant A. The optimal parking fee here (FO) is acting as a rationing device and
its size could be adjusted accordingly if a suboptimal road price were initiated
at some later date.
Parking policy has its own limitations. First, it only affects certain traffics. It
obviously has little impact on through-traffic and, indeed, by deterring stopping
traffic, it may encourage movements through the city centers. It also has distribu-
tional consequences. It bears more heavily on those making short journeys since
the parking fee will form a relatively large part of the overall costs of their trips
compared to those driving a greater distance. The policy is also likely to be rela-
tively insensitive to actual levels of congestion since it is acting on a complement
to road use rather than road use itself. From a practical perspective, many parking
places are privately owned and hence direct control of prices is difficult for policy-
makers, although indirect measures, such as taxing land used for parking, provide
a mechanism for tackling the problem.
There is also the question of just how sensitive parking decisions are to
charging: the elasticity of demand. Table 9.6 provides some calculations for

BUTTON 9781786435668 PRINT.indd 343 27/04/2022 11:07

 344 TRANSPORT ECONOMICS, 4TH EDITION

Table 9.6 Parking elasticities in Sydney

Preferred Less preferred CBD fringe

CBD CBD

Car trip, preferred CBD 20.541 0.205 0.035

Car trip, less preferred CBD 0.837 20.015 0.043
Car trip, CBD fringe 0.965 0.286 20.476
Park-and-ride 0.363 0.136 0.029
Ride public transit 0.291 0.104 0.023
Forgo CBD trip 0.469 0.150 0.029

Source: Hensher and King (2001).

Sydney, Australia, dividing parking locations up into more and less preferred
sites. It provides both direct elasticities and cross-elasticities between locations.
It shows, for example, that while a 1 percent rise in parking charges at preferred
urban locations will reduce parking there by 0.54 percent it will increase parking
at less preferred sites by 0.84 percent and fringe parking by 0.97 percent. In other
words, simply raising parking charges at the most congested parts of the city
can spread the demand for parking to other locations. Stated-preference analysis
in Israel (Albert and Mahalel, 2006) finds that the demand elasticity regarding
parking fees is considerably lower than for road pricing, again indicating a lack
of traffic sensitivity to their introduction.

9.7 Congestion Pricing at Airports

Congestion in air transport arises at airports and during flights. As with road
transport, the lack of a genuine benchmark for measuring congestion means that
estimating congestion costs at different traffic levels or excess congestion levels
in either context is difficult. The United States Federal Aviation Administration
(FAA) and many other civil aviation organizations use a 15-minute delay on the
published schedule as an indication of congestion, and on this basis 25 percent
of flights suffered from excess congestion at the United States’ major airports
in 2006.
The economic cost of these delays is difficult to estimate because, unlike
road traffic congestion where car drivers are trying to minimize their travel time
and vehicle costs, commercial airlines are the direct customers for most airports’
slots and in a deregulated world, airlines are largely motivated by profits. In other
words, the nature of the opportunity cost associated with congestion is perceived
differently when the user of a facility is a commercial entity than when it is an
individual – the effects of congestion on an airline affect its bottom line, whereas
the effects of road congestion on an individual are on that person’s utility, which
is usually modeled in terms of financial costs and the monetary value of lost time.
Airports vary in their physical form, ownership, and motivations. While
there has been a gradual trend towards privatization in Europe, and especially in

BUTTON 9781786435668 PRINT.indd 344 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 345

the United Kingdom, and modifications to the regulations regarding how they
operate, United States airports are largely owned by local authorities of one
kind or another. There are few, if any, fully privately owned airports in the world
driven purely by unrestrained profit motives and this may, in part, explain why
their efficient operations are often an issue of public interest.
Many airports are already congested (Table 9.7 gives data for major
European facilities) and it is unlikely that runway capacity will in the foreseeable
future expand in line with the projected increase in traffic, and certainly there will
be many facilities where capacity will remain stagnant or grow much more slowly.
There are often airports where demand is high but additional capacity will not be
constructed because of local opposition to adverse local environmental effects.
There are few quick-fix technologies on the horizon. Enhanced air navigation
equipment and air traffic control practices, such as those being adopted in the
United States under the NexGen initiative and by EUROCONTROL under the
Single European Sky program, are likely to offer only temporary relief in terms of
controlling flows into and out of airports. The challenge in these circumstances is
to make the best use of the capacity that is available, and to adopt more efficient
management practices.
The existing mechanisms for allocating take-off and landing slots generally
rely upon administrative procedures. Many countries, or groups of countries such
as the European Union, have regulations governing the ways slots are allocated,
embracing a combination of formal and informal arrangements often overseen by
a scheduling committee that includes incumbent users. It is normal to have some
grandfathering element that favors existing users, and often there are limitations
on the ability of incumbents to trade their slots with other carriers. To assist in
coordination across airports, every flight requires a take-off and landing slot.
There are also various inter-airport and airline meetings and, for international

Table 9.7 
Delays suffered by major European airlines at large airports in 2006

Airport Departures Arrivals

Flights (%) Minutes (average) Flights (%) Minutes (average)

London Heathrow 31.8 34.9 30.3 38.9

Madrid 31.3 45.7 33.1 44.3
London Gatwick 30.7 32.5 31.5 37.8
Paris CDG 28.6 38.1 23.4 37.4
Barcelona 27.2 43.4 28.0 41.2
Rome 26.3 41.8 22.8 42.0
Athens 25.2 40.2 27.7 41.0
Frankfurt 25.0 36.9 24.6 40.0
Dublin 24.4 44.6 29.5 39.7
Larnaca 24.3 50.9 30.1 49.8
Milan Malpensa 24.2 44.3 22.1 41.6
Lisbon 24.1 42.6 30.2 37.5

Source: Association of European Airlines.

BUTTON 9781786435668 PRINT.indd 345 27/04/2022 11:07

 346 TRANSPORT ECONOMICS, 4TH EDITION

routes, since 1947 the International Air Transport Association (IATA) has held
twice-yearly schedule conferences now involving some 300 airlines and 200
airport representatives.
The situation is different in the United States where anti-trust laws mean that
airlines simply schedule flights taking into account expected air traffic control and
airport delays. An exception initiated in 1969 under the High-density Rule tem-
porarily designated five ‘slot-controlled airports’ (JFK and LaGuardia in New
York, O’Hare in Chicago, Newark in New Jersey, and Washington DC’s Reagan
National), where the federal government limited the number of aircraft move-
ments during certain hours although the airlines from 1986 could buy and sell, or
lease, the slots designated for domestic use amongst themselves. This regime has
been modified several times since.
Although there are variations by country and airports, the fees charged for
a slot are based largely on the size of aircraft, with some variations that reflect,
for example, noise emissions, and determined on cost recovery criteria. This is a
sort of unequally weighted Ramsey Pricing concept. Where there are fee differen-
tials, they seldom embrace a full consideration of congestion costs. There is thus
no reason to assume that this type of arrangement results in anything like the
optimal use of slots at one airport, let alone across a network.
There are few incentives for re-allocation between carriers as the demand for
their services fluctuates even when this is permitted. While there are wide varia-
tions, it is often difficult for new carriers to enter specific markets, and especially
those dominated by a major airline. Equally, there are less than maximum incen-
tives for incumbents to make efficient use of slots they retain in terms of the con-
gestion that is imposed on other operators. Indeed, congestion may be used as an
entry deterrent measure to limit competitive market entry.
While there is a temptation to simply transfer the road pricing concept to air
transport infrastructure, there are differences. Roads are, for example, used by an
atomistic set of independent ‘customers’ who cannot control the trips of others,
whereas air transport facilities are used by a relatively small group of airlines that
often have significant control over their actions. In some cases, a carrier is a near-
monopsonist buyer of services at an airport. This is so at many of the most con-
nected international hub airports in Europe and the United States (see Table 9.8).
What one has in these circumstances are bi-lateral monopolies or oligopolies,
with airlines providing countervailing power to the airports. Further, while it
may be reasonable to apply continuous functions to the estimation of road traffic
congestion, although cars are clearly discrete units, this is less sensible regarding
aircraft movements that are far less numerous.
As alluded to earlier, there are also other differences that affect the measure-
ment of congestion costs. While the analysis of production externalities is similar
to the user-on-user externalities that form the basis of the economic analysis of
road congestion, the motivation of the actors involved is not consumer surplus
maximization (usually dominated by travel time costs) but rather producer
surplus maximization. Commercial airlines in most markets are concerned only
with profits when making operational decisions, and may well impose significant

BUTTON 9781786435668 PRINT.indd 346 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­347

Table 9.8 Share of international flights by airline at Europe’s and America’s ten largest
airports (2019)

Airport Connectivitya Carriers Share of

flights (%)

Europe
London Heathrow 317 British Airways 51
Frankfurt 309 Lufthansa 63
Amsterdam 279 KLM 52
Munich 259 Lufthansa 59
Paris Charles de Gaulle 250 Air France 50
Istanbul Airport 187 Turkish Airlines 79
Madrid 154 Iberia 46
Moscow Alexander S. Pushkin 150 Aeroflot Russian Airlines 80
Rome Fiumicino 139 Alitalia 43
Zurich 114 SWISS 54
North America
Chicago O’Hare 290 United Airlines 46
Toronto Pearson Airport 251 Air Canada 59
Hartsfield–Jackson Atlanta 247 Delta Air Lines 79
Los Angeles 219 American Airlines 22
New York John F. Kennedy 186 Delta Air Lines 37
Dallas–Fort Worth 185 American Airlines 85
Miami 184 American Airlines 75
Houston George Bush 179 United Airlines 81
Newark Liberty 169 United Airlines 70
Vancouver 155 Air Canada 47

Note: a. Direct international destinations served.

Source: Official Airline Guide.

time costs on large numbers of their own passengers if this ensures high profits
from the remainder, including those using connected services.
Figure 9.4 provides a simplified version of the use of a congestion charge
regime for take-offs on a single runway on the premise that there are numerous
airlines seeking rights: the atomistic case. The airport just charges an account-
ancy fee to cover costs. Functions are linear for ease of drafting and this does not
affect the argument. With each additional flight, the marginal aircraft impacts
other flights that the initiating airline does not take into account in its decision-­
making – seen as the divergence of the MC and the MC* curves. The result is
a flow through the runway of F1 involving a marginal cost of MC1. The flight,
however, will impose congestion on other carriers’ activities affecting their profit-
ability as their costs rise and as customers’ lower willingness to pay for a poorer
service adversely affects revenues. The outcome is an aggregate producer surplus
for the airlines at the airport of {a + c} – {g + h + i} when there are F1 flights. This
will be zero if there is an atomistic market.
If there were a monopsony airline at the airport bearing congestion effects
internally, it would take MC* as its reaction function and manage a flow of F2.

BUTTON 9781786435668 PRINT.indd 347 27/04/2022 11:07

 348 TRANSPORT ECONOMICS, 4TH EDITION

$
MC*

Congestion charge i
MC2

f g h
MC
MC1
c d e
a b

Demand = AR

0 F2 F1 Flights per hour

Figure 9.4 Congestion costs at airports

Such an airline can prioritize flights for take-offs at the airport and, to a lesser
extent, en route by allowing overtaking. The resultant producer surplus to the
airline is then {a + c + f} which will be above zero. Similarly, with an airport
employing marginal cost pricing akin to a road congestion charge, and thus
levying a take-off fee that embraces congestion, the flights would be F2.
While this simple analysis, essentially just a transfer of the Pigouvian frame-
work to a situation where the infrastructure users are interested in economic rent
maximization, can be insightful, it does miss some essential elements of runway
economics. In the analysis, the degree to which any individual flight by an airline
affects other flights depends, therefore, on the nature of the carrier operating the
services and on the number of slots an airline uses. In many cases, where there
is a dominant carrier, the impact of any flight is on the other operations of that
carrier. What this intuitively means for charges is that, when there is a single
dominant carrier operating from a not-for-profit airport not actively seeking to
exploit monopoly power, there is no need to initiate congestion charges. There are
potential problems, however, if the airport is itself rent-seeking. As with most bi-
lateral monopoly situations, the slot prices will be bounded, but indeterminate, in
this case. This is because a market dominated by a profit-maximizing monopoly
tends to have high prices from users whereas a market dominated by a profit-
maximizing monopsony tends to seek lower prices for those services.
When combined into a bi-lateral monopoly, the buyer and seller cannot
both maximize profit simultaneously and are forced to negotiate. The resulting
price could be anywhere between the higher monopoly and the lower monopsony
prices depending on the relative negotiating power of each side. What theory sug-
gests, however, is that, in formulating its case, the monopoly airline will think in
terms of being able to internalize its congestion costs.
Where there is a degree of competition amongst airlines, however, conges-
tion costs are not fully internalized and congestion pricing become relevant. Most

BUTTON 9781786435668 PRINT.indd 348 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 349

airports are used by several airlines of different sizes with diverse scheduling
desires leading to some degree of imperfect competition. In these cases, there
will be some internalization of congestion costs but, from a social welfare maxi-
mization perspective, a need for congestion charges remains to internalize the
remainder. The situation is made more difficult to analyse because the airports
themselves usually confront some degree of competition from other airports or,
especially in Europe, high-speed rail services.
In addition to this issue of pure market power, the air transport network is
generally more complex than a road system in a very specific way. While most road
movements are point-to-point (car drivers seldom change vehicles en route), most
large airlines operate hub-and-spoke systems where people make trips via a hub,
and changes of hardware for them are the norm. This means that carriers are con-
cerned not only about the immediate implications of flight delays but also about
the effects of this on connecting traffic and its own further network operations.
Further, whereas a car can wait almost indefinitely in traffic, an aircraft
cannot be airborne indefinitely, reducing the flexibility of infrastructure to handle
connecting traffic flows. Passengers would also be less than happy if an airborne
flight was diverted to an alternative airport because its congestion fees were lower
at that time.
Added to this is an issue examined initially by Eric Pels and Erik Verhoef
(2004), namely the possibility in an oligopolistic business environment for stra-
tegic interaction between competitors that leads to non-competitive pricing.
Basically, without congestion this will lead to prices in excess of the Pareto
optimum for the airlines that collude, with the policy implication that, in the
absence of any direct remedial action on the part of the authorities, there is a
second-best case for subsidies.
But perhaps the biggest difference is that the literature on road pricing gener-
ally makes simplifying assumptions of homogeneous traffic and identical types of
driver, which allows the MC curve also to be the cost curve related to the MC*
curve. This gives a determinant solution for the congestion charge. Put another
way, with MC also representing the congestion-embracing-average-cost curve,
MC1 is both the average revenue in equilibrium F1 and the average cost. The types
of assumption made regarding roads, however, seem less applicable to runways,
and the MC curve unlikely to be identical to the average-cost curve. We just list a
few of the complications.
First, as noted previously, it may not seem reasonable to assume the curves
are continuous; with 5,000 cars an hour on a road, continuous functions may
seem an acceptable assumption, but with 40 landings on a runway they are much
less justifiable. Second, many runways involve mixed flows of take-offs and land-
ings with different time and fuel requirements. Third, typically air traffic is highly
heterogeneous involving different sizes and types of aircraft. Fourth, because of
laws regarding flying hours, the crew costs may vary between flights if connect-
ing movements are missed. What these and other inconveniently realistic features
would typically mean is that there would be no stable AC and MC curves (that is,
adding a marginal flight may change the average cost for infra-marginal flights,

BUTTON 9781786435668 PRINT.indd 349 27/04/2022 11:07

 350 TRANSPORT ECONOMICS, 4TH EDITION

so the curve is not stable). In this case there is no way of estimating an optimal
congestion charge.
While in some ways more complex than roads, air transport infrastructure
is also simpler in dimensions that make some forms of congestion charging, and
especially auctioning, easier. It involves relatively small networks compared with
the complexity of most urban road systems and, despite some differences in
technical requirements, have a much more homogeneous set of users. A look at
the vast number of theoretical papers on urban networks seeking optimal traffic
flows, and ipso facto charges, all of which are based on a plethora of simplifying
assumptions, illustrates the complexities involved. Further, while this work no
doubt maximized the adrenalin flow of the authors that accompanies dancing an
intellectual jig, it would seem to have little impact on policy. The road prices intro-
duced in places such as Singapore and London involve nice round numbers, and
the speed targets on the Interstate 15 in California’s variable charging projects
have been equally robust rather than being founded in economic science.
Considerable intellectual effort has been expended on defining appropriate
pricing principles for airport slots, focusing mainly on the ways administrative
fees could be modified to make better use of them. The resultant output has
included looking at making greater use of pricing for cost recovery when there
is no congestion, but has largely been focused on situations where there are at
least some periods where demand under the current charging regime exceeds the
capacity of the system.
Moving to the empirical analysis of airport congestion, much of this has
involved looking at whether dominant airlines do internalize their costs. Daniel’s
pioneering work (Table 9.9) suggests that there is very little internalization by
the larger carriers, but his analysis suffers from being based on a single airport
and from focusing on just Northwest Airlines within a narrow cost-minimization
context. The focus on one airport may be a reasonable approach if it is the only
one subjected to serious congestion, but if there are other facilities in a similar
situation then a network analysis becomes germane. Additionally, the simulation
models deployed do not have a closed-form solution and thus it is not possible to
discern what drives the non-internalization.
Subsequent work, notably by Jan Brueckner (2002) and Mayer and Sinai
(2003), has looked across a range of airports to gain empirical breadth, and modi-
fied the models they used to meet the technical limitations of earlier formulations.
New data sources also allow for econometric analysis of causes for different
behavior patterns of airlines rather than just simulations. Using the number of
flights that are delayed for 15 minutes or more as the indicator of excess conges-
tion, Brueckner finds that delays do decrease where there is airline domination at
an airport but the degree of internalization is small.
Adopting the excess of flight time over the minimum feasible flight time as
their indicator of delay, Mayer and Sinai come to the same conclusion. Where
there is competition between airlines at an airport, then each should take a share
of the congestion costs imposed on others in proportion to their market share –
for example, American and United Airlines each has roughly half the traffic at

BUTTON 9781786435668 PRINT.indd 350 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 351

Table 9.9 
Econometric studies of internalization of congestion at airports

Study Case Model Findings

Carlin & Park LaGuardia Numerical calculation While full marginal cost pricing is not
(1970) Airport, New and regression possible, proportional marginal cost
York, 1967–68 pricing does offer some efficiency gains
Morrison US technical Numerical calculations Equal weighted Ramsey Pricing would
(1987) data on airline redistribute welfare from commuter and
costs for 1980 local service carriers to cargo,
international, and trunk carriers
Morrison & US delay data Looked at a road- Atomistic pricing would generate
Winston for May 1988 pricing-regime-type significant welfare benefits
(1989) model of atomistic
pricing
Daniel (1995) Minneapolis–St Bottleneck congestion Demand peaking means that congestion
Paul Airport in model and a time- pricing can generate large savings by
1990 homogeneous smoothing out demand
stochastic queuing
model
Martin-Cejas Uncongested Simulations looking at Allows for cost recovery
(1997) Spanish airports Ramsey Pricing
Daniel (2001) Minneapolis–St Stochastic bottleneck Congestion pricing would transfer
Paul Airport in model welfare back from private aircraft
1990 operators and their high-income
passengers to common travelers on
commercial aircraft
Daniel & Minneapolis–St i. Standard peak-load Similar traffic patterns are produced by
Pahwa Paul Airport in pricing model the models using weight-based fees, but a
(2000) 1990 ii. Deterministic variety of patterns emerge when
bottleneck model congestion pricing is used
iii. Stochastic
bottleneck model
Brueckner 25 most Econometric analysis When an airport is dominated by a
(2002b) congested of delays using six monopoly airline then the costs of
airports in the different specifications congestion are fully internalized
US in 1999
Mayer & Flights between Econometric analysis Hubbing is the primary cause of traffic
Sinai airports with at of delays using delays and this largely reflects optimal
(2003) least 1% of US regression with and use of runway capacity
flights, without fixed effects
1988–2000
Harback & 27 major US Deterministic The idea that dominant carriers
Daniel airports, 2003 bottleneck model of internalize self-imposed congestion costs
(2007) congestion is rejected and full congestion pricing is
found to be optimal
Morrison & 74 US airports Morison & Winston Optimal and atomistic congestion pricing
Winston for 2005 (1989) congestion generate welfare gains with the latter
(2008) measure being slightly greater

Source: Button (2008). This paper contains the full references to the studies cited.

BUTTON 9781786435668 PRINT.indd 351 27/04/2022 11:07

 352 TRANSPORT ECONOMICS, 4TH EDITION

Chicago O’Hare International Airport, and should thus each pay half of the costs
that it imposes.
This is not, however, the finding of Harback and Daniel (2007), who looked
at 27 large United States airports using structural models of landing and take-off
queues, rather than econometric techniques, for empirical verification. It also
differs from Brueckner’s and Mayer and Sinai’s findings in that an input measure
of delays (the delays generated by each aircraft) is used rather than an output
measure (delays experienced by each aircraft). The result, tested on data for 2003,
is that in most cases there is no internalization of congestion costs by dominant
airlines. The rationale seems to stem from Stackleberg oligopoly game theory,
although other gaming models may be relevant. Essentially, the dominant carrier
uses its ability to manipulate congestion to deter market entry by competitors,
a limit pricing, which leads the dominant firm to act as if it had the costs of its
rivals or potential rivals. If it were to internalize its congestion it would release
capacity that could be taken by a competitor, allowing the latter the potential of
reaping economies of scope and density, and of market presence.
What much of the work on airport congestion pricing does not completely
embrace is the issue of full cost recovery. In conditions of a competitive supply of
runway capacity and constant cost, the link is a simple one: appropriate charges
indicate where capacity is needed and provides the revenues to complete new
works. This quickly breaks down where supply is not competitive and there are
obvious technical non-linearities in the long-term cost function; runways are not
divisible. The empirical work on this is limited, but Morrison and Winston (2008)
indicate that only charging for congestion caused to other carriers à la Brueckner
would lead to $4.6 billion less revenue at United States airports than a regime
of atomistic congestion charges akin to road pricing. Cost recovery then poses a
range of additional issues in terms of the supplementary methods of finance, be
these subsidies of one kind or another, or Ramsey Pricing.
There are difficulties in the general approach of these studies, besides specific
limitations inherent in each. Here we focus on the data and definitional problems
in trying to determine congestion costs. Consideration of transaction costs and
information availability is, as Coase (1960) pointed out, important in determining
the best mechanisms for asset allocation. Much of the slot analysis work focuses
on the operating impacts of congestion on airlines’ costs and time costs for pas-
sengers, but airlines are the final purchasers of the infrastructure.
Airlines are rent-seekers and use a variety of practices to extract rent from
their customers – most notably second-degree price discrimination with frequent
flyer programs that offer effective discounts for multiple flights and a combina-
tion of second- and third-degree price discrimination through yield management
practices that involves posting a menu of fares, but where this menu changes
by user groups according to their preferences about the timing of bookings.
Because airlines take advance bookings, the immediate effects of congestion
charges on revenue would be minimal, but the longer-term implications may be
larger. In this context, even if an airline is profit-maximizing, it may prioritize
its actions by net revenue flow rather than cost savings in terms of the ways it

BUTTON 9781786435668 PRINT.indd 352 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 353

adjusts its flight sequencing. The empirical problem is that suitable information
on revenues per flight is not available, although this is needed to estimate the
congestion cost.
Added to this, as Brueckner (2005) points out in a theoretical contribution,
on the cost side there are indivisibilities in the provision of airline services that
make assessments of flight prioritization difficult, and especially so when there
are hub-and-spoke operations. There are problems of scheduling different sizes
of aircraft over various stage lengths and ensuring equipment compatibility as
well as fulfilling technical requirements regarding such things as periodic refu-
eling and maintenance inspections. In addition, crew changes need to conform to
statutory flying hour regulations and there are operation constraints such as gate
availability. Much more than in the road traffic congestion case, the internaliza-
tion process within this context is inevitably one of a constrained optimization
process and assessment of the degree to which this is done requires more complex
assumptions about relevant functions. Whilst these considerations may be built
into theoretical models, they are difficult to incorporate into empirical analysis
and, in turn, into estimation of actual charges.
As mentioned earlier, one difference between road congestion and airport
congestion is that the airline market is not atomistic and this difference offers an
alternative practical approach to handling congestion. An underlying problem
with congestion charging and its analysis is that it involves making external judg-
ments about the optimal congestion charge. While the approaches examined in
Table 9.9 are essentially about situations where there is excess demand under
current charging regimes, and the role a Pigouvian-style price may play in opti-
mizing this, they do not focus on how optimal slot charges may be implemented.
They do not look at arrangements that would automatically reveal these fees and
impose them on users. The alternative approach to setting congestion charges is
to create appropriate institutional structures that coincidentally reveal optimal
prices while allocating efficiently a given capacity.
One suggestion for significantly improving the slot allocation situation is
the adoption of some form of auctioning system (Button, 2007). This is a mirror
image of congestion charging where some expert sets a congestion charge and
users willing to pay this price purchase capacity. Instead, the expert sets a level
of capacity and then allows bidding for that capacity, generally in the form of
an auction. Auctions have become a subject of increasing interest as markets
have been deregulated and authorities have become interested in attracting more
private finance. Indeed, the idea of using auctions for allocating airport slots is
not new and the Nobel laureate Vernon Smith and his associates (Rassenti et al.,
1982) used the problem in their early work on experimental economics. Their use
for allocating air transport facilities is now attracting the interest of the European
Union and some national governments.
Strictly, the standard Coasian line of reasoning implies that the initial allo-
cation of slots, be it by lottery, auction, a handover to incumbents, or whatever,
should, if, subsequent secondary markets are permitted, make no difference to
the long-run efficiency with which they are used. Auctions are but one way of

BUTTON 9781786435668 PRINT.indd 353 27/04/2022 11:07

 354 TRANSPORT ECONOMICS, 4TH EDITION

allocating property rights. There is no need for an expert economist or engineer

to externally set congestion prices once the allocation has been made. There will
be a short-term transfer of rents between airports and carriers, and subsequently
between various carriers after secondary trading, but the outcome is that carriers
that can make the most efficient use of slots will be able to obtain the numbers
they require either through their bidding at an auction or through secondary
trading, and will buy and sell slots as markets change.
In practice, there are matters of political economy to consider as well as the
technical nature of subsequent secondary markets. The way that slots are allo-
cated is a concern; windfall gains accrue if incumbents are given property rights
that would not be enjoyed by those wishing to join the market. It is thus no acci-
dent that many legacy carriers find voice in supporting this give-away approach.
Lotteries spread the windfall rents randomly but there do arise issues about who
can participate and whether there are any payments to be made by winners. There
is also the matter of whether the lotteries should be for pairs of slots, combina-
tions of slots, or groups of slots. An extensive secondary market would be needed
because the initial allocation is likely to be far from optimal.
But even if auctions are adopted, there are issues akin to those inherent in
road pricing, including that of ensuring that costs of maintaining existing capac-
ity are recovered, and how to finance and prioritize capacity expansion. With
charges, this involves notions of maintenance cost allocations between users
and cost recovery that have to be embraced within the charging regime. While
auctions of runway slots are certainly not free from problems, they do partially
circumvent the need for excessive calculations.
The primary aim of a well-structured periodic auction of slots is to make
sure that the winner is the carrier that will generate the most profit from the use
of a slot or bundle of slots. Ideally, it is a neutral process whereby exogenous
information has no influence. In addition, an appropriately structured auction,
through the revenues it raises, both provides signals as to whether additional
capacity is needed and provides initial resources for capacity expansion. Indeed,
the collection of a bid from the auction transfers the value of a slot to the airport
and avoids the concerns with arbitrary initial allocations whereby airlines enjoy
the wide-fall gains. Incorporating secondary markets into the structure allows for
fine-tuning as markets evolve and to make technical adjustments for imperfec-
tions in the auction procedure.
Thus, as with congestion charging, auctioning of runway slots has its prob-
lems and it is perhaps helpful to discuss these in a comparative context. The
number and duration of slots that is seen as making efficient use of a runway have
to be determined and this, like a congestion charge, is an external judgment. But
it is very much akin to the real economic decisions made by industry every day:
coffee is normally sold in cafés in cups and there is the issue of defining exactly
what constitutes a ‘cup’. While modeling, by definition, abstracts from reality,
there are institutional constraints that implicitly shape both, including the defini-
tion of slots as well as ‘coffee cups’. Roads are designed to engineering standards
that affect, for example, the speeds at which one can drive and remain on the

BUTTON 9781786435668 PRINT.indd 354 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 355

surface, and this is in addition to institutional issues such as speed limits and
driving restrictions for safety and environmental reasons. These are largely taken
as axiomatic in the road pricing analysis.
Runways are similarly restricted with stall speeds limiting the number of
take-offs and landings per hour, as well as institutional requirements dictat-
ing headways and hours of operation. While in the road case these factors can
be incorporated in the assumptions needed for a continuous traffic flow (for
example, in the maximum engineering flow) without much disruption, they are
important for runways that only involve 60 or so movements an hour. The tech-
nical and institutional constraints that are part of airport operations essentially
determine the capacity in terms of very discrete units. This makes the modeling
of congestion pricing difficult.
If auctions are adopted, there are further issues to consider. One approach
is a once-and-for-all auction with rights transferred to the airlines and the other
is a concession arrangement with periodic auctions of slots but with airports
retaining property rights. The former has considerable theoretical attraction, in
that there is no requirement to externally determine the optimal frequency of auc-
tions for concessions, but it may well be preferable to avoid the ‘New York Cab
Medallion Problem’. The latter involves existing holders resisting any expansion
in capacity to protect their monopoly powers and highlights difficulties in increas-
ing and allocating new capacity.
In practice, slot delineations, again as with estimation of a road price, cannot
be treated as independent of wider considerations. In the case of road pricing there
is the problem of the allocation of capacity to non-vehicle users, most notably
pedestrians. Their pricing seems to be singularly ignored in the academic literature
but they, and cyclists in some countries, probably constitute the largest number of
movements on urban roads, and the greatest congestion. The parallel at airports is
non-aircraft movements on aprons. Added to this, there are matters of interactions
between traffic at junctions and how this should be handled. In the case of urban
traffic systems, light sequencing is partly designed to facilitate high flow but also to
take into account access needs of those on secondary roads and the overall safety
of road users. The use of complicated runway configurations and the need to use
runways for both take-off and landings are the aviation parallels.
Another difficulty often raised is the matter of how auction revenues should
be used; they effectively transfer economic rent from carriers to the airport
authority (be that a private or public agency). This is not a difficulty absent from
congestion charging in atomistic markets, and deciding what to do with road
pricing revenues has generated a small library of publications. If the airport is a
rent maximizer then the auction becomes a form of price discrimination with each
slot being sold at a price that reduces its value to the purchasing airline to that
yielding a normal profit. The airline holding a slot would use it efficiently, includ-
ing taking account of any internal congestion considerations. The airport would
use the revenue to make investments that offer the highest return irrespective of
the sector involved. Strictly, if an airport is congested then there are grounds for
using the revenue for capacity expansion and the return would justify this.

BUTTON 9781786435668 PRINT.indd 355 27/04/2022 11:07

 356 TRANSPORT ECONOMICS, 4TH EDITION

Finally, while the general literature, which is not considered here, on auctions
is extensive, no fully efficient and practical auction structure for allocating slots
emerges, just as there is no ideal form of congestion charging. Although it poses
some problems of its own, secondary trading in slots can provide a viable adjust-
ment mechanism to correct for the worst misallocations of a primary auction and
allow airlines to adjust their slot portfolios to meet take-off and landing require-
ments. But these also need structuring. As institutional economists frequently
point out, markets do not arise and function in a vacuum but rather operate with
a structure of formal laws and governances.
In practical terms, normally there is a place for trade to take place – in the
twenty-first century this can be an electronic exchange – and there is the need for
oversight of transactions, and methods of recording to ensure property rights
are transparent. These are not problematic and, indeed, exist in embryonic forms
in several airport slot markets already and seem to impose minimal transaction
costs. A more important concern is the possibility that the market will prove to be
imperfect, with monopoly power distorting its efficient functioning. There may be
serious concerns that participants will ‘bank’ certain slots and that other airlines
need to make efficient use of those slots they already have. The extent to which
this can occur often depends not only on the underlying nature of the airline
market and the details of the secondary slot market, but also on the generic
nature of competition laws in the country.

9.8 Seaports Congestion

Roads are not the only area where congestion pricing has been advocated.
Walters (1976), for example, argued 46 years ago that appropriate marginal cost
pricing (including congestion charges) is ‘no panacea for ailing or congested
ports, but it does supply a useful set of principles to deploy in the discussion
of port pricing policy’. Although the general economic principles for optimal
port pricing are identical to road pricing, in some circumstances the nature of
the shipping industry may result in complications. In the road context there is
a monopoly supplier adopting social pricing policies coupled with competition
for road space amongst many, uncoordinated potential users. While most ports
conform to this type of market situation, in some instances the port authorities
are confronted by a monopoly (or, more likely, a cartel) of shipping compa-
nies while in others there may be competition between hub ports for business
(Haralambides, 2002).
Figure 9.5 shows the demand curve for shipping (the demand for port ser-
vices may be seen as proportional to this) in terms of total import and export
traffic. The port is assumed to have constant marginal handling costs, 0H, which
are passed to the ship owners as port charges. The shipping companies, if com-
petitive, would then charge these customers an additional amount, RH, to reflect
their own average costs, to give a total shipping rate of 0R. The AC of shipping
will itself rise after a certain point as port congestion forces queuing to load

BUTTON 9781786435668 PRINT.indd 356 27/04/2022 11:07

 357
OPTIMIZING TRAFFIC CONGESTION ­

MC (port & ship)

AC (port & ship)
S
B

R
U

A
Demand

H
MC (port)
MR

0 T3 T2 T1 Tons of freight
Figure 9.5 Port congestion pricing

and discharge. Since the AC curve does not reflect the true costs of increasing
traffic the port authority should, on welfare economic grounds, levy a congestion
charge of AB. Assuming there is no potential for modifying the types of ships
in service or the methods of cargo handling this will reduce the tonnage passing
through the port from 0T1 to the optimal level 0T2. This is identical to the road
pricing case.
Suppose that instead of a competitive shipping market, the port was used
exclusively by a closed liner conference. There are now two important differences.
First, the conference, being the sole operator, will bear the costs of congestion
itself – the congestion costs are internalized. Second, the conference is likely to act
as a monopolist (although, as we have seen in the previous chapter, countervail-
ing powers act as a limited constraint in practice) and be more concerned with
the marginal revenue curve than with demand. Thus, the ship owner will charge
customers a rate of 0S for their services which comprise: port fees, 0H; their own
costs, including that of congestion, HU; and economic, monopoly, rent, US. The
tonnage passing through the port is now suboptimally small at 0T3.
Although it may be argued that in this situation the optimal use of the port
could in some situations be achieved by not charging a congestion toll and by
reducing port fees below 0H, this rather evades the real problem, namely the
monopoly power of the shipping conference. Such a policy also places excessive
power in the hands of the conference when negotiating with port authorities the
fees (and, ipso facto, the subsidy) to be charged. The solution here is to tackle
distortions at source, namely in the shipping market, rather than maladjusted
port prices.
There are occasions when ports do add a surcharge for exceptional con-
gestion. This, for example, occurred in many British, Chinese, and American
container ports after the outbreak of the Covid-19 pandemic, although these
surcharges have also been a reflection of higher operating costs due to labor

BUTTON 9781786435668 PRINT.indd 357 27/04/2022 11:07

 358 TRANSPORT ECONOMICS, 4TH EDITION

shortages. These and other costs are passed through to shipping rates and affect
the overall costs in the supply chain. Shippers have thus prioritized their consign-
ments in the short term.

9.9 Non-pricing Options for Reducing Congestion

The traditional approach to handling traffic congestion, be it on the roads or

elsewhere, in the absence of appropriate economic pricing, has been to increase
capacity (which may entail civil engineering works or softer options such as the
use of informatics) or to try to attract passengers or freight to less congested
facilities such as public transport or the railways. These approaches have largely
been advocated because they do not alienate road users, who, in terms of coalition
theory, form a powerful lobbying group, and partly because of the capture of the
system by engineers and others who have a vested interest in infrastructure invest-
ments. We look at the implications of some of these alternatives.

Invest in More Capacity

The oft-called ‘engineering approach’ to traffic congestion that tended to domi-

nate thinking in the post-Second World War period was to build more capacity
to meet demand; basically to ‘predict and provide’. But, as was seen in Chapter 8
and earlier in this chapter, because there is congestion does not necessarily mean
that there is a shortage of capacity; it is often the result of inappropriate use and
incorrect pricing. Adding capacity in these circumstances can be wasteful and
seldom reduces the congestion problem in the long run.
Figure 9.6 replicates the standard congestion diagram (essentially Figure 9.1)
with a suboptimal traffic flow of F1. An optimal road price would bring traffic
levels back to the flow where MC1 is equated to Demand (that is, at point B). In
efficiency terms this means that without an appropriate charge, the traffic flow F1
results in a social, or ‘dead-weight’ loss equal to the area ABC.
Increasing capacity affects the cost curves in the diagram; it pushes them
down and to the right because at any flow the generalized cost is lower. The
extra capacity need not come from physical investment but may be the result of
introducing improved traffic control or information systems. If the additional
investment pushes the marginal and average costs down to AC2 and MC2 then
the traffic flow will increase to F2. At this flow, however, there is still excessive
congestion resulting in a dead-weight loss of abc, quite simply because the
optimal use of the road after its expansion or the introduction of some enhanced
informatics would be where MC2 is equated with demand (that is, at point b in
the figure).
If the demand curve is highly elastic, it is clear from the diagram that conges-
tion costs would be little reduced by pulling the cost curves down. This situation
is often couched in terms of a sort of Parkinson’s Law of traffic at work: the so-
called the Pigou–Knight–Downs paradox, or simply Downs’ Law. Basically, this

BUTTON 9781786435668 PRINT.indd 358 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­359

$
MC1
Dead-weight loss before
road investment AC1
Demand
MC2

A AC2
a
B
P1 b
C Dead-weight loss after
P2 c road investment

0 F1 F2 Flow
Figure 9.6 The effects of expanding capacity on road congestion

says that road traffic expands to fill the road space available. And there is empiri-
cal support for this.
One way of looking at the effects of infrastructure expansion on traffic con-
gestion is to examine the effects of national construction programs. In an econo-
metric study of highway spending and the cost of traffic congestion (as measured
by the Texas Transportation Institute) in 74 urban areas in the United States
covering 1982 to 1996, Winston and Langer (2006) found that, ‘on average, one
dollar of highway spending in a given year reduced the congestion costs to road
users only eleven cents in that year’. Part of this may, of course, be because the
investments were not optimal, but nevertheless the congestion-reducing effects do
not appear large. In another study examining United States urban data from 1983
to 2003, Duranton and Turner (2011) found support for the fundamental law in
highway building, the extension of the Interstate highway system being met with
a proportional increase in traffic.
While aggregate expansion of capacity may not have much impact on con-
gestion levels, investment in specific categories of highway may. Mixing traffic
(buses, cars, trucks, bicycles, etc.) can worsen traffic flows because of their dif-
fering engineering features, such as speed and acceleration, and compound the
problems of vehicle interaction. One policy approach is to separate traffic types
and to construct mode-specific infrastructure. Truck-only lanes, for example, are
under study as tools to combat road congestion, enhance safety, and reduce other
external costs of road traffic.
The benefits of separating cars and trucks depend on several factors: the
relative volumes of cars and trucks, the congestion delay and safety hazards that
each vehicle type imposes, the values of travel time for each type, and the lane
capacity indivisibilities. The optimal assignment of heavy vehicles to truck lanes,

BUTTON 9781786435668 PRINT.indd 359 27/04/2022 11:07

 360 TRANSPORT ECONOMICS, 4TH EDITION

however, would still need the support of an allocative mechanism to optimize

traffic flows on those lanes. In this context André de Palma et al. (2008) finds that
priced lane for one vehicle type is generally more effective than hard restrictions
on the number of entrants. The benefits of all forms of intervention, though, are
sensitive to whether the proportions of cars and trucks are commensurate with
lane capacities.
But even when it was appreciated that unrestrained use of infrastructure
means its inefficient use, decision-makers were still reluctant to adopt economic
pricing, their focus often continuing on maximizing throughput rather than opti-
mizing its use.
One of the reasons that capacity expansion has been a way of optimizing
congestion is that an implicit assumption of many traffic forecasting models is
that the demand curve for road use is perfectly elastic, whereas the evidence from
the road pricing schemes that have been initiated suggest that this is far from
being the case. We shall return to this issue when discussing the economic aspects
of transport forecasting modeling.
Improved traffic management, including the adoption of electronic ITSs, has
the same effect as building more capacity. It allows more vehicles to use a road,
but, without any system of limiting entry, simply leads to additional traffic bring-
ing the flow back to the initial level. As an example, the perennially congested
south-western stretch of the M25 ring road around London has been fitted with
an experimental automated traffic control system called Motorway Incident
Detection and Automatic Signalling (MIDAS). This consists of a distributed
network of traffic and weather sensors, speed cameras, and variable-speed signs
that control traffic speeds with little human supervision, but has done little to
alleviate traffic problems.

Subsidize Fewer Congestion-causing Modes

Some modes of transport can technically make more efficient use of infrastruc-
ture than others, and thus there are arguments advanced for stimulating their use
rather than congestion-inducing modes such as the motorcar. In the road context,
the most common approach is to subsidize public transport modes. Buses, trams,
and metro systems can in most circumstances when fully loaded move more
people than the motorcar for a given land-take. In terms of the standard conges-
tion diagram, subsidizing, say, a bus service shifts the demand for car use down
and to the left, as seen in Figure 9.7. The welfare implications are that, whereas
without any congestion initiatives there is a loss of ABC, by attracting some
travelers from the car to a bus the loss declines to abc. Congestion costs remain,
because the lower generalized costs of road use will cause some people who did
not travel before to make trips, but they are normally reduced.
This type of policy has attractions, especially in circumstances when the sub-
sidies also benefit low-income public transport services. There are, however, often
difficulties. For the subsidies to stimulate a significant traffic effect, there must be
a relatively high-fare cross-elasticity of demand. As we have seen in Chapter 4,

BUTTON 9781786435668 PRINT.indd 360 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­361

MC
Demand1
Dead-weight loss
after transit subsidy AC

Demand2 A

B Dead-weight loss
before transit
P1 C subsidy
b a
P2
c

0 F2 F1 Flow
Figure 9.7 The effects on road traffic of subsidizing rapid transit

the cross-elasticities between modes often tend to be relatively low. The general
evidence suggests that lowering fares often has little impact when it comes to
attracting traffic to buses; at the extreme one can witness the reluctance of school-
children to use free and convenient ‘yellow buses’ once they can drive a car. There
is some limited support for the position that improved service (speed, frequency,
and reliability), and especially in terms of rapid transit systems, can affect modal
transfers to some extent, but this is an expensive option.
The second issue is that there are generally dead-weight losses associated
with subsidies that we touched upon in Chapter 7, and just highlight again in a
slightly different way.
First, a subsidy will result in a simple loss in terms of waste in the public
transport sector. In Figure 9.8 it is assumed that the public transit undertaking,
which for simplicity we assume has its own track, is run effectively and is eco-
nomically priced. Further, and just for ease of exposition, it is assumed that the
marginal cost of each public transport passenger is constant, thus giving an initial
fare of F1. A subsidy is then introduced to reduce fares to F2 to attract travelers
from their automobiles. The cost of that subsidy is the shaded area F1F2bc: the
subsidy times the number of passengers that ultimately use the public transport.
Part of the subsidy goes to the Pas passengers who are already riders, and the
1
rest to those who give up their cars. Those already on public transport enjoy a
transfer of money from tax-payers; this has no resource effects if prices elsewhere
in the economy are optimal, although there are differential effects on bus-riders
and tax-payers.
The amount that is needed to stimulate the Pas – Pas transfer to public
1 2
transport entails a degree of waste. In this case it is equal to an amount abc in
the figure. This is a loss because while the Pas th transfer needs the full subsidy
1

BUTTON 9781786435668 PRINT.indd 361 27/04/2022 11:07

 362 TRANSPORT ECONOMICS, 4TH EDITION

Dead-weight loss of
$ Demand
transit subsidy

a b
F1
Subsidy
per
passenger c
F2

0 Pas1 Pas2 Passengers

Figure 9.8 The dead-weight losses of subsidies

to cause the transfer, the Pas th person needs very little. But, in addition to this,
2
there is also the potential for significant X-inefficiency in terms of the ways that
subsidies are used. Much depends, as we saw in the previous chapter, on the way
in which subsidies are awarded to public transport suppliers.

9.10 ‘Micromobility’

The notion of ‘micromobility’ is far from new. Walking and horse-riding are, of
course, the traditional forms, but more recently the term refers to small, light-
weight vehicles operating at speeds typically below 15 mph and driven by users. It
includes the likes of bicycles, push-scooters, e-bicycles, e-scooters, e-skateboards,
Segways, and electric pedal-assisted bicycles. While favored because of many
positive environmental attributes, such as being low noise generators and fuel
efficient, they are also seen as a way of reducing local traffic congestion. They not
only take up little space while in motion but also when they are parked. Added
to this, many of them are amenable to ride-sharing, and especially so with the
development of GPS systems for locating them, smartphones for booking them,
and electronic payment allowing the use of dockless technologies.
As a public transport mode, some forms of micromobility, such as Segways,
were limited in their uptake probably because they require some practice to ride
and involve high investment costs. But the short-term rental of bicycles, initially
in Copenhagen in 1995 but then in Washington DC and other United States
cities, became popular. By 2016 there were 42,000 shared bicycles across the
United States. A key development occurred in 2013 when dockless bicycle-­sharing
emerged. Low initial and operating expenses, combined with local authority
financial support in some cities, made the mode convenient and economical for
customers. The arrival of the e-scooter soon followed, superseding the dockless

BUTTON 9781786435668 PRINT.indd 362 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 363

bicycle in many places. Bird, a company based in Santa Monica, California, first
deployed e-scooters in 2017. Within a year, the company expanded to over 100
United States cities and 11 international cities, and supplied over 10 million rides.
Spin, based in San Francisco, started as a bicycle-sharing business and began
to switch to e-scooters when the Ford Motor Company purchased it for nearly
$100 million in 2018. Lime, which also started as a bicycle-sharing business,
expanded into e-scooters in 2018 and by early 2020 was operating in over 100
cities in the United States, along with 27 international cities, accumulating over
11.5 million rides. The arrival of Covid-19 in 2020 stymied the rapid growth of
the mode for at least a period.
E-scooters often meet a demand that current services do not, or only do so
in a second-best way (Button et al., 2020). They may, therefore, not be a good
substitute for congestion-causing automobiles or conventional forms of public
transit. Their use may be as an alternative to walking or using another micro­
mobility mode. Scooter rides are, for example, typically under two miles, which is
often a distance too short for hailing a taxi or a transportation network company
ride with, say, Uber. Robert Noland (2019), studying journey purposes of about
80,000 Bird and Lime e-scooter trips in Louisville, Kentucky, where 82 percent
of commuters drive alone and 9 percent carpool, found that the average trip
duration by scooter between August 9, 2018 and February 28, 2019 was 15.59
minutes and distance was 1.33 miles. Similar average distances were recorded in
Indianapolis.
E-scooters, and most other micromobility modes, also generate a variety of
safety and other issues. Their use on side-walks poses problems for pedestrians
by creating congestion and endangering them. Nikan Namiri et al. (2020), for
example, found that between 2014 and 2018 more than 39,000 e-scooter injuries
occurred in the United States, with the annual number involving millennials
increasing from 582 to 5,309 over the period. One-quarter of the injuries involved
a broken bone, and one-third were to the head, double the rate of bicyclists’ inju-
ries. There is also a question mark over the financial viability of e-scooter com-
panies. They operate in a very competitive market with free entry and exit, and
few made a profit prior to the Covid-19 disruptions. These are common features
of a lack of an economic core of the type discussed in Chapter 7. The resultant
market instability reduces the likelihood that commuters and others will adopt
e-scooter-riding on a large scale.

References

Albert, G. and Mahalel, D. (2006) Congestion tolls and parking fees: a comparison of the
potential effect on travel behavior, Transport Policy, 13, 496–502.
Arnott, R., de Palma, A., and Lindsey, R. (1993) A structural model of peak-period
congestion: a traffic bottleneck with elastic demand, American Economic Review, 83,
161–79.
Börjesson, M., Eliasson, J., and Hamilton, C.J. (2016) Why experience changes attitudes to
congestion pricing: the case of Gothenburg, Transportation Research Part A, 85, 1–16.

BUTTON 9781786435668 PRINT.indd 363 27/04/2022 11:07

 364 TRANSPORT ECONOMICS, 4TH EDITION

Börjesson, M., Eliasson, J., Hugosson, M.B., and Brundell-Freij, K. (2012) The Stockholm
congestion charges – 5 years on: effects, acceptability and lessons learnt, Transport
Policy, 20, 1–12.
Brueckner, J.K. (2002) Airport congestion when carriers have market power, American
Economic Review, 92, 1357–75.
Brueckner, J.K. (2005) Internalization of airport congestion: a network analysis,
International Journal of Industrial Organization, 23, 599–614.
Button, K.J. (1984) Road pricing: an outsider’s view of American experiences, Transport
Reviews, 4, 73–98.
Button, K.J. (2006a) How should the revenues from congestion pricing be spent? in
G. Roth (ed.), Street Smart: Competition, Entrepreneurship, and the Future of Roads,
The Independent Institute.
Button, K.J. (2006b) The political economy of parking charges in ‘first’ and ‘second-best’
worlds, Transport Policy, 13, 470–78.
Button, K.J. (2007) Auctions: what can we learn from auction theory for slot allocation? in
A. Czerny, P. Forsyth, D. Gillen, and H.M. Niemeier (eds), How to Make Slot Markets
Work, Ashgate.
Button, K.J. (2008) Issues in airport runway capacity charging and allocation, Journal of
Transport Economics and Policy, 42, 563–85.
Button, K.J. (2020) The transition from Pigou’s ideas on road pricing to their application,
Journal of the History of Economic Thought, 42, 417–38.
Button, K.J. and Vega, H. (2007) The costs of setting up and operating electronic road
pricing in cities, Traffic Engineering and Control, 48, 6–10.
Button, K.J., Frye, H.F., and Reaves, D. (2020) Economic regulation and e-scooter net-
works in the USA, Research in Transportation Economics, 84, No 100973.
Coase, R.H. (1960) The problem of social cost, Journal of Law and Economics, 3, 1–44.
De Palma, A., Kilani, M., and Lindsey, R. (2008) The merits of separating cars and trucks,
Journal of Urban Economics, 64, 340–61.
Duranton, G. and Turner, M.A. (2011) The fundamental law of road congestion: evidence
from US cities, American Economic Review, 101, 2616–52.
Goodwin, P.B. (1989) The rule of three: a possible solution to the political problem of
competing objectives for road pricing, Traffic Engineering and Control, 29, 495–7.
Graham, D.J. and Glaister, S. (2006) Spatial implications of transport pricing, Journal of
Transport Economics and Policy, 40, 173–201.
Gu, Z., Liu, Z., Qixiu, C., and Saberi, M. (2018) Congestion pricing practices and public
acceptance: a review of evidence, Case Studies on Transport Policy, 6, 94–101.
Haralambides, H.E. (2002) Competition, excess capacity, and the pricing of port infra-
structure, International Journal of Maritime Economics, 4, 323–47.
Harback, K.T. and Daniel, J.J. (2007) (When) Do hub airlines internalize their self-
imposed congestion delays? Journal of Urban Economics, 63, 613–30.
Hensher, D.A. and King, J. (2001) Parking demand and responsiveness to supply, price
and location in Sydney central business district, Transportation Research Part A, 35,
177–96.
Jones, P. (1998) Urban road pricing: public acceptability and barriers to implementa-
tion, in K.J. Button and E.T. Verhoef (eds), Road Pricing, Traffic Congestion and the
Environment: Issues of Efficiency and Social Feasibility, Edward Elgar Publishing.
Leape, J. (2006) The London congestion charge, Journal of Economic Perspectives, 20,
157–76.
Lehe, L. (2019) Downtown congestion pricing in practice, Transportation Research Part C,
100, 200–23.
Mayer, C. and Sinai, T. (2003) Network effects, congestion externalities, and air traffic
delays: or why not all delays are evil, American Economic Review, 93, 1194–215.
Morrison, S.A. (1986) A survey of road pricing, Transportation Research, 20A, 87–97.

BUTTON 9781786435668 PRINT.indd 364 27/04/2022 11:07

 OPTIMIZING TRAFFIC CONGESTION ­ 365

Morrison, S.A. and Winston, C. (2008) Another look at airport congestion pricing,
American Economic Review, 97, 1970–77.
Namiri, N.K., Lui, H., and Tangney, T. (2020) Electric scooter injuries and hospital admis-
sions in the United States, 2014–2018, Journal of the American Medical Association
Surgery, January 8, doi:10.1001/jamasurg.2019.5423.
Noland, R.B. (2019) Trip patterns and revenue of shared e-scooters in Louisville,
Kentucky, Transport Findings, April, https://doi.org/10.32866/7747.
Pels, E. and Verhoef, E.T. (2004) The economics of airport congestion pricing, Journal of
Urban Economics, 55, 257–77.
Quddus, M.A., Carmel, A., and Bell, M.G.H. (2007) The impact of the congestion charge
on retail: the London experience, Journal of Transport Economics and Policy, 41,
113–34.
Rassenti, S.J., Smith, V., and Bulfin, R. (1982) A combinational auction mechanism for
airport time slot allocation, Bell Journal of Economics, 13, 369–84.
Sharp, C.H. (1966) Congestion and welfare, an examination of the case for a congestion
tax, Economic Journal, 76, 806–17.
Shoup, D. (2005) The High Costs of Parking, American Planning Association.
Small, K.A. (1992) Using revenue from congestion pricing, Transportation, 19, 359–81.
Sorensen, P.A. and Taylor, B.D. (2006) Innovations in road finance: examining the growth
in electronic tolling, Public Works Management and Policy, 11, 110–25.
Verhoef, E.T. (2002) Second-best congestion pricing in general static transportation net-
works with elastic demands, Regional Science and Urban Economics, 32, 281–310.
Verhoef, E.T., Nijkamp, P., and Rietveld, P. (1995) The economics of regulatory parking
policies, Transportation Research Part A, 29, 141–56.
Verhoef, E.T., Nijkamp, P., and Rietveld, P. (1997) The social feasibility of road pricing:
the case study of the Randstad Area, Journal of Transport Economics and Policy, 31,
255–67.
Walters, A.A. (1976) Marginal cost pricing in ports, Logistics and Transportation Review,
12, 99–105.
Winston, C. and Langer, A. (2006) The effect of government highway spending on road
users’ congestion costs, Journal of Urban Economics, 60, 463–83.
Young, W. (2001) Parking and traffic control, in K.J. Button and D.A. Hensher (eds),
Handbook of Transport Systems and Traffic Control, Pergamon.

BUTTON 9781786435668 PRINT.indd 365 27/04/2022 11:07

 10 Economics and Transport Logistics

10.1 Introduction

The novel coronavirus (Covid-19) pandemic of the early 2020s highlighted the
importance of good transport logistics. Globally, the need to move medicines,
equipment, and sometimes patients put strains on the conventional supply
chains in the medical field. Official lockdowns and quarantines, in addition to
the sickness itself, affected workers and closed down many industrial production
nodes, as well as removing numerous links in national transport networks (Guan
et al., 2020). Tackling this required a rapid reorientation of complex supply lines.
This, however, occurred after a period of considerable advances in the way that
transport logistics was viewed and applied, and this aided the flexibility of its
response.
Rapid changes in computer technology and communications over the past
half-century and the arrival of ‘big data’, together with more flexible and rapid
transport, has inevitably produced a more complicated market for transport ser-
vices. Overlapping this have been developments in production technology and the
emergence of a sophisticated financial system. And spatially coinciding with this
has been the creation of global markets. To take advantage of these developments
there has been considerable integration of various supply chains, with, at the
same time, greater diversity in the nature of these chains. Economic factors have
been instrumental in the forms by which transport has fitted into these chains
while economists have been heavily engaged in analysing them.
Economists have long recognized that transport services are not provided
in isolation. The very fact that they appreciate that transport providers are con-
fronted by a derived demand curve makes this explicit. It has also been made clear
in our earlier discussion of the value chain concept. For a variety of reasons, not
least of which were the technical ability to do it and a rapidly growing demand
for international trade facilitation, the larger notion of transport logistics began
to be developed in the 1990s. This is effectively a more holistic way of looking at
transportation, in effect redefining its boundaries to move outside of the simple
movement of goods or people from A to B.
Optimizing supply chains is very much a matter of maximizing the efficiency
of resource allocations and considering the best way of using scarce resources. It
is about opportunity costs and thus it is about economics. Many of the findings
of the work that has been done in the supply-chain field, however, have tended to
appear in the management literature, or specialized logistics publications, rather
than in mainstream economics journals. This is not the place to consider why this
is so, although there has certainly been an element of the practical leading the

BUTTON 9781786435668 PRINT.indd 366 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 367

theoretical in many of the developments that have taken place in supply-chain

management, and much of the recent economic literature, supply modeling, and
analysis has tended to be rather abstract in its orientation (Friesz and Kwon,
2007). Here we focus on the main areas where there are differences or refinements
in the way that economics is applied, compared to that which is often covered in
the more conventional transport economics texts.
The chapter is largely about the private provision of logistics services, but
this provision is always within institutional and policy frameworks. The latter are
largely taken as given, and only discussed when of immediate relevance to the eco-
nomics of an issue. But they are relevant and one reason why modern transport
logistics has grown in importance in business is that there have been significant
changes in the legal constructs under which transport services, and trade more
generally, can be delivered.
In addition, the focus is largely on the private sector businesses, although
there are also issues involving the provision and management of public sector
transport infrastructure. For example, the scale and quality of publicly owned
transport road and rail networks, airports, and seaports affect such things as
warehouse distribution, trans-shipments, and the extent to which just-in-time
management of the supply chain is possible. This also raises economic issues
of a different kind, namely how to appraise public infrastructure investments.
The conventional cost–benefit procedures that we consider in Chapter 11 would
appear inadequate to the extent that they do not fully encapsulate the benefits
throughout the supply chain. This is not a topic dealt with in any depth here, but
really reflects the limited partial equilibrium nature of the economic cost–benefit
approach.

10.2 Transport Logistics

Transport logistics has a long history. In many ways, it is the adoption of trans-
port economic principles within a larger supply-chain concept. Definitions of
what exactly transport logistics entails are not always precise, but the general
idea is that it involves relationships between transport and integrated approaches
to logistics and supply-chain management. In other words, it involves the
­movement of goods through the supply chain, but is more than just the freight
transport aspect and embraces the full commercial and operational frame-
works within which the movement of goods is planned, managed, and finally
carried out.
The modern, large-scale academic study of civilian transport logistics started
around the early 1990s, when many of the current approaches were originally
developed, but there was both an implicit interest in it which went back centuries,
and a more systematic study of military logistics from classical times.
Logistics has always been important for the military, and transporting men
and equipment to the battle-front in a timely and coordinated fashion has since
the time of Alexander the Great given armies a decided advantage. The advent

BUTTON 9781786435668 PRINT.indd 367 27/04/2022 11:07

 368 TRANSPORT ECONOMICS, 4TH EDITION

of mechanized transport and of massive, often conscripted armies has made the
task more complex but possibly also more important. One reason for the failure
of the German army to capture Paris in 1914, for example, was the inadequacy
of transport logistics support; in effect it was impossible to move the number of
men and their supplies through northern France to conform with the needs of the
Schlieffen Plan. More recently, the overwhelming military success of Operation
Desert Storm in 1990 was in part due to the transport logistics expertise of US
General William (Gus) Pagonis.
The crucial role of logistics in the civilian sector has been recognized by
the appointment of logistics experts to the boards of many major corporations,
positions formerly largely reserved for financial, marketing, and production
executives. There are also many companies that specialize in transport logistics,
in addition to the in-house divisions of companies, and in many instances this
involves international activities as the ‘global economy’ grows and the supply
chains grow longer and more complex. The educational establishment has also
not been slow to recognize the modern role of logistics and many of the prestige
business schools around the world offer options in logistics as elements of their
MBA programs.
This change has come about for a variety of reasons; some are economic in
their nature while others stem from the emergence of new technologies. Before
looking at some of these in a little more detail, two over-riding trends have been
important in shaping the form of modern logistics. First, the length of freight
hauls has increased due to wider outsourcing and new approaches to centraliza-
tion, inventory holding, and terminal capacity. This means that production is now
more transport-intensive. Second, there has been a ‘time compression’ in produc-
tion and distribution that has put pressures on the system for more rapid delivery
of intermediate and final goods. This, in turn, has reduced inventory holdings,
making the production chain more vulnerable to disruptions in the transport
system.
At the more meso level, in terms of technology, the advent of the container
in the 1960s, and the fact that Malcom McLean did not seek any patent on his
invention, revolutionized the way that movement of semi-bulk cargoes could
be conducted, reducing the costs of movement almost immediately. In 1983
United States foreign ocean-borne commerce was 694.4 million metric tons;
ten years later, this had increased to 884.4 million metric tons. Ten years after
that, it had magnified to 1,167.9 million metric tons, nearly doubling in just
20 years. Growth of this magnitude is only physically possible with the use of
containers.
The container not only cuts the costs of individual movements by a single
mode of transport, but speeds up and cuts the cost of trans-shipments and con-
solidations between modes, making inter-modal transport a viable form of freight
movement. Inter-modal freight transport is movement of cargo in a container
using multiple modes of transport (rail, ship, and truck) without any handling
of the freight itself when changing modes. It also allows for easier handling of
mixed types of cargo on a single mode: a container can have its own refrigeration

BUTTON 9781786435668 PRINT.indd 368 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 369

unit or can be a ‘tank’ within a container-sized frame. The standardization of

the unit being carried allows for economies of scale in the cost of manufacturing
ships, wagons, etc. A container is also relatively easy to track, and can be sealed
to enhance security.
The types of goods being carried have also changed, in part because of the
availability of container transport and in-vehicle storage facilities (for example,
refrigeration). There are more high-value, less-bulking commodities being moved,
and their origins and destinations are more dispersed than with most primary
products. Their value means that holding large inventories can be costly and that
there are often economies in minimizing warehousing and storage. Inventory
management requires reliable delivery to ensure that goods and components are
available at the right time and undamaged. ‘Just-in-time production’ requires
complex logistics, including information systems, to ensure the generalized cost
of the supply chain is minimized.
Supply chains vary in their form and scale depending on factors such as
whether they are domestic or international, involve perishable goods, are mainly
bulk commodities, the costs of inventory holding, and whether special services
are required, such as refrigeration (Meixell and Gargeya, 2005). They can also
vary by direction, for example involving the movement of bulk commodities in
one direction and exotics, such as agricultural products, in the other. They are also
frequently multi-modal, often involving collection and delivery by one mode but
trunk haul by another. In some circumstances there may be considerable inter-
action between the freight supply chain and passenger transport; for example,
trucks and cars share road space, and many ‘passenger’ aircraft carry freight in
their belly holds.
In terms of economics, the logistics supply chain may be thought of as a set
of interdependent markets, or in some cases a single market, for transportation
services. Figure 10.1 provides a very simple illustration of the air cargo supply
chain.
In depicting this, we assume that air is the chosen mode of transport, which
will itself be an economic decision; in some circumstances road or rail may be
an option, and in a few cases, maritime transport. The shippers may either send
goods directly through their own regional distribution centers, may engage an
integrated carrier, such as UPS or FedEx, that offers door-to-door multi-modal
services, or may go to a freight forwarder that basically offers wholesale services.
The decision on which option to take is economic and involves considering the
generalized costs of the alternatives vis-à-vis their respective services. This is a
market decision. But there are also submarkets in the chain. While the shipper
may select a specific airfreight forwarder, that forwarder will obtain capacity from
air cargo carriers by bidding for space.
Looked at from the supply side, the various air cargo chains that co-exist
reflect the different business models that exist which in turn can be explained
in economic terms. Ronald Coase (1937) provides the classic economic reason
for different types of industrial structure, and why firms exist. The main reason
stems from the existence of transaction costs. He argues that a firm’s interactions

BUTTON 9781786435668 PRINT.indd 369 27/04/2022 11:07

 370 TRANSPORT ECONOMICS, 4TH EDITION

Shippers

Air transportation, Door-to-door logistics

logistics, and customs Regional and customs services
distribution
Additional revenue
through air carrier
functions

Freight forwarders Integrators

Express services through

integrators
Hub networks and
Warehousing
terminal stations

Airport logistics centers

Air cargo terminals

Air carriers Inter-line cargo

Source: Derived from Kasarda et al. (2004).

Figure 10.1 The supply chain for air cargo

within a market may not be under its control, but the firm’s internal allocation
of resources are. Consequently, the integrated suppliers that manage the chain
internally cater mainly for small, high-value goods that require rapid and reliable
service, whereas the use of forwarders reflects the need for less structured but
cheaper services that accompany more general cargoes.

10.3 The Costs of Warehousing and Inventory Holdings

Savings in freight travel time are an important, although not always the most
crucial, indicator of improvements in the efficiency of freight transport. The
methods of valuing such savings for use in the economic analysis of firms, or for
forecasting aggregate traffic flows in transport infrastructure planning, are akin
to those that are used in valuing passengers’ travel time savings (see Chapter 4).
The savings, however, can be related to several units –for example, to truck trips,
ship sailings, or rail wagons – in addition to the actual item being transported.
Table 10.1 offers a survey of the sorts of results that have emerged across a range

BUTTON 9781786435668 PRINT.indd 370 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 371

Table 10.1 Value of time in goods transport by rail, inland waterways, trucking, and air transport

Publication Country Data Method Value of travel time

savings

Rail transport € per transport hour

Inregia (2001) Sweden SP Logit 0 (shipment)
De Jong et al. (2004) Netherlands SP Logit 918 (train)

Air transport
Inregia (2001) Sweden SP Logit 13 (shipment)
De Jong et al. (2004) Netherlands SP Logit 7935 (full carrier)

Trucking
Small et al. (1999) United States SP Logit 174–267
Bergkvist (2001) Sweden SP Logit+WML 3–47
De Jong et al. (2001) France SP+RP Logit 5–11
Fowkes et al. (2001) United Kingdom SP Logit 40
Inregia (2001) Sweden SP Logit 0–32
Kurri (2000) Finland SP Logit 1.53
De Jong et al. (2004) Netherlands SP Logit 4.74

Rail transport
Vieira (1992) United States SP/RP Ordered Logit 0.65
Kurri et al. (2000) Finland SP Logit 0.09
De Jong et al. (2000) Netherlands SP Logit 0.96

Inland waterways
 Blauwens & Van de Voorde (1998) Belgium RP Logit 0.09
De Jong et al. (2004) Netherlands SP Logit 0.05

Trucking
Kurri et al. (2000) Finland SP Logit 1.53
De Jong et al. (2004) Netherlands SP Logit 4.74
Fowkes et al. (2004) United Kingdom SP Logit 0.08–1.18

Notes: SP = stated preference models; RP = revealed preference models.

Source: Adapted from De Jong (2008). This chapter contains the full references to the studies cited.

of studies covering inland waterways, air transport, railroads, and trucking,

taking account of the country to which the study relates, the type of data used,
and the methodology adopted.
There are wide variations in the results, and drawing general conclusions is
not simple. Much depends on what is being carried: saving a few hours for a ton
of coal may not be worth a great deal, but delays in supplying a crucial compo-
nent for a damaged piece of equipment may be very costly. Modes of transport
can also be important because of their particular security and safety features as
well as their position in the supply chain. We also note that most of the studies
that have been undertaken have used stated-preference techniques and these
would in general seem to generate relatively high values for time savings than the
small number of revealed-preference studies cited.

BUTTON 9781786435668 PRINT.indd 371 27/04/2022 11:07

 372 TRANSPORT ECONOMICS, 4TH EDITION

Freight is not always on the move in the logistics supply chain, but is periodi-
cally held in storage. The economics of warehousing is essentially about efficient
inventory holding. The role of a warehouse is that of storage. Referring again to
Figure 1.2 in Chapter 1, warehouses can also be seen as providing for consoli-
dation and points between elements in the value chain, but we leave discussion
of that role until the next section. Warehousing, within the narrow, traditional
context, serves several functions (Ackerman and Brewer, 2001):

• Accumulations of raw materials for distribution further down the supply

chain.
• Storage of in-process inventory at various points in the logistics pipeline.
• Storage of finished goods near the point of production.
• Storage of wholesale or retail inventory pending distribution to final
customers.

From an economics perspective, warehousing is a clear cost. There are the

fixed and variable costs of the warehousing activity, but there are also the costs of
holding inventory in the system. There are the direct costs of having assets sitting
in a warehouse, but added to these are the costs of physical decay of the prod-
ucts, their potential redundancy in the supply chain (a major concern for fashion
items), and the danger of theft or damage being done. The benefits of warehous-
ing come from reducing risk in the system, for example by having spare stock to
meet contingencies, and from allowing economies of scale in production and in
transport. In the context of the latter, it allows larger units to be transported less
frequently.
One aspect of inventory optimization that has gained in importance in
recent years relates less to the area served by a producing unit and rather more to
the area served by warehouses. Changes in transport technology and information
systems combined with the increased importance of high-value, low-weight man-
ufactures in the economy has brought forth new physical distribution systems.
The traditional model (Figure 10.2) assumed a trade-off between warehousing
and transport with the costs of operating warehouses rising with their number
but transport costs falling, assuming a constant throughput of goods. The
optimal number of warehouses would be determined by minimizing the com-
bined costs of warehousing and transport (that is, Qw).
Modifications have incorporated the costs of holding inventories, essentially
warehousing and handling, into the calculations that, in the simplest of terms,
argue that variations in the costs of holding inventories follows the ‘square root
law’. This law effectively says that safety and cycle inventory requirements are
related to the square root of the number of warehouses in the system. Hence,
moving from a system of ten depots to an entirely centralized system would
reduce inventory requirements by 68 percent. This has provided justification for
companies reducing the number of warehouses and with this the market area
served by each. The break-even point for degrees of warehousing, however, is
sensitive to prevailing costs of holding inventories – in effect to the interest rate.

BUTTON 9781786435668 PRINT.indd 372 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 373

Operating costs
10,000 tons
Combined costs

Transport costs

Warehousing costs

0 QW Number of warehouses

Figure 10.2 Trade-off in transport and warehouse operating costs

Reducing warehousing essentially means that more inventory is being held in the
narrower transport system, in containers on trucks or rail cars, or in the holds
of ships or aircraft (transit inventory). This has been made more attractive as
the cost of tracking and tracing have fallen as telecommunications and related
technologies have advanced in sophistication and fallen in cost. This has not only
made the transport system more internally efficient, but, by providing more infor-
mation to consignors, has given the ability to reduce inventory holdings. This is
the basis for just-in-time production, or lean logistics. Some of the claimed gains
that can result from just-in-time, as opposed to conventional, production are
listed in Table 10.2.
The challenge is to find the optimal mix between using more transport to
facilitate greater frequencies of deliveries, and greater reliability in those deliver-
ies, and the reduction in inventory holdings. What a manager is trying to do is to
optimize his or her stock, which consists of three elements:

Target stock = Forecast demand next period + Forecast demand in lead time +
Safety stock

The problem is that the forecast elements have inherent uncertainty in them,
and so the safety stock acts as a buffer. Improved transport systems, packaging,
and information systems have effectively reduced the scale of inventory holdings
required for safety purposes. Much of the rise in just-in-time practices came
with the advent of ‘dependent demand inventory’ systems from the late 1960s,
which, rather than traditional approaches that relied on forecasts premised on
total demand being made up of a number of independent separate demands,
focused attention on the interdependent needs for delivering a target near-term
output.

BUTTON 9781786435668 PRINT.indd 373 27/04/2022 11:07

 374 TRANSPORT ECONOMICS, 4TH EDITION

Table 10.2 The benefits of just-in-time production

Area of improvement Description

Productivity Manufacturing productivity can improve by 20 to 40%

Lead times Lead times can be reduced by more than 90%, reducing the
need to carry finished goods inventories
Flexibility Improves production flexibility, by allowing changes to the
manufacturing schedule to meet changes in market demand
Market power Market position is strengthened by the ability to supply on
short notice, with greater flexibility and greater reliability
Quality Consistency of quality improves
Inventories Substantially reduces inventories, often by 60 to 80%
Floor space Manufacturing floor space requirements are reduced by 30 to
40% because there is no ‘waiting’ inventory stacking on the
floor
Investment Capital required to operate a business is significantly reduced,
estimated reductions being 25 to 50%

Source: Warne and Rowney (1986).

The possibility of more frequent deliveries means that the period over which
demands are made is shorter and thus uncertainty is reduced. It is not just
reduced lead times (the period between an order being placed and the delivery of
that order) that are important, but also the reliability of delivery. Rail transport,
for example, in the United States is often used in just-in-time manufacturing
systems because of its reliability rather than its speed. Of course, higher costs
of meeting short lead times need to be taken into account in the arithmetic.
Figure 10.3 offers a simple illustration of the situation.
Conventional management requires that there is always a large safety stock
of inventory to cope with unexpected contingencies. Alternatively, it has large and
infrequent deliveries of stock that it holds in its warehouses. Just-in-time inven-
tories are small in magnitude but the number of deliveries is larger. To maximize

Inventory ($)
Conventional
deliveries
Just-in-time
deliveries

Conventional
management
Just-in-time
management

0 Time

Figure 10.3 The economics of just-in-time production

BUTTON 9781786435668 PRINT.indd 374 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 375

efficiency, the manager has to set the cost of alternative delivery patterns against
the level of stock that is being warehoused.

10.4 Consolidation and Trans-shipment

Warehousing is simply a storage function and adds little value to the supply chain
other than providing insurance through the holding of safety stocks. In other
words, a more dynamic form of warehousing involves consolidation and trans-
shipment, a direct element of value added to the logistics supply chain. In this
sense, when warehousing is used in this way, it becomes part of a hub-and-spoke
structure that enables the suppliers in the system to benefit from economies of
scope and density in the logistics chain, just as airports do in the provision of air
passenger services (see Chapter 5).
To look at the optimum level of consolidation, it is useful to break down the
costs involved along the lines of Figure 10.4. The diagram considers the problem
in terms of the optimal payload for a delivery vehicle. The line-haul costs of
moving goods to and from the consolidation point fall with the size of consign-
ment carried due to scale effects, until the maximum physical or legal average
payload on a vehicle is reached. If only haulage costs were to be considered in the
warehousing decisions, then this would represent the optimal payload. However,
there are also the resource costs involved with warehousing and consolidation
itself: the provision of depots, handling staff, administrative costs, etc. These
terminal costs are likely to rise with the level of warehousing and trans-shipment.
Consequently, the logistics supply-chain manager, when considering options and

$ per ton
carried External costs

Inventory holding costs

Maximum attainable
tonnage

Line-haul costs

Terminal costs

0 Lp Lt Le Ls Average
payload
Figure 10.4 Freight consolidation costs (represented cumulatively)

BUTTON 9781786435668 PRINT.indd 375 27/04/2022 11:07

 376 TRANSPORT ECONOMICS, 4TH EDITION

taking both broad elements of cost into account, will feel the optimal level of
consolidation would imply an average payload of Lt in the diagram.
So far, we have only looked at the terminal and movement costs confronting
the transporter; however, the final customer awaiting delivery will also have costs
that vary with trans-shipment levels. The greater the amount of warehousing and
the larger the final average payload per vehicle, the fewer the number of deliveries
that will be needed. Longer frequencies between deliveries push up the costs of
stock-holding for customers and the overall level of inventories held. Thus, the
time costs of increased consolidation rise with the average payload, suggesting
that, overall, final recipients of goods would prefer a level of consolidation con-
sistent with an average payload of Lt in Figure 10.4.
The diagram shows a clear distinction between the direct costs influencing
the transporter’s optimum and those factors affecting the final customer of the
service, and it is the bringing together of these elements that supply-chain logis-
tics embraces.
As we have seen in Chapter 6, there are also wider external costs influenc-
ing those not directly concerned with transport operations; these include those
affected by vehicle noise or fumes or who have their own travel disrupted by freight
vehicles. Generally, increased consolidation and higher payloads will reduce these
costs, because fewer trips are needed, even if just-in-time production is part of
the process, to transport the same volume of goods, and consolidation generally
means less environmentally intrusive vehicles can be used in sensitive areas (Button
and Pearman, 1981). Hence, from society’s point of view, the optimal level of
consolidation in the diagram is when all costs are minimized, that is, at point Le.

10.5 Mode Choice

Logistics is not a transport-mode-specific activity and entails the use of a diver-

sity of modes, either within a single logistics provider’s own fleet or using for-
warders and consolidators who make use of third-party transport suppliers. The
way mode choices are made within a supply chain are, in economic terms, not that
conceptually different from the way individuals make decisions about whether to,
say, make a trip by car or bus. It is all about opportunity costs. We consider the
more technical aspects of modeling and forecasting in greater detail in Chapter 12
and here just provide a flavor of the type of models that are used. We also focus
on the more micro models, rather than models that are used, for example, in
national network planning (Button, 2005).
One way of looking at the decision-making process of logistics service provid-
ers is to treat the alternative modes as a bundle of attributes – availability, speed,
cost, capacity, security, safety, and so on – and then select the mode that comes
closest to meeting the providers’ requirements given their financial and other
constraints. This was the approach developed some years ago by Baumol and
Vinod (1970) when considering a situation where speed and economy of service
are important; adding more factors complicates the mathematics but not the logic.

BUTTON 9781786435668 PRINT.indd 376 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 377

Speed is important to minimize the amount of ‘inventory on wheels’, essen-

tially the lead time, and regular services to termini are usually preferred by con-
signors for their own production purposes, the possibility of delays necessitating
keeping a larger safety stock. Initially the problem of safety stocks is assumed
away and an iso-cost equation of this general form is used:

e = vp/(Kvp – 1) (10.1)

where:

e is the reciprocal of the shipping cost per unit of commodity;

v is the reciprocal of the average time required to complete a shipment;
p is the reciprocal; and
K is set equal to the expected annual variable cost of handling.

Allowances need to be made for uncertainty in delivery and safety stocks,

and doing this provides an operational model of the form:

C = T/e + T/(pv) + a/s + wsT/2 + wh{[s + (1/v)]}T0.5 (10.2)

where:

C is the expected annual variable cost of handling;

T is the amount transported per year;
a is the cost of ordering and processing per shipment;
w is the warehousing cost per unit per year;
s is the average time between shipments; and
h is a parameter derived from the Poisson distribution used to simulate the
stochastic elements of the problem.

The final items in equation (10.2) reflect the fact that when the timing of
deliveries is uncertain, slow deliveries require larger safety stocks which increase
the costs of warehousing. A shipper will select the mode of transport that will
complete the delivery at the lowest annual variable cost of handling.
This approach provides guidance as to the mode that would be selected for a
consignment, but not the total demand for the services of any given mode. The latter
may, however, be of interest for the providers, either in the public or private sectors
that provide infrastructure and operational capital (truck fleets, rail cars, ships, etc.).
Aggregate demand can be assessed if it is assumed that profit maximization is the
goal of the providers (which is largely realistic, say for United States railroads but
not for state-owned roads); that the demand curve is linear, of the form Δc = α – βT;
and that the safety stock is proportional to the total volume of shipments (rather
than the square root specification used in equation (10.2)). This then gives:

T = (1/b){Δc – 1/e – 1/(pv) – ws/2 – wh[s + (1/v)]0.5} (10.3)

BUTTON 9781786435668 PRINT.indd 377 27/04/2022 11:07

 378 TRANSPORT ECONOMICS, 4TH EDITION

where Δc is the price difference between origin and destination and b is a constant
equal to –dΔc/dT.
This illustrates the fact that the annual tonnage shipped will be larger, the
greater the price difference between the destination and the origin, the smaller the
time taken between shipments, the smaller the time taken by a shipment, and the
flatter the slope (b) of the demand curve for the commodity. However, these intui-
tively satisfying conclusions come at a price. The linear specification of the model
is convenient for estimation purposes but may deviate significantly from reality.
The analysis of Baumol and Vinod (1970) is about warehousing and
inventory management; it does not fully take account of consolidation and
trans-­shipment that is an integral part of many supply chains. This is often a
multi-modal (inter-modal) activity – for example, using shipping for the long
haul, and trucks and rail for access and egress to and from ports. The question
then often becomes one of whether to use the same transport mode for the entire
trip – say the truck as in Figure 10.5 – or to use a combination of modes and
trans-ship or consolidate at some point in the system.

Truck
Inter-modal
(p and BC)

B
A

0 d d* Distance

Truck
Inter-modal
(p and BC)
Inter-modal
(p* and BC*)
C
C*
B
A

0 d d** d* Distance

Figure 10.5  ost per ton of direct and inter-modal transport with and without support for
C
trans-shipping and rail rates

BUTTON 9781786435668 PRINT.indd 378 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 379

Figure 10.5 offers a simple example of what this means and the implications
of technical change, or policy shifts, in terms of mode split (Beuther and
Kreutzerberger, 2001). This considers an option of moving freight from A to B
using either a truck or a rail/truck combination. The upper diagram shows a com-
parison between trucking which is a loading/unloading cost of 0A plus a trucking
rate of r per mile. The cost by inter-modal road and rail is 0A plus AB (the cost
of using a truck over distance 0d) plus the trans-shipment cost of BC plus a rail
rate of p per mile. The cut-off point in the figure, when inter-modal transport
becomes less costly than rail, is at 0d*. The relative mile rates of the r and p, com-
bined with the trans-shipment costs, determines this point. If there are technical
changes, which one may think of in terms of the advent of container technology,
whereby these parameters are reduced so that BC* < BC and p* < p, then this will
shift the comparative advantages of the alternatives; the threshold distance would
decrease to d* < d for inter-modal transport.
A further aspect of mode choice is the extent to which third-party logistics,
the employment of an outside transport operator and logistics suppliers, should
be engaged and the degree to which a company should operate its own private
transport fleet (‘own account operations’ in the United Kingdom). Many of the
largest fleets of trucks are used to haul a company’s own products, as can be seen
in Table 10.3. There were more than 200,000 companies, excluding farmers, oper-
ating private fleets in the United States in 2018 and these accounted for 68 percent
of the outbound freight.
Surveys conducted over the years, such as that in the United Kingdom and
reported in Table 10.4, suggest that simple costs of carriage do not dominate
decisions regarding whether third-party carriage use, but broader notions of
costs and quality of service, such as reliability, security, and flexibility, also play a
role. While the exact weights placed on the individual items inevitably vary with
industrial sector and the size of companies, as well as changing over time, the
items listed remain much the same. Globalization, lead-time reductions, customer
orientation, and outsourcing have, for example, been important considerations

Table 10.3 Largest private truck fleet operators in the United States (2018)

Company Industry City Vehiclesa

AT&T Utilities & like services San Antonio 66,879

PepsiCo Food products Plano 62,400
Comcast Corp. Business or home services Philadelphia 37,000
Waste Management Sanitation Houston 32,056
Verizon Communications Utilities & like services Elmont 20,011
Time Warner Cable Business or home services Charlotte 19,903
Quanta Services Inc. & subs Construction Houston & subs 18,920
Republic Services Inc. Sanitation Phoenix 18,652
Waste Connections & Operating Co. Sanitation The Woodlands 15,574
CenturyLink Utilities & like services Denver 15,319

Note: a. The composition of vehicle fleets can vary considerably between operators. AT&T, for example, had 49
tractors and 66,830 trucks whereas PepsiCo had 14,300 tractors and 48,100 trucks.

BUTTON 9781786435668 PRINT.indd 379 27/04/2022 11:07

 380 TRANSPORT ECONOMICS, 4TH EDITION

Table 10.4 Reasons for electing to use a company’s own truck fleet

Factor Score

Reliability 14.9
Control 13.0
Customer relations 9.4
Speed of delivery 9.2
Flexibility 7.8
Cost versus price 7.4
Ability of ‘own account’ to meet timing constraints 6.6
Price is subordinate to service considerations 6.5
Specialized capacity 5.5
Speed of response 5.1
Adaptability 3.6
Consistency 3.5
Avoidance of damage of consignment 3.4
Security 2.6
Other (non-financial) 1.1
Other (financial) 0.5

Source: UK Department of Transport (1979).

since the 1990s. Further, new firms from different fields have entered the market,
competing with traditional transport and warehousing firms to offer different
packages of service attributes (Hertz and Alfredsson, 2003).

10.6 Urban Logistics

Many of the challenges in logistics involve collection and delivery in urban areas;
often called the ‘last-mile problem’. This may be misleading because urban freight
transport can involve four types of movement: in, out, within, and through, and
clearly the latter may not be the last mile of a trip. At the same time, because
collection and delivery is often done by road transport that uses a common infra-
structure with passenger modes, and can involve numerous stops, urban distribu-
tion can add more than proportionately to larger traffic congestion problems.
What constitutes urban freight transport in this context is not, however, clearly
defined. There has, for an example, been a considerable reduction compared with
50 years ago, in the number of household deliveries of goods; people collect them
from stores, and particularly large hyper- and supermarkets. Should this be treated
as part of this ‘last-mile problem’ or is it part of consumption behavior? The issue
may be seen in terms of whether the purpose of a trip is explicitly to transport
freight, but even here it is not always easy to separate out the urban logistics part
of a multi-purpose trip from, say, that part involving getting to or from work
(D’Este, 2007). We have a joint product situation in these circumstances.
Table 10.5 provides a categorization of the various commercial transport seg-
ments serving the urban market. It shows quite distinct market segments meeting

BUTTON 9781786435668 PRINT.indd 380 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 381

Table 10.5 The market sectors of urban freight

Market sector Truck type Commodity Load type Route Trip type

Courier Small Mixed LTL Variable Very complex linked

trips
General carrier Intermediate Mixed LTL or FTL Variable Variable – simple or
size linked trips
Specialist commodities Large Specific FTL Regular Mostly simple trips
(e.g. bulk liquid)
Over-sized/hazardous Large Specific FTL Fixed Simple trips
External Large Mixed or FTL Regular Simple trips – external
specific to/from a single point

Note: LTL = less than truck-load; FTL = full truck-load.

the demands for carriage of various types of commodity. From a transport eco-
nomic perspective, no single market structure emerges. Couriers, for example, that
often include integrated suppliers such as FedEx, DHL, and UPS, offer extensive
networks of service with multiple local pick-up and delivery points – often a
de facto personal service – and there are clear economies of scope and density
involved. Other, more specialized types of service have simpler networks and are
thus more amenable to competitive supply.
Improving urban logistics can have significant economic benefits for their
users. Table 10.6 offers a few examples of both the importance of local freight
transport, but also some of the gains that have come about by enhancing the
system. Of course, there have also been failures when inappropriate changes have
been made. As with any activity, the costs of change must be set against the ben-
efits, and in the urban logistics case there are such issues as the relative costs of
changing the entire distribution network as opposed to making marginal changes
to elements of it. The more complex the network, the more difficult it is generally
to change the entire system.

Table 10.6  The estimated benefits of improved urban distribution systems

Supply-chain study Scope of study Estimated savings

Kurt Salmon US dry grocery sector 10.8% of sales turnover (2.5% financial,
Associates 8.5% cost); total supply chain $30 billion,
warehouse supplier dry sector $10 billion;
supply chain cut by 41% from 104 to 61 days
Coca-Cola supply 127 European companies 2.3–3.4% of sales turnover (60% to retailers,
chain focused on cost reduction from 40% to manufacturers)
the end of manufacturing line
ECR Europe 15 value-chain analyses 5.7% of sales turnover (4.8% operating
(10 European manufacturers, costs, 0.9% inventory cost); total supply
5 retailers), 15 product chain saving of $21 billion; UK savings
categories, 7 distribution $2 billion

Source: Adapted from Fiddis (1997). This paper contains the full references to the studies cited.

BUTTON 9781786435668 PRINT.indd 381 27/04/2022 11:07

 382 TRANSPORT ECONOMICS, 4TH EDITION

The nature of urban logistics systems and retail systems differs between countries,
and this can partly be explained in terms of the market power within the spe-
cific supply chain and the motivations of the various actors. For example, Alan
Mitchell (1997) pointed to the fact that most German and French retailers tradi-
tionally tended to be privately owned or franchised operations and largely moti-
vated by volumes of sales and price when planning their strategic positioning.
In contrast, United Kingdom firms tend to be publicly quoted companies
motivated by margins, and thus have a more constructive approach to their sup-
pliers, including transport services. The United States system also differs because
there is greater market segmentation, wholesalers have more power in the supply
chain, and there is more focus on customer promotions and greater government
intervention in the market, especially regarding anti-trust issues. These factors
determine the bargaining ability of the transport suppliers in the market and
the types of services that are demanded from them. In the United Kingdom, for
example, the shift has largely been to one controlled by retailers, with their needs
being pushed through to the transport market. This, combined with the long-
established deregulated trucking market (following the 1968 Transport Act) that
offers significantly lower cost services, has meant the widespread use of third-
party logistics in the United Kingdom, but a slower uptake in the United States.

10.7 Green Logistics

Chapters 6 and 8 focused on some of the major environmental challenges imposed

by transport, and on the economic implications of some of the ways of reducing,
if not always removing, them. Here we focus specifically on the economic aspects
of reducing the environmental costs of the larger freight supply chain. This has
been a subject of increasing interest as trade, both within and between countries,
has grown and the supply chain has become more extended. Interest has extended
beyond traditional environmental groups and government to embrace trade
bodies and many logistics service suppliers.
The supply chain makes use of a diverse range of transport inputs that in
turn impact upon the environment. These differ sometimes from the environmen-
tal externalities associated with passenger modes because they can involve what
is carried as well as the pure transport-generated environmental externalities,
but from an economic perspective can largely be treated in the same way. While
there are market incentives for the private sector to adopt greener logistics supply
chains, official policies have also been deployed, mainly to reduce energy con-
sumption. A range of policy options are available, but while there are a variety
of approaches, the pros and cons of the alternatives, with some minor nuances,
are largely the same as with passenger transport (Martin et al., 1995). They often
entailed the types of measures outlined in Chapter 8 for transport more gener-
ally, but in addition there are often specific measures covering elements of the
logistics supply chain. Table 10.7 provides details of the potential impact of some
strategies that have been considered to foster fuel consumption and thus also

BUTTON 9781786435668 PRINT.indd 382 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 383

Table 10.7 Macro studies on fuel consumption of impacts of alternative logistics

structures

Country Time period Main measures Changes in energy

consumption
Business as With
usual rationalization

Netherlands 1990–2015 Increased fuel efficiency; +47 246

consolidation of loads;
reduced empty running;
large modal shift to rail
Germany 1988–2010 Improved fuel efficiency; +55 +28
large modal shift to rail;
higher vehicle utilization;
shorter distances traveled
United 1995–2020 Halving the rate of ton-miles +56 24.4
Kingdom growth to 10% per decade;
large modal shift to rail &
water

Source: Adapted from McKinnon (2003).

CO2 emissions. What they do not show, however, are the detailed mechanisms for
bringing this about, nor the costs that these changes would impose on the supply
chains involved.
Logistics also plays a reverse role in the larger production process in terms
of product returns, source reduction, recycling, material substitution, reuse of
materials, waste disposal, and refurbishing, repair, and remanufacturing. This is
known as ‘reverse logistics’. The combination of making the transport element
of the overall supply chain more environmentally benign and the role of reverse
logistics in optimizing the use of resources in the entire production process
through the facilitation of recycling, refurbishment, and so on of products is
known as ‘green logistics’. It is a wider approach to the interactions of transport
and the environment than simply making transport per se less environmentally
intrusive (Rodrigue et al., 2001). From a narrow economic perspective, the com-
mercial advantage to the transport industry of such things as recycling and
refurbishment is that it can provide return loads and thus reduce the back-haul
problem.
Recycling and refurbishment is growing in importance, not just for envi-
ronmental reasons but also as a simple commercial proposition. Some exam-
ples of the gains from recycling compared to entirely new extraction and
processing include 95 percent energy and air pollution savings for aluminum,
between 40 and 73 percent for paper, 5 and 30 percent for glass, and 20 and
60 percent for steel. The amount of recycling that takes place varies by com-
modity and between countries; for example, the United Kingdom recycled
42 percent of steel packaging in 2002, with Belgium recycling 93 percent,
Germany 79 percent, and the Netherlands 78 percent. The European average

BUTTON 9781786435668 PRINT.indd 383 27/04/2022 11:07

 384 TRANSPORT ECONOMICS, 4TH EDITION

Table 10.8 Amounts of hazardous goods moved in the United States (2017)

Mode Million tons Billion ton-miles Average miles per shipment

For-hire truck 932.7 92.1 153

Private truck 882.2 34.7 28
Rail 90.4 61.7 640
Water 304.2 60.9 72
Air (incl. truck & air) 0.3 0.2 1,333
Pipeline 679.8 n.a. n.a.
Parcel, postal, & courier 0.3 0.2 949
Other 78.1 75.3 4

Note: n.a. = not available.

Source: US Bureau of Transportation Statistics.

was 60 percent. Outside of Europe, the figure for Australia was 43 percent, Korea
47 percent, the United States 59 percent, South Africa 63 percent, and Japan
86 percent.
Some materials moved are also hazardous – especially certain chemicals –
and command-and-control instruments involving routing, timing, and packaging
are often adopted to minimize any external costs of spillage (Hancock, 2001).
This applies to cargoes carried by water, pipeline, and air, as well as road and rail.
The amounts involved are quite large, about 10 percent of the total ton-miles
of freight movement being done in the United States. Table 10.8 offers some
indication of the ways in which this is moved and the distances involved. The
importance of railroads and multi-modal transport for long-distance movements
is clear.
Command-and-control instruments are preferred over pricing largely
because of the transaction costs involved. It is relatively easy to highlight haz-
ardous loads by signs on vehicles and this makes enforcement of such things as
routings and use of specified infrastructure, such as designated lanes on freeways,
relatively easy. There is clearly, however, a quasi-pricing mechanism involved
through the insurance market and in many cases compulsory insurance coverage
is used to complement command-and-control instruments.

10.8 International Logistics

The major growth in international trade has both been facilitated by innovations
in international logistics and has been a driving force for new approaches to logis-
tics. The introduction of the container from the late 1960s significantly cut both
the financial and the time costs of freight movements. Institutional changes, both
at the global level (including tariff reforms under the auspices of such agencies as
the World Trade Organization) and domestically in many countries (which have
enhanced the efficiency with which goods can be moved to and from gateways,
international ports), have been complementary to this.

BUTTON 9781786435668 PRINT.indd 384 27/04/2022 11:07

 385
ECONOMICS AND TRANSPORT LOGISTICS ­

There has also been a move towards internationalization of freight transport

service suppliers. This has involved the formation of strategic alliances in ship-
ping and cargo airlines, and in the United States and abutting countries between
the freight railroads. The economic rationale for this is parallel to economies
of scope, density, and market presence found in the passenger airline sector
discussed earlier in the book. In addition, there have been direct international
mergers and take-overs that have brought about even greater integration. Some
of these have involved very large companies, such as the mega-shipping lines P&O
Nedlloyd and AP Moller Maersk, which merged in 2005.
Figure 10.6 offers a very stylized picture of what generally has happened
regarding freight movements before and after either technology changes (such as
the adoption of the container), mergers, or, more often, the relaxation of insti-
tutional barriers to international trade. Not only was there a shift to more cross-
border trade, but also the amount and complexity of goods movements within the
nations involved increases.
This type of pattern has been observed in the European Union since the
removal of internal restrictions of trade from the early 1990s and as additional
countries have joined periodically since 2000. While the creation of the European
Economic Community under the Treaty of Rome in the late 1950s had intended
to introduce free trade between member states, progress was slow. After 30 years,
as we see in more detail in Chapter 14, there remained serious obstacles to the
movement of goods. Some indication of the administrative costs of international
border crossings within Europe prior to the Single European Market initiative
are seen in Table 10.9. Estimates made by the European Union in 1996, after

Before integration After integration

Network

Flow

Figure 10.6 Freight movements before and after international integration

BUTTON 9781786435668 PRINT.indd 385 27/04/2022 11:07

 386 TRANSPORT ECONOMICS, 4TH EDITION

Table 10.9 
Estimated annual cost of administrative formalities and border controls in
Europe before the Single Market initiative

$ billions (1988)

Administration 9
Delays 1
Public spending on customer operations 1
Lost business opportunities 5 to 19
Total 16 to 30

the removal of border controls under the Single European Market initiative that
came into effect in 1992, suggest that logistics costs fell by as much as 27 percent
between 1987 and 1992. About half of this saving was realized in the transport
sector as idle time at borders was reduced and service reliability improved.

10.9 Big Data, Supply Chains, and Economics

The economic efficiency of any supply chain is heavily dependent on the quality
and relevance of the data on both sides of the various micro markets involved.
Exact definitions in any dynamic field of study or application is seldom easy.
Broadly, big data can be seen as a field that analyses systematically extracted
information relating to datasets that have traditionally been too large or complex
to be dealt with by traditional data-processing application software. Entwined
with this is the increased use of business analytics. This offers the ability to gain
insight from data, applying mathematics, econometrics, simulations, and applied
statistics to improve the quality of decisions (Selod and Soumahoro, 2020).
Transport logistics examples include global positioning systems and other
location-aware technologies that are producing data which are specific down to
particular latitude and longitude coordinates and seconds of the day. Another is
the data continually gathered from large commercial aircraft in flight, monitoring
their technical performance and allowing more efficient planning of highlighting
needs and aircraft rotation. From the public sector perspective, collecting detailed
information of traffic flows and delays at junctions across a network can enhance
the efficiency of traffic signal sequencing.
Just considering the two aviation examples given, the ability to more rapidly
and accurately predict weather patterns allows for adjustments in flight paths,
while engine monitoring facilitates pre-emptive actions to prevent technical fail-
ures. Both help reduce costs and enhance the value of aviation in supply chains,
effectively pushing up the production function of transport service suppliers.
The take-up of big data by those engaged in logistics and supply-chain man-
agement has been gradual (Wang et al., 2016). While definitions remain a little
fuzzy, in 2014 survey data collected by the consultants Accenture from over 1,000
executives engaged by global companies found 97 percent of respondents were
interested in implementing big data analytics. But it was also found that only 17

BUTTON 9781786435668 PRINT.indd 386 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 387

percent had implemented it internally or were outsourcing its implementation.

The rationale for implementation was that not only can big data and business
analytics reduce costs, but by improving visibility it can increase the logistics
services demanded and, with that, revenues. Better information regarding con-
sumers’ habits and preferences allows both services to be tailored more closely to
those being sought as well as pricing that extracts maximum revenue. Those still
waiting were concerned, however, about investment costs, security, and privacy.
Deciding how to incorporate big data analytics into companies’ business models
was a consideration as well.
Big data is also an almost necessary requirement for competitive modern
logistics markets. The neoclassical ideal of perfect information is something of
a straw man, but big data and associated analytics can enhance competition and
increase economic welfare. Just as it can increase information enjoyed by trans-
port service suppliers in the chain, it can make more visible to customers the rela-
tive qualities, prices, and other attributes of these transport firms. Big data can
provide benchmarks. Often, those seeking logistics service have multiple criteria
for making their selections. Big data is increasingly being used in conjunction with
techniques such as analytical hierarchic processes which break down complex
problems into separate small-size subproblems, to help with this (Ekici, 2013).
In the context of modeling transport markets, while much of the data used are
quantitative, big data does offer sources of qualitative data in some circumstances.
This generally involves the use of subjective assessments on the part of consumers
or users when past series of hard data are not available or there are gaps in them.
There are challenges in using big data even when data series are complete:

• Its volume accumulates data noise and incidental homogeneity.

• Volume creates high computational costs and algorithm instability.
• High variety requires different techniques and methodologies.

These factors generally produce heterogeneity, experimental variation, and statis-

tical bias that are not well handled by standard econometric techniques designed
to analyse smaller datasets. More robust and adaptive methodologies are, there-
fore, often adopted.

10.10 Security

One of the traditional concerns of supply chains is that of security. At one time
or another, all elements of transport networks have historically found themselves
prone to attack both in conventional wars and by terrorists (broadly defined).
While the focus of these attacks has in the past often involved links and termi-
nals in inter-urban networks – airlines/airports, railway lines/railway stations,
etc. – the local, urban, and suburban elements of logistics chains have not been
left unscathed. In the twenty-first century, attacks have also extended beyond the
hardware of logistics networks to include their management. Examples include

BUTTON 9781786435668 PRINT.indd 387 27/04/2022 11:07

 388 TRANSPORT ECONOMICS, 4TH EDITION

ransomware attacks on such things as pipelines. The 2021 ransomware attack on

the Colonial Pipeline Company IT system resulted in the shutting down of the
pipeline supplying oil to most of the eastern United States.
Security is a large subject and the focus here is on the vulnerability of shop-
ping malls – important urban and suburban interfaces between inbound flows of
goods to retail outlets and outbound flows from retail outlets to final consump-
tion (Button, 2008). They are an integral part of the supply chain and also an
element that is both high-profile and difficult to secure. We also add a few words
on shipping matters.
Shopping malls provide important retail outlets that offer shoppers econo-
mies of scale and scope in their purchasing and economies of market presence for
retailers in their selling. They have grown significantly in number in the United
States and in many other countries over recent years, with increases in their size
and range of facilities provided. They also often act as major recreation centers
containing cinemas and food outlets in addition to stores selling commodities to
take away. Strolling and ‘hanging out’ in malls can also provide a distraction for
teenagers. Overall, from a security perspective, they represent large gatherings of
citizens that offer possible targets for terrorists. Indeed, the US Department of
Homeland Security categorizes shopping centers, along with other easily acces-
sible public places, as ‘soft targets’.
Traditional concerns with security, involving theft and attacks on individu-
als, have focused on the location of malls. Older malls located in middle-income
groups and accessible only by automobiles were previously considered relatively
‘safe’ but with increasing suburbanization and changing demographics this is now
less the case: 65 percent of United States and Canadian malls of over 400,000
square feet are located in areas with above-average crime risk, with 25 percent
being in areas with three times or more the national crime risk. Mall design and
that of its surrounding area also plays a role in crime rates. Crime tends to be
higher in malls with spots that are attractive to juveniles and ‘non-shoppers’, and
where there are crowded stores offering easy access.
Much of the discussion of mall security and terrorism has relied on
­anecdotal evidence, perhaps not unexpectedly given the nature of the ‘games’
that are played with potential terrorists. Their features would seem to make
them very vulnerable to terrorist attacks of various kinds: they have multiple
entrances and exits, are open to the public, have numerous potential places for
­concealment, often have adjacent car parking, have frequent deliveries, attract
numerous shoppers, and employ large numbers of workers. Additionally, since
their initial introduction in 1956, many of the larger malls in the United States
(some 1,200 or so) and elsewhere are enclosed with extensive air conditioning
and ventilation systems making them potential targets for toxic chemical or
biological attacks. Other developments in the supply chain have seen the greater
containerization of consignments reducing considerably the transparency of
what is being delivered.
Security has thus always been an issue at malls and it is the emphasis that
has changed. Traditional security has been focused on preventing petty crimes

BUTTON 9781786435668 PRINT.indd 388 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 389

Exhibit   Costs and benefits of transport security

The September 11, 2001 attacks in the United States on the World Trade Center and
the Pentagon brought forth a 21-layered response from the Transportation Security
Administration. This involved 15 pre-boarding and deterrence and 6 in-flight security
measures; see the table.

Pre-flight In-flight

Intelligence Passenger resistance

International partnerships Trained flight crew
Customs & border protection Hardened cockpit doors
Joint terrorism task force Federal Air Marshal Service
No-fly list & passenger pre-screening Law enforcement officers
Crew vetting Federal Flight Deck Officer Program
Visible inter-modal protection response teams
Canines
Behavioral detection officers
Travel document checker
Checkpoint/transportation security officers
Checked baggage
Transportation security inspectors
Random employee checks
Bomb appraisal officers

Subsequently, Mark Stewart and John Mueller sought to estimate the costs and benefits
of three of the main security measures designed to reduce the likelihood of a direct
replication of the 2001 attack. They broke down the net benefits of each into the following
components:

Net Benefit = pattack × Closs × ΔR − Csecurity

where:
pattack is the probability of a successful attack is the likelihood that a successful terrorist
attack will take place if the security measures were not in place;
Closs is the losses sustained in the successful attack include the fatalities and other damage –
both direct and indirect – that will accrue as a result of a successful terrorist attack;
ΔR is the reduction in risk is the degree to which the security measure foils, deters, disrupts,
or protects against a terrorist attack; and
Csecurity is the costs are those of providing the risk-reducing security measure required to
attain the benefit.

The three security measures examined were: the installation of physical secondary barriers
(IPSB) to restrict access to the hardened cockpit door during door transitions, the use
of the Federal Air Marshal Service (FAMS), and the Federal Flight Deck Officer (FFDO)
Program. Measures such as scanning checked bags or canines were not explored because

BUTTON 9781786435668 PRINT.indd 389 27/04/2022 11:07

 390 TRANSPORT ECONOMICS, 4TH EDITION

they were not relevant to the 2001 attacks. Only measures aimed at stopping undetected
boarding of aircraft, hijacking of aircraft, and success in entering the cockpit were
examined.
Specifically, in doing this they examined doubling the budget of the FFDO program to
$44 million per year, installing IPSBs in all aircraft at a cost of $13.5 million per year,
and reducing funding for FAMS by 75 percent to $300 million per year. The break-even
cost–benefit analysis then finds the minimum probability of an otherwise successful attack
required for the benefit of each security measure to equal its cost.
It was found that the IPSB is cost-effective if the annual attack probability of an otherwise
successful attack exceeds 0.5 percent or one attack every 200 years. The FFDO program is
cost-effective if the annual attack probability exceeds 2 percent. However, more than two
otherwise successful attacks per year are required for FAMS to be cost-effective. A policy
that includes IPSBs, an increased budget for FFDOs, and a reduced budget for FAMS may
be a viable policy alternative, potentially saving hundreds of millions of dollars per year with
consequences for security that are, at most, negligible.
See also: M.G. Stewart and J. Mueller (2013) Terrorism risks and cost–benefit analysis of
aviation security, Risk Analysis, 33, 893–908.

of theft, minor material damage to the fabric of buildings, and assaults on

­individuals – collectively called ‘crimes’ hereafter for brevity. These problems
continue. For example, in a 1995 survey for the National Retail Federation, the
University of Florida found that the average discount store in 1994 had 0.09
crimes against customers, ranging from simple assault to rape and murder. That
translates into a rate of about 20 crimes against customers per year for chains the
size of Walmart and Kmart. This problem of criminality is now overlapped by
concerns about the possibility of terrorist attacks.
Globally, malls have not been immune from attacks in the past: there have
been over 60 terrorist or terrorist-inspired attacks of various sizes at shopping
malls since 1998 in 21 countries including the United Kingdom, Turkey, Israel,
Russia, and Spain (LaTourrette et al., 2006). Most of these attacks have involved
explosives of various types, but there have also been incidents involving firearms
and chemicals. The events of September 2001 have heightened concerns in the
United States and elsewhere that shopping malls may more frequently become
explicit targets for major attack. There is also increased concern about other
forms of major assault at malls. For example, a bombing took place in October
2002 in Vantaa, Finland, in a local Myyrmanni shopping mall killing seven
people, including the bomber, and injuring 166. The shopping center was particu-
larly crowded, with almost 2,000 people in it, including many children who had
come to see a clown performance.
There have already been some actual and thwarted terrorist attacks at
American malls, although on a relatively minor scale. Five people were killed

BUTTON 9781786435668 PRINT.indd 390 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 391

and four injured in a shooting at a Salt Lake City mall in 2007 by a Bosnian, for
example. There have also been abortive attempts at terrorism. In November 2007
Nuradin Abdi, a Somali living in Ohio, pleaded guilty to planning, with other al-
Qaeda operatives, to blow up a Columbus-area shopping mall.
Nevertheless, and for a variety of economic, political, and social reasons, the
attention paid to shopping mall security has been much less than to many other
parts of the transportation supply chain such as airports, airlines, and chemical
and oil movements.
One of the problems is that the retail industry has treaded warily for com-
mercial reasons. Customers expect shopping centers to be free and open, and mall
owners and operators have been reluctant to introduce stringent security measures,
as airports have done, that might limit shoppers’ access, or scare them off altogether.
Though security officers are usually uniformed, for example, they are not intended
to appear threatening and frequently serve a variety of roles as well as security.
Although the Nobel Prize-winning economist, Gary Becker (1968) and others
have enhanced our understanding of crime prevention in general, the analysis of
terrorism and its prevention is a relatively new area of study for economists, much
of the social science interest in the field coming from sociologists and political sci-
entists. Much of the traditional economic analysis of crime has focused on deter-
ring criminals, but this is of limited relevance in situations when the perpetrator is
intent on committing suicide. Inevitably there has been some theoretical economic
analysis of the appropriate role of public sector and private sector insurance,
essentially revolving around the degree to which the threat of terrorism constitutes
risk or uncertainty, on the moral hazard issues of various compensation policies,
and on the optimal ways to allocate the costs of counterterrorism measures. There
has also been applied analysis seeking to cost the macro costs of terrorism on eco-
nomic performance. Here we are more concerned with embracing some of these
things within an assessment of how optimal security levels may be defined.
One of the problems in trying to apply economics to a subject such as mall
security is that many of the actions are driven by subjective factors, often influ-
enced by only partial information, which are difficult to quantify given the lack
of large amounts of prior data. This makes it difficult to work out probabilities
of attacks and their costs.
The situation is compounded because, unlike the consideration of safety
where there is no feedback mechanism, the perpetrators of terrorist acts modify
and adjust their methods and targets as security measures reduce the vulnerability
of previous types of target. They essentially play games with the security agen-
cies; for example, in some cases they switch targets because they have achieved
their objectives by having diverted considerable security efforts to the old cat-
egory of target. This means that much of the difficulty lies in the uncertain nature
of terrorism in Frank Knight’s (1921) sense of there being no easily calculable
probability distributions associated with attacks. Initially, however, we look at a
Gaussian approach and assume that such probabilities can be calculated.
Figure 10.7 provides a diagrammatic representation of the situation with
security measured in some undefined way. This allows an assessment of the

BUTTON 9781786435668 PRINT.indd 391 27/04/2022 11:07

 392 TRANSPORT ECONOMICS, 4TH EDITION

C* C

B*

0 S# S** S* S Security
Figure 10.7 The determinants of optimal security provision

optimal level of security that should be provided in simple cost–benefit terms.

C is an upward-sloping marginal-cost-of-security curve based on the reasonable
assumption that each increment of security costs more as the most basic meas-
ures are exhausted. The B curve indicates the marginal benefits of additional
security with the flattening of the slope reflecting the decline in additional benefits
associated with the more detailed security measures. The optimal level of security
in this case is at S.
These curves, however, can for analytical purposes usefully be decomposed.
Increased security provides benefits in terms of both a reduced threat of material
(including physical injuries to people) damage in the mall (separated out as B*)
and greater psychological ‘comfort’ to those shopping, or having family or friends
shopping in the mall (the distance between the B and B* curves). The extent to
which this psychological ‘comfort’ exceeds the actual dangers that exist depends
to a large extent on the information that is available to potential shoppers. While
there may be no good estimate of the reduced chance of being involved in an
attack on a mall, there are benefits from simply seeing security measures in place
even if there is no objective method of assessing their effectiveness, although con-
versely people are sensitive to such things as media coverage of threats that push
out the mental benefits of greater security actions. There may also be benefits that
are external to those using and working in a mall, for example to those living in
the neighborhood.
The cost curve also has its complexities. Attacks on malls can inflict both
material damage on the fabric of the buildings and the vitality of the shops in
it, and injury and death on individuals. The C* curve is drawn to separate out
the minimum marginal costs of incremental units of security, including the
additional inconvenience costs to all parties as well as financial outlays, but the
actual full economic costs may deviate from this. (While the B and C curves are
drawn as smooth functions, in practice there may well be kinks if, for example,

BUTTON 9781786435668 PRINT.indd 392 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 393

shoppers have threshold tolerance levels to the degrees of intrusion that security
will impose on them.) The incentive to provide security at the lowest cost may not
be there and, in consequence, X-inefficiency may be present in providing security
measures pushing the cost curve up. Potential inefficiency arises in these types of
situations because objectives tend to be opaque and many of the costs are only
indirectly borne by those responsible for the security provided.
The problems of providing security on the lowest cost curve are com-
pounded if there is asymmetrical information concerning the effectiveness of
security measures – for example, security experts and consultants have an incen-
tive to exaggerate the challenges being faced (consultants after all being rent-
seeking economic agents operating in a commercial market). In effect, there is
the potential for a degree of ‘regulatory capture’ of the security system by those
involved in providing it.
On the other hand, the costs of security measures may reduce insurance pre-
miums and other forms of crime in the mall. The costs may also be off-set to some
extent by the supply chain adjusting at other points, either up or down, as other
actors adjust their behavior. In these circumstances the cost curve will be lower
than that depicted in the figure. The trade-offs involved are empirical questions
and inevitably will differ between malls.
Given these complexities, some suboptimal outcomes may emerge. For sim-
plicity, we assume that gains at other points in the supply chain off-set some of
the costs of enhanced security and any X-inefficiency associated with it. In other
words, {C* – C} is the difference between the actual and the narrowly defined
minimum cost curve of on-site security.
If the attention was purely on the commercial damage that can be caused
by acts of terrorism, as for example may be the case with private insurance
companies, then security will be under-supplied by {S – S*}. But even if the
psychological benefits of more assured security are not ignored, then there may
be inefficiencies in the provision of security measures leading to inadequacies
of {S – S**} in their provision if the agencies responsible do not minimize their
costs. If both full benefits are under-estimated and costs and the provision of
security measures are not carried out efficiently then the resultant level of secu-
rity, S#, could be well below the social optimum.
The conventional public policy approach to this sort of optimization problem
is to apply cost–benefit or cost-effectiveness analysis, but this is difficult in this
case because terrorism involves both uncertainty owing to the limited number of
previous events, and game-playing owing to the reactive function of the terrorists.
To get a genuine handle on the shapes of these curves perhaps the best tools avail-
able are experiments. These may involve some form of conventional simulations
that can help assess static uncertainty, or more interactive approaches deploying
experimental economics that can capture the adaptive behavior of terrorists to
security measures.
While it is generally impossible to identify and quantify many of the costs
and benefits of securing transport facilities, there are important questions about
who should pay the costs if the terrorist threat is to be handled efficiently.

BUTTON 9781786435668 PRINT.indd 393 27/04/2022 11:07

 394 TRANSPORT ECONOMICS, 4TH EDITION

Shopping malls, as an example, are generally owned by the private sector and
traditionally they have provided significant amounts of security within their facil-
ities. Indeed, much of the protection is provided by private security (Davis et al.,
2006). Mall owners and operators have had a vested interest in doing this, and in
cooperating with police forces and other private parties (for example, the security
personnel employed by individual retailers with a mall and the transportation
companies involved in deliveries). Public confidence in the security of people and
property is an important marketing tool for a mall operator.
Given the nature of traditional crime, mall operators have the option of
passing on some of the risk of any breakdown in this type of security – for
example, compensation claims – through insurance markets. They can also, as
most have done, internalize some of the risk by enhancing their private policing.
On the premise that they act rationally, shoppers in the mall also have a personal
interest in protecting themselves (after all, compensation for being killed by a
mugger is hardly ideal!) and their property, and often have the option of taking
out insurance. Individual shops within malls have differing levels of attraction to
criminals and may thus add additional security.
In another context, namely international freight movements, the United
States government (though authorities in other countries have adopted similar
policies) have played upon this private sector interest and developed initiatives
such as the Customs–Trade Partnership Against Terrorism (C-TPAT) that fosters
public–private cooperation throughout the global supply chain (Brooks and
Button, 2006). Firms that participate conduct a comprehensive self-assessment
of their security practices using guidelines developed after consultation with
the national government. This avoids excessive checking by the government and
affords the logistics supply flexibility in its operations on the premise that any
failure in the security arrangements will hit the supplier commercially.
The general problem with security is that the threat posed by terrorists
is different in that it largely involves uncertainty, with little knowledge about
any underlying probability functions, if indeed any exist, rather than risk. The
actuarial approach to the challenge is thus far from perfect, and certainly incom-
plete. The potential costs of a major terrorist incident are also extremely high.
Returning to our primary example, these factors make it virtually impossible for
a mall operator to either completely insure internally, or through the financial
market, against terrorist attacks. This is one reason why Figure 10.6 is very broad
in its nature; the precise quantification of any of the curves shown is virtually
impossible. We can point to additional costs associated with security measures
in several transport contexts. Table 10.10, for example, provides a comparison of
the costs explicitly borne by airline travelers for security measures in 2002 imme-
diately after the September 11 attacks in the United States, and these are only
partial. They do not include costs embedded in higher fares.
There is the added problem that a major attack on a mall could have serious
and much wider economic and geographical repercussions than those simply
relating to the shopping facility. This can be seen and treated as a standard exter-
nality problem if these effects have relatively small impacts on overall price levels

BUTTON 9781786435668 PRINT.indd 394 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 395

Table 10.10 
Average security charges per traveler at airports (2002)

Country Average charge per passenger

Canada $14.50
Germany $10.57
Israel $8.03
France $6.88
Australia $5.19
United States $5.00
Netherlands $4.13
Russia $2.04
Italy $1.90
United Kingdom $0.00

Source: Waters and Yu (2003).

and incomes in the larger economy, but, if this is not so (which may well be the
case), then the conventional tools of microeconomics break down. The stand-
ard ceteris paribus assumptions of neoclassical analysis do not hold. The issue
becomes one of so-called ‘political risk’, although strictly, given the subjective
nature of the issue, it is more about ‘political uncertainty’.
These market failures necessitate a degree of public involvement in terms of
specifying and possibly providing security and (if this fails) resources to cope with
many of the aftermath effects. These are obviously not problems unique to trans-
port security, but they certainly do exist when it comes to looking at appropriate
economic strategies concerning transport infrastructure and operations.
Where the line is drawn between private and public sector responsibility
is partly a technical one, which largely depends on where uncertainty becomes
the dominant concern but also embodies normative judgment involving public
perception of what the exact role of government should be. While normally the
actual implementation of policies is seen in terms of narrower economic effi-
ciency considerations – essentially the undertaking of a cost–benefit calculation
of the effectiveness of public and private actors – when security arises there is
often a high degree of subjectivity involving public perceptions. This was seen in
the debates in the United States over whether private operators should implement
new airport security measures after the 2001 terrorist attacks or whether it should
be by a federal agency. Also indicative of this is that, while there is evidence that
private security does not reduce security in malls, many shopping mall operators
in the United States actively seek a local police presence in their facilities, often a
police station, to give added public confidence.

References

Ackerman, K.B. and Brewer, A.M. (2001) Warehousing: a key link in the supply chain, in
A.M. Brewer, K.J. Button, and D.A. Hensher (eds), Handbook of Logistics and Supply-
chain Management, Pergamon.

BUTTON 9781786435668 PRINT.indd 395 27/04/2022 11:07

 396 TRANSPORT ECONOMICS, 4TH EDITION

Baumol, W.J. and Vinod, H.D. (1970) An inventory theoretical model of freight transport
demand, Management Science, 16, 413–21.
Becker, G. (1968) Crime and punishment: an economic approach, Journal of Political
Economy, 76, 169–217.
Beuther, M. and Kreutzerberger, F. (2001) Consolidation and trans-shipment, in A.M.
Brewer, K.J. Button, and D.A. Hensher (eds), Handbook of Logistics and Supply-chain
Management, Pergamon.
Brooks, M.R. and Button, K.J. (2006) Market structure and shipping security, Maritime
Economics and Logistics, 8, 100–20.
Button, K.J. (2005) Can freight transport models be transferred across the Atlantic?
in A. Regiani and L. Schintler (eds), Methods and Models in Transport and
Telecommunications: Across Atlantic Perspectives, Springer-Verlag.
Button, K.J. (2008) The economics of shopping mall security, Journal of Transportation
Security, 1, 127–39.
Button, K.J. and Pearman, A.D. (1981) The Economics of Urban Freight Transport,
Macmillan.
Coase, R.H. (1937) The Nature of the Firm, Economica, 4, 386–405.
Davis, R.C., Ortiz, C., Rowe, R., Broz, J., Rigakos, G., and Collins, P. (2006) An assessment
of the preparedness of large retail malls to prevent and respond to terrorist attack,
Report, US Department of Justice.
De Jong, G. (2008) Value of freight travel time savings, in D.A. Hensher and K.J. Button
(eds), Handbook of Transport Modeling, 2nd edn, Elsevier.
D’Este, G. (2007) Urban freight movement modeling, in D.A. Hensher and K.J. Button
(eds), Handbook of Transport Modeling, Elsevier.
Ekici, A. (2013) An improved model for supplier selection under capacity constraints and
multiple criteria, International Journal of Production Economics, 141, 574–81.
Fiddis, C. (1997) Manufacturer-retailing Relationships in the Food and Drink Industry:
Strategies and Tactics in the Battle for Power, Parson Publishing.
Friesz, T.L. and Kwon, C. (2007) Strategic freight network planning models and dynamic
oligopolistic urban freight networks, in D.A. Hensher and K.J. Button (eds), Handbook
of Transport Modelling, 2nd edn, Pergamon.
Guan, D., Wang, D., Hallegatte, S., Davis, S.J., Huo, J., Li, S., Bai, Y., Lei, T., Xue, Q.,
Coffman, D., Cheng, D., Chen, P., Liang, X., Xu, B., Lu, X., Wang, S., Hubacek,
K., and Gong, P. (2020) Global supply-chain effects of COVID-19 control measures,
Nature Human Behavior, 4, 577–87.
Hancock, K.L. (2001) Hazardous goods, in A.M. Brewer, K.J. Button, and D.A. Hensher
(eds), Handbook of Logistics and Supply-chain Management, Pergamon.
Hertz, S. and Alfredsson, M. (2003) Strategic development of third-party logistics provid-
ers, Industrial Marketing Management, 32, 139–49.
Kasarda, J.D., Green, J., and Sullivan, D. (2004) Air cargo: engine of economic develop-
ment, Center for Air Commerce, Report, University of North Carolina at Chapel Hill.
Knight, F.H. (1921) Risk, Uncertainty and Profit, Houghton Mifflin.
LaTourrette, T., Howell, D.R., Mosher, D.E., and MacDonald, J. (2006) Reducing terror-
ism risk at shopping centers: an analysis of potential security options, Report, RAND
Corporation.
Martin, D., Cannell, W., and Gwilliam, K. (1995) Reducing the impact of freight transport
on global warming: the potential of technical solutions, World Transport Policy, and
Practice, 1, 11–17.
McKinnon, A.C. (2003) Logistics and the environment, in D.A. Hensher and K.J. Button
(eds), Handbook of Transport and the Environment, Elsevier.
Meixell, M.J. and Gargeya, V.B. (2005) Global supply chain design: a literature review and
critique, Transportation Research Part E, 41, 531–50.

BUTTON 9781786435668 PRINT.indd 396 27/04/2022 11:07

 ECONOMICS AND TRANSPORT LOGISTICS ­ 397

Mitchell, A. (1997) Efficient Consumer Response: A New Paradigm for the European
FMGG Sector, Pearson Publishing.
Rodrigue, J-.P., Slack, B., and Comtois, C. (2001) Green logistic, in A.M. Brewer, K.J.
Button, and D.A. Hensher (eds), Handbook of Logistics and Supply-chain Management,
Pergamon.
Selod, H. and Soumahoro, S. (2020) Big Data in Transportation: An Economics Perspective,
World Bank.
UK Department of Transport (1979) Road haulage operators’ licensing report of the inde-
pendent committee of enquiry into road haulage operators’ licensing, Report, HMSO.
Wang, G., Gunasekaran, A., Ngai, E.W.T., and Papadopoulos., T. (2016) Big data analyt-
ics in logistics and supply chain management: certain investigations for research and
applications, International Journal of Production Economics, 176, 98–110.
Warne, J. and Rowney, M.J. (1986) The Toyota just in time production system and early
experience with ZIPS (Zero Inventory Production System), Report, TTC.
Waters, W.G. and Yu, C. (2003) Air security fees and highway safety, Proceedings of the
38th Canadian Transportation Research Forum Annual Conference.

BUTTON 9781786435668 PRINT.indd 397 27/04/2022 11:07

 11 Investment Criteria: Private and
Public Sector Analysis

11.1 Transport and Infrastructure

Adam Smith in his classic book, The Wealth of Nations, certainly thought that
transport in 1776 had a major and positive role to play in fostering economic
well-being, and spent considerable time explaining this, and that government
has an important responsibility in ensuring adequate transport infrastructure is
provided:

The third and last duty of the sovereign or commonwealth is that of erecting and
maintaining those public institutions and those public works, which, though they
may be in the highest degree advantageous to a great society, are, however, of such a
nature that the profit could never repay the expense to any individual or small number
of individuals, and which it therefore cannot be expected that any individual or small
number of individuals should erect or maintain.

The preceding chapters have been primarily concerned with making the best
use of an existing transport network or fleet of vehicles. They have, therefore,
principally focused on short-term problems involving the management, regula-
tion, and pricing of an established transport system. They were concerned with
emphasizing the central role of marginal cost pricing (including social costs) in
encouraging the optimal utilization of transport facilities. The discussions have
thus largely taken the stock of transport infrastructure as given and been con-
cerned with making the most efficient economic use of it.
There is, however, a longer-term aspect to be considered, namely possible
changes in the size or nature of the basic transport system by either investment
or disinvestment. (The latter may take the form of allowing quality deterioration
by reduced maintenance of a transport network rather than physical removal of a
link.) In the case of road haulage, and airline and shipping operations, the com-
mercial nature of decision-making bodies means that changes are normally ana-
lysed in terms of their financial repercussions. With road track, railways, and port
authorities, which in most countries are owned by public agencies, the provision
of basic infrastructure is usually determined by looking at much wider economic
considerations.
Before moving on to look at some of the key techniques employed in invest-
ment analysis, it is helpful to consider the scale of transport infrastructure that
exists, and to provide some information on variations in modal provision and
technology. We have already seen in Chapter 2 the considerable amount of

BUTTON 9781786435668 PRINT.indd 398 27/04/2022 11:07

 399
INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­

transport infrastructure – roads, railway track, etc. – that has been developed over
the years; basically, the stock of network capital. It is also clear from what we saw
in Tables 2.6 and 2.7 that road infrastructure provision far exceeds that of rail in
most of the larger, wealthier industrial nations. This stock is also growing. The
United States railroad network, for example, expanded from 178,000 kilometers
in 1993 to 202,500 in 2020, that in Japan from 20,300 to 27,300, and China’s from
59,000 to 146,000. Perhaps more pronounced has been the investment in road
networks.
The global situation is, however, variable. The picture is, for example, slightly
different in the former communist states in Europe (and even more so in many
less developed countries) that have relatively extensive rail networks vis-à-vis their
road systems, although things are changing with developments in logistics and
higher levels of car ownership.
In addition, it is not simply a matter of the quantity of the capital stock; its
quality is also of considerable importance. The market-based economies of the
Organisation for Economic Co-operation and Development (OECD) states, for
example, recorded 63 percent of the Netherlands’ rail network as being double-
tracked, compared with 69 percent of the United Kingdom’s and 45 percent
of France’s, while Bulgaria and Hungary only had 23 percent double-tracking,
and Poland just 15 percent. Taking another parameter, while there are consider-
able variations in the electrification of rail networks in industrialized countries
because of the type of traffic they carry, and the freight/passenger mix, they
generally still have more extensive systems than in the post-communist states. For
example, out of 31,000 kilometers of line, nearly 12,000 were electrified in the
western states of pre-1990 Germany, but out of 14,024 kilometers in the former
East Germany only 3,475 kilometers were electrified, and in Poland, of 26,545
kilometers only 6,296 were electrified. With political and economic reforms,
Poland now has about 11,900 kilometers electrified, and Hungary has increased
its electrified system from 2,300 to 2,800 kilometers.
There have always been significant variations between countries in terms of
the proportion of national resources that are invested in transport infrastructure.
This can be explained in terms of different geographies and industrial needs.
While the capital stock of transport infrastructure is large and huge sums
are spent on its expansion and maintenance, the evidence is that investment in
transport, and especially rail transport, tended to slow down in the latter part
of the twentieth century compared to the 1960s and 1970s, and, in some cases,
disinvestment may well have taken place. In the case of rail, this manifested itself
as a 4.1 percent reduction in the rail network of OECD countries between 1970
and 1985, although effective capacity may well have increased as electrification
and double-tracking programs were carried through. One reason for this was
that many networks, such as the United States Interstate highway system, neared
completion, and most industrialized countries had extensive transport networks
in place by the 1990s. This trend has been reversed somewhat (Table 11.1) during
the first decades of the twenty-first century, although the pattern still varies across
countries.

BUTTON 9781786435668 PRINT.indd 399 27/04/2022 11:07

 400 TRANSPORT ECONOMICS, 4TH EDITION

Table 11.1 Investment in transport infrastructure in selected countries (€ millions)

Country Infrastructure investment as % of GDP

Rail Road

2016 2019 2016 2019

Canada 796.5 1,366.3 7,042.7 7,543.2

France 8,614.3 11,630.0 9,216.0 9,816.0
Germany 5,192.0 6,948.0 12,870.0 16,650.0
Italy 3,847.0 2,855.0a 3,755.0 6,555.0a
Japan 9,174.7 n.a. 32,575.9 30,027.2b
United Kingdom 13,511.1 13,298.4 8,561.4 9,642.6
United States 12,473.3 11,587.1 85,589.4 93,962.2

Notes: a. 2018. b. 2017.

n.a. = not available.

Source: Organisation for Economic Co-operation and Development.

The growth in globalization and the removal of trade barriers at the macro level,
and the spread of urbanization at the more local levels, added impetus to the
notion that additional capacity was needed in the early part of the twenty-first
century. The changes in Europe, and the Trans-European Networks (TENs) ini-
tiative of the European Union which we discuss in Chapter 13, can also be likened
to the development of transport networks of the kind fostered in Canada and the
United States in the nineteenth century as a tool for political integration.
Economists look at transport investment from two broad perspectives. At the
macro and meso levels there is interest in the ways and extent to which it contrib-
utes to the economic advancement of a region or a country, a topic we deal with
later in the book. In the 1930s, for example, the construction of the autobahns in
Germany were used as a way to stimulate employment, and the construction of
the Federal Highway System in the United States was a major macroeconomic
stimulus. More recently, the American Recovery and Reinvestment Act of 2009
had a significant transport investment component to it, as has the macroeconomic
policy following the Covid-19 pandemic. At the microeconomic level, which is our
main concern here, the interest is in the efficiency of individual transport invest-
ments, and the returns that they generate in either narrow financial terms or in
wider, social terms. Project appraisal is necessary because investment resources
are not infinite and there are many potential options in terms of the ways in which
they may be used. Looking at the efficient use of these resources is our focus here.

11.2 Basic Theories of Investment Policies

Simple economic theory provides straightforward guidelines for investment

­decision-making; essentially, it involves pricing and output decisions where the
constraints of a fixed production capacity (for example, a given fleet or rail

BUTTON 9781786435668 PRINT.indd 400 27/04/2022 11:07

 401
INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­

network) cease to be binding. This is something that was introduced in Chapter 5

when considering long-run costs.
In Figure 11.1, which is a more complete picture of the cost curves shown
in Figure 5.3, we consider an example of a profit-maximizing airline with a fleet
exhibiting short-run average and marginal cost characteristics of SRAC1 and
SRMC1 respectively, and confronted by the demand curve D (with marginal
revenue MR). Ideally, a price P1 will be set and seat-kilometers Q1 offered. The
long-run marginal cost (LRMC) is, however, below marginal revenue at this
output, and, with this size of fleet, gives an inducement to expand output in the
long term by acquiring more capacity.
Greater seat availability will, however, force prices down, but – and this
is important in the example – it may also make it more economical to increase
the aircraft fleet size. In the diagram, and assuming throughout that profit-­
maximizing prices are charged, the fleet could be expanded to correspond to
SRAC2 and SRMC2. Here the long-run optimum situation is achieved with
marginal revenue equated to long-run and short-run costs and with profit maxi-
mization resulting. If the firm were concerned with social rather than commercial
profit-maximization criteria and adopted instead marginal cost pricing, then the
SRMC3 and SRAC3 curves become relevant because with this objective function
it is the setting of D = LRMC = SRMC which is important. Greater capacity is
still needed for this (hence the SRMC3 and SRAC3 curves) and optimal long-run
output will be higher at Q3 seat-miles with fares lowered to P3. At this price and
output, social surplus is maximized, although, since P3 < LRAC, in the long run
a financial loss will be made.
The basic principles are simple, and come straight from the elementary eco-
nomic literature; the difficulty comes in the transport context in putting them into
$ D
P1
MR
P2
LRAC

SRAC1

SRMC1 SRAC2

LRMC

SRMC2 SRAC3
P3
SRMC3

0 Q1 Q2 Q3 Traffic flow

Figure 11.1 Optimal investment: profit maximization and social surplus maximization

BUTTON 9781786435668 PRINT.indd 401 27/04/2022 11:07

 402 TRANSPORT ECONOMICS, 4TH EDITION

practice. In many cases investments are not divisible and, hence, the LRAC and
LRMC curves are disjointed segments, or even points, which do not intersect with
demand. This is an extremely common situation in transport and it does pose
serious problems in many operational cases.
It is not difficult, for example, to envisage routes where the available vehi-
cles (be they planes, buses, or whatever) are either too small or too large to be
optimal, and it is even more common in the case of infrastructure where, for
instance, a two-lane motorway may be inadequate to cope with normal demand
but a three-lane one is too capacious; you also must build a complete runway at
an airport because planning to construct 20 percent of one does not make a great
deal of sense. Further, there is the problem of what exactly is meant by ‘cost’; is it
simply the costs borne by the investor, or, because of market imperfections such
as externalities, are there other costs that should be taken into account? There
is also the matter of how inclusive the number of alternative investments to be
considered is; in many cases there is a wide variety of alternative ways in which
resources can be invested and selection processes can become important (Mackie
and Simon, 1998).
While we have treated the commercial and social criteria of profit maximiza-
tion and marginal cost pricing as amenable to presentation on one diagram, in
practice most socially orientated undertakings look at a much wider range of
costs (notably many of the externalities discussed in Chapter 6) when deciding
upon investment than do those motivated by purely financial considerations.
Coupled with this is the fact that the diagrammatic analysis assumes that, irre-
spective of the operational criteria, prices are optimal in the short term and thus
can act as an aid and guideline to investment decision-making. Also, despite the
sophistication of forecasting techniques (see Chapter 12) it is unlikely that the
transport provider is completely aware of the exact form of the long-run demand
curve confronting it; and may have a very poor idea of the full costs of con-
structing new facilities. Indeed, the fluctuating nature of demand for transport
(especially long-term cycles in demand associated with national and international
economic conditions) means that it is rather more of a stochastic concept than a
deterministic phenomenon as depicted.
There is the added institutional problem that transport investments are
seldom made in isolation from other changes to the transport system. For
example, larger suppliers of transport services in the private sector that have the
potential to exercise monopoly power are generally subjected to economic regula-
tions to prevent such exploitation. Regulatory policies regarding external costs
may also change during the construction of a large project. The result is that in
practice a sort of ‘muddling through’, to use Charles Lindblom’s (1959) jargon,
turns out to be the reality with continual modifications being made to the project
even after its construction has begun.
Given all these difficulties, together with the general inadequacy of informa-
tion enjoyed by most transport suppliers of their current levels of cost, let alone
future costs, it is not surprising that investment analysis in transport has received
considerable interest. The high costs and long-term implications of infrastructure

BUTTON 9781786435668 PRINT.indd 402 27/04/2022 11:07

 403
INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­

investments in road and rail track, and sea- and airports, has led to attention
being directed at these areas. At the academic level they also pose difficult ques-
tions because, in many cases, facilities are provided at prices unrelated to cost, or
made freely available to users. Additionally, there are frequently widespread rami-
fications for transport users elsewhere or for the non-users living in surrounding
areas.

11.3 Commercial and Social Approaches to Investment

Transport infrastructure is provided in a variety of institutional settings, ranging

from purely commercial structures to ones that are totally socially oriented,
and is financed in many ways, ranging from user fees to taxes. Just considering
air traffic control systems, the United States Federal Aviation Administration
(FAA) is financed largely from taxes and has no strict financial remit, whereas
NAV Canada is a non-profit, private corporation that levies user fees, the
Australian air traffic control body is a government corporation financed from
user fees, and the United Kingdom’s National Air Traffic Services (NATS) is a
price-regulated, for-profit, public–private partnership (PPP) that imposes user
fees (Button and McDougall, 2006). But there is also diversity within countries,
and sectors within them.
The administrative structure of transport in many countries means that most
types of infrastructure are supplied with the intention of maximizing overall eco-
nomic efficiency – that is, they are appraised in terms of their social value, assum-
ing they are optimally utilized at marginal cost prices. There are clearly exceptions
to this where profit maximization is seen as the primary objective, as with the
United States railroads. Even when conducting assessment of a single project, or
a series of closely related ones, the complexity of funding mechanisms can mean
that a combination of techniques are adopted to meet the requirements of differ-
ent stock-holders and this can even involve different approaches by various levels
of government that have a financial stake in the project A simple review of the
ways in which highways are financed in the United States (Table 11.2) highlights
the complexity of the problem.

Financial Appraisal

Historically a lot of transport infrastructure has been provided by the private

sector, and even some that has been publicly provided has been done using com-
mercial criteria. Indeed, there are arguments that there are cycles in which the
private sector, and, ipso facto, financial considerations, have played a greater or
lesser role in transport investments. While this may apply when there are new
modes emerging, for example the railroads taking over from canals for the move-
ment of freight in the mid-nineteenth century, there have also been shifts of
emphasis within modes. An example is the private engagement in turnpike invest-
ments in the eighteenth and early nineteenth centuries, reverting to government

BUTTON 9781786435668 PRINT.indd 403 27/04/2022 11:07

 404 TRANSPORT ECONOMICS, 4TH EDITION

Table 11.2 Sponsors and features of highway financing in the United States

Sponsor Major features of financing Examples

Private equity Finance and develop the project using Dulles Greenway (VA)
investors private resources 91 express lanes (CA)
Private, non-profit Issues tax-exempt debt backed by tolls TH 212 (MN)
entity (and without recourse to taxes) and Southern Connector (SC)
oversees the project under the terms of Interstate 895 (VA)
the agreement between the state and a Tacoma Narrows Bridge
private developer (WA)
Arizona toll project (AV)
Special-purpose Issues tax-exempt debt backed by tolls E-470 (CO)
public (and without recourse to taxes) and Orange County
oversees the project under the terms of transportation
the agreement with private developer corridor agency (CA)
State agency Issues tax-exempt dept backed by tolls Some turnpikes
(and without recourse to taxes)
State agency Issues tax-exempt dept backed Most highway projects
by taxes that are financed by debt
State agency Finances highways on a pay-as-you-go Most highways
basis using states taxes and fees plus
federal aid

Source: US Congressional Budget Office (1998).

investments in highways in the early twentieth century, to the concessions for

private investors that became common in the 1990s.
Differences in the perspectives of the investors depend largely on whether the
focus is on the stock-holding (concerning only the people who make the invest-
ment) or the stake-holding (embracing those who not only put the finance into the
system but also are affected by its larger outcomes). The former focuses primarily
on commercial criteria and the financial returns that an investment can generate,
and the latter on wider social criteria.
In technical terms the simple differences between the basic commercial
and social welfare-maximizing approaches are seen in Figure 11.2. The profit
maximizer is only interested in ‘producer surplus’, the difference between
money paid out to make the investment and the subsequent revenue earned,
whereas the broader approach takes account of both the ‘producer surplus’ and
the ‘consumer surplus’, the amount society would have been willing to pay for
the investment beyond the costs incurred, the combination forming the social
surplus.
The distinction between the two approaches may be seen in more detail by
contrasting the discounting approach used by large profit-orientated firms (and
public corporations instructed to operate commercially) with the discounting
approach of undertakings concerned with economic efficiency. (The discounting
process is a simple weighting of different items of cost and income according to
the time period at which they occur, more distant items being given less emphasis
in the calculations.)

BUTTON 9781786435668 PRINT.indd 404 27/04/2022 11:07

 INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­405

$
Consumer
surplus
Cost

Producer Demand
surplus

0 F Traffic flow

Figure 11.2 The simple difference between the financial and social cost–benefit approaches

More formally, a commercially motivated firm will, in the absence of a budget

constraint, accept investments when the financial net present value is p
­ ositive;
that is:
t =k
 P (Rt ) − P (Ct ) 
NPVP = ∑   (11.1)
t =1  (1 + r )t 
where:

NPVP is the financial net present value;

P(Rt) is the probable revenue that would be earned in year t from the
investment;
P(Ct) is the probable financial cost of the investment in year t;
r is the rate of interest reflecting the cost of capital to the undertaking; and
k is the anticipated life of the investment.

A positive NPVP, therefore, tells the businessperson that it is worthwhile

undertaking an initial investment, that is, it tells them that a movement from
zero output in Figure 11.1 to output Q1 with the associated short-run costs of
SRAC1 and SRMC1 is commercially desirable and that a profit above both long-
and short-run costs will be earned. A more normal case, where an expansion of
operations is being considered involving some new capital outlay, requires the
additional discounted profits from the investment to be compared with the addi-
tional discounted costs. If the resultant incremental NPVp is positive, then the
investment is justified on profitability grounds. In terms of the diagram, a move-
ment down the LRMC curve to output level Q2 with the short-run marginal costs
of SRMC2 would yield a positive incremental NPVp but subsequent investment to
take one down the LRMC curve to output Q3 would not.

BUTTON 9781786435668 PRINT.indd 405 27/04/2022 11:07

 406 TRANSPORT ECONOMICS, 4TH EDITION

The Cost–Benefit Approach

In contrast to the stock-holder approach, investments involving the public sector

generally involve economic efficiency being assessed using some form of cost–
benefit (or ‘benefit–cost’) analysis (CBA) that, again in the absence of a budget
constraint, suggests that schemes with a positive social net present value should be
undertaken where:
G=x T = K  
aP( BGT ) − bP( LGT )
NPVSW = ∑ ∑   (11.2)
( )
T
G =1 t =1  1+ i 
 
where:

NPVSW is the social net present value;

amP(BGT) is the probable social benefit to be enjoyed by individual m in year
  T as a result of the investment’s completion, and BGT is given a weighting
a, to reflect society’s welfare preference;
bmP(LGT) is the probable social cost imposed on individual m in year T as
 a result of the investment’s completion, and LGT is given a weighting b, to
reflect society’s welfare preference;
i is the relative social weight attached to a cost or benefit occurring each year;
K is the anticipated life of the investment; and
G is the number of individuals affected.

Again, in terms of Figure 11.1, a positive NPV implies that the social surplus
associated with an investment exceeds the discounted costs – that is, the demand
curve at the final output is equal to or above the LRMC curve. An additional
investment will be economically justified if the discounted value of incremental
social benefit exceeds incremental costs. Contrasting this with the commercial
criteria, the NPVp associated with moving down the LRMC curve from output
Q2 to Q3 is negative but the incremental NPV would be calculated to be positive.
Not only is the cost–benefit type of analysis more comprehensive in terms of
the items considered, but it also redefines many of the items retained from com-
mercial criteria. For example, the costs of imported raw materials used in a poten-
tial road construction project in a Third World country would be valued at market
prices if a commercial undertaking were responsible for road investment decisions.
If a public body undertakes road investment using wider social criteria, then
it would look beyond the immediate financial indicators and at the ‘shadow’
prices of imports so that the scarcity of foreign exchange and the limitations of
adequate finance for imports is reflected in the decision-making. In some invest-
ments use is made of formerly unemployed factor services – for example, unem-
ployed labor where the opportunity cost of employment in a transport scheme is
really zero or the opportunity cost of the leisure they now forgo. A commercial
concern would cost such inputs at the wages that have to be paid, but in a cost–
benefit study they may not be considered a cost at all, or, more probably, would be
costed so that genuine resource costs are incorporated in the calculations.

BUTTON 9781786435668 PRINT.indd 406 27/04/2022 11:07

 407
INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­

A further very important distinction is that the social efficiency approach

takes cognizance of the distributional effect of the investment (the am and bm
terms in equation (11.2)). This is often difficult to do although various schemes
for weighting costs and benefits have been advanced by theoreticians. In practice
there is a tendency to employ rather crude methods. Often, as in the case of the
planning balance sheet approach discussed below, such methods have been used
in urban infrastructure investment appraisals, which involve the simple setting
out in tabular form of the impacts of a scheme on the different user and non-
user groups affected, or, as with the COBA inter-urban road appraisal methods
previously used in the United Kingdom, which carried out a partial CBA with
no allowance for distributional effects and subjected the results of this to further
debate at public inquiry.
In other cases, the social analysis may not be complete – it may focus only on
the internal social efficiency of the transport system and ignore its wider implica-
tions on the environment, etc. In an evaluation, a new mode of high-speed ground
transport, the magnetic levitation rail system (Maglev), Elhorst and Oosterhaven
(2008) compare the results of an integral CBA that embraces environmental and
wider economic effects with those of a conventional partial CBA that only looks
at transport effects, and conclude that the additional economic benefits due to
market imperfections vary from –1 percent to 38 percent of the direct transport
benefits, depending on the type of regions connected and the general condition
of the economy. Initially, however, it is useful to explain in a little more detail
the broad differences between a financial appraisal and a social analysis of an
investment.

11.4 Public–Private Partnerships

Public–private partnerships (PPPs) are a well-established part of transport policy.

PPPs involve a government, or its agent, signing an agreement with a private
company or consortium to supply it with services, with the private sector actor
being involved in various elements of designing, building, temporarily ‘owning’,
and running the physical assets. PPPs are long-term development and service
contracts between government and a private partner. The European Union,
for example, supports PPP initiatives by channeling funds through the Loan
Guarantee Instrument for Trans-European Transport Network Projects. This
consists of a European Investment Bank guarantee for subordinate debt in the
form of a stand-by liquidity facility to be provided by commercial banks.
PPPs have a relatively long history. Their modern origins are often traced to
the construction of Amsterdam town hall in 1662, and PPPs existed in France
in 1666 with their use in the building of the Canal du Midi in Toulouse. In the
United States, they date back to the Philadelphia and Lancaster Turnpike in
Pennsylvania, which was chartered in 1792. The global upsurge in interest in
them came in the 1970s when there was a focus on accelerating infrastructure
­investment, with transport forming a large part of the initiative.

BUTTON 9781786435668 PRINT.indd 407 27/04/2022 11:07

 408 TRANSPORT ECONOMICS, 4TH EDITION

PPPs were seen as part of an alternative service delivery mechanism

designed to provide new ways of providing public services. Initially, they largely
involved contractual arrangements between the private and public sectors to
provide and operate relatively small-scale physical infrastructure (roads, sea-
ports, etc.), but this situation broadened out somewhat in the late 1990s to
embrace information technology and social facilities. The widening was partly
due to concerns about the problems that straightforward privatization of large
scale had encountered in places like Latin America. The public sector saw PPPs
as a way of ­accelerating investment beyond traditional tax and borrowing chan-
nels, while the private sector saw it as a way of spreading risk. A listing of some
of the main heavy transport PPPs that were initiated is seen in Table 11.3 (see
also Button, 2016).
These differences in the perspectives reflect varying interests between stock-
holders (concerning only the people who make the investment – the private sector
element of a PPP) and stake-holdings (embracing those not only putting finance
into the system but also affected by its larger outcomes: the public more gener-
ally). The former focus on commercial criteria, and the financial returns that an
investment can generate, and the latter on wider social criteria – a subject covered
more rigorously below. In welfare economic terms, a profit-maximizing private
investor is only interested in producer surplus, the difference between money
paid out to make the investment, and the subsequent revenue earned, whereas the
broader approach takes account of both producers’ and consumers’ surpluses, the
latter being the amount society would have been willing to pay for the investment
beyond the costs incurred.
A major motivation for the private sector engaging with the public is to
reduce their risk. The private sector pays more in interest, a risk premium, to raise
capital because of the possibility of it defaulting. The public sector, because of its
power to tax, does not have the same problem. This leads, with other factors, to a
series of problems of drawing up contracts between the two parties (Hart, 2003):

• The nature of partnerships. Virtually all PPPs involve a degree of bi-lateral

oligopoly and often, where highly specialized infrastructure is concerned,
bi-lateral monopoly. In terms of large-scale investments, and the limited
number of private sector investors capable of financing such projects,
the negotiations underlying a PPP involve the exercise of market power.
Conventional, neoclassical economic ideas of competitive equilibrium, the
outcome of a bi-lateral monopoly or oligopoly situation, as discussed in
Chapter 7, are normally indeterminate, depending on the bargaining skills
of the parties involved.
In many cases, the private sector developer (or a strong alliance of devel-
opers), by dint of size and specialism, have considerable monopsony power,
while the government agency is confronted with competition for its scarce
resources from alternative uses. In cases where the public sector can foster
competition between private developers for the market, the bargaining posi-
tion is reversed. The resultant price and output combination will, in either

BUTTON 9781786435668 PRINT.indd 408 27/04/2022 11:07

 INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­409

case, depend on the bargaining skills of the two parties, and that, of course,
will be a function of their respective game-playing abilities.
In general, however, whatever the outcome, the final cost of supplying
the investment to the economy will be higher than that which would emerge
with perfect competition on both sides of the market. Thus, irrespective of
the power of the private firms and the government agency involved, the final
consumer does not generally benefit greatly from a bi-lateral monopoly situa-
tion in terms of immediate costs. The gains are generally seen in acceleration
in investments.
• Initial contract. The main challenge with the initial contract concerns its
clarity and coverage (Hart, 2003). To be successful, a PPP must indicate the
roles, responsibilities, financial liabilities, etc., of both the public and private
partners.
• Boundary of contract. The degree to which unbundling (the passing over to
the private sector of traditional public sector responsibilities) occurs can be
looked at in a number of ways, and indeed the ways it can be carried out
are also numerous. They may involve design, build, finance, operate, and
maintain (DBFOM) concessions, or design, build, finance, and maintain
(DBFM) concessions, or build, own, operate, and transfer (BOOT) conces-
sions, and there are other variants. Additionally, details within each PPP, of
any type, can vary considerably. There are also various categories of leasing
that may be seen as PPPs, and the financial commitments of the two sectors
can take a variety of forms.
From a practical perspective, there are few independent elements of
infrastructure, and there are also often interdependencies involving the
investment in and operation of facilities. This raises questions about the
appropriate boundaries of any PPP: which elements should be public and
which private? For example, should the investment be public and the main-
tenance of a road be private? Forms of concessions can also differ in terms
of the structure of the market imposed. In the context of South American
airports, for example, there have been diverse approaches to concessions
with some countries combining a number of airports within a concession
package, while others have separate PPPs for each airport (Button, 2008).
Much depends upon the extent to which the public sector believes there are
economies of scope and density in having an integrated airport system as
opposed to airports competing with each other.
Many concessions involve commitments to build and finance, as well
as manage, existing infrastructure. The decisions regarding the optimal
approach in such cases have as much to do with the inherent nature of trans-
action costs within firms as they have to do with their method of provision.
It is essentially a practical matter. If there are managerial synergies or high
costs of unplanned coordination, then there are arguments for grouping
activities in a PPP arrangement. The transaction costs of many standalone
entities are higher than those when they are bundled. In terms of unbun-
dling infrastructure investment from its subsequent operations, a situation

BUTTON 9781786435668 PRINT.indd 409 27/04/2022 11:07

 Table 11.3 Examples of heavy rail infrastructure public–private partnership projects

BUTTON 9781786435668 PRINT.indd 410

Project Design/ construction Concession Route length Investment Public funds Contract
period period (years) (kms) costs type

Channel Tunnel rail link 1966–2003/2007 90 109 £5.8 billion £2,010 million DBFM
Öresund road–rail link 1991–2000 25–30 38 €2 billion 100% state guarantee DBFM
HSL-Zuid Amsterdam to Belgium 2000–2007 25 100 €6 billion €110 million per year DBFM
Alice Springs–Darwin rail 2000–2004 – – A$1.2 billion A$559 million –
Taipei–Kaohsiung high-speed rail 2000–2007 35 335 US$18 billion – DBFOM
Perpignan–Figueres high-speed rail 2005–2009 50 45 €1.1 billion €540 million + €62 million per year DBFM
Diabolo rail link Brussels 2007–2012 35 3 €540 million €250 million + track & signaling DBF
Liefkenshoek rail link Antwerp 2008–2013 38 16 €840 million €50 million per year DBFM
Tours–Bordeaux high-speed rail 2010–2016 44 340 €7.2 billion < 50% DBFM
GSM-R network France 2009–2015 15 14,000 €1 billion €160 million DBFOM
Lisbon–Madrid high-speed rail 2009–2013 40 165 €7.8 billion – DBFM
Nimes–Montpellier high-speed rail 2011–2016 – 80 €1.62 billion DBFM
Brittany–Pays de Loire high-speed 2011–2017 – 182 €2.85 billion – DBFM
rail
Rio de Janeiro–São Paulo– 2011–2017 40 – US$18.7 billion – DBOF
Campinas high-speed rail
Ylivieska rail link 2011–2014 20–30 76 €660 million – DBFM

Notes: DBFM = design, build, finance, and maintain. DBOF = design, build, operate, and finance. DBFOM = design, build, finance, operate, and maintain.
DBF = design, build, and finance.

Source: Hanson (2011).

27/04/2022 11:07
 411
INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­

more akin to conventional provision, again transactions costs are involved,

but in this case there is the matter of whether better-designed infrastructure
reduces operating costs; essentially interactions are viewed sequentially
rather than at the same time.
• Honesty. Corruption may take several forms regarding PPPs. In terms of
the private sector, corruption can stem from the large incentive required to
pull in the necessary private resources. In terms of the public, corruption
is more in terms of economic capture. Officials and public servants subju-
gate the public’s interest to those of their private interests, although overt
breaking of laws is also present in some cases. Often, this type of corruption
entails governments making inefficient decisions to improve the chances of
re-election, and in particular continuing or bailing out projects that should
be terminated. One reason that the issue of pet projects can occur, making
simple comparisons between public and private ownership, is that the public
sector is generally intrinsically more complex than the private, although
the situation can vary on a case-by-case basis. Because of this, it is hard to
make rules regarding ownership. To contain these sorts of problems, and
in particular those of cost over-runs as officials seek to complete their ‘pet’
projects, fixed-price contracts for the private party and a budget cap for the
public financing can temper political misbehavior by allowing scrutiny of
public liabilities.
• Renegotiation. Regarding transportation, Guasch et al. (2008) found that 53
percent of concessions awarded in Latin American countries during 1989–
2000 required renegotiation, and on average this was done just 3.1 years after
an award. Vining et al. (2005) provide some North American transporta-
tion examples of the detailed issues that can lead to renegotiations and the
various costs involved.
There are often quite legitimate reasons why a PPP breaks down once
physical construction has begun; for example, unexpected rises in costs, or
the encountering of unforeseen technical problems; basically the contract is
incomplete. In these cases, there are often justifications continuing with the
project; these may be centered in terms of bygones-being-bygones and the
marginal gains to both parties of the additional costs are justified, or because
of high transaction costs in seeking new partners to continue the work after
relationship-specific investments have been made. In other cases, however, a
private contract may ‘low-ball’ a contract by making an exceptionally low
offer to contract for the work with the expectation that it can subsequently
renegotiate at a higher price. The challenge is then to find the appropriate
level of flexibility: too much, and undesirable opportunistic renegotiations
are likely to be necessary; too little, and opportunities for welfare-enhancing
renegotiations will be lost.
Long-term contracts can be one way of circumventing some of these
problems, especially if the private actor can reduce its risks by having long-
term contracts with its suppliers: a form of hedging. But, given the nature
of risk, defining comprehensive contracts that allow for all contingencies is

BUTTON 9781786435668 PRINT.indd 411 27/04/2022 11:07

 412 TRANSPORT ECONOMICS, 4TH EDITION

almost impossible. One way around rigid contractual structures, which can
minimize transaction costs but which sacrifice opportunities to make PPPs
more economically efficient, is by allocating and addressing future downside
risks appropriately, to make them more flexible. But even this can fail, thus
PPP contracts are seldom set in stone and renegotiation may be anticipated,
given the longevity of many concessions and the temporal variability of
demand.

11.5 The Theory of Cost–Benefit Analysis

While there are many complexities in undertaking commercial investment

appraisal (for example, allowing for risk of unexpected changes in demand,
deciding upon appropriate methods of raising capital, etc.), it is public sector
transport investment that has attracted the greatest attention. The wide-ranging
and long-term effects of most major changes in transport infrastructure neces-
sitate the employment of sophisticated methods of project appraisal and of
comprehensive techniques for decision-making. The underlying notion of CBA
that forms the explicit (and, on occasions, implicit) foundation for much of this
work has already been alluded to in previous sections. While the algebra set down
there suggests a comparatively simple set of standard calculations, the theoretical
model is itself based upon a set of much more complex assumptions which makes
the application a far more tortuous exercise than it might at first appear. Indeed,
there is evidence that the optimism once felt for CBA as the panacea for all trans-
port investment appraisal problems has gradually evaporated and the confidence
felt in the strength of CBA calculations no longer exists.
This and the following sections attempt, in broad terms, to explain the CBA
methodology and to point to recent innovations in theory and practice. The
subject matter of CBA has now become so vast that the treatment here must, by
necessity, be rather limited. We begin by looking at the basic welfare economics
underlying the technique, and the direction this is moving (Atkinson, 2011). This
highlights some of the key assumptions upon which it is based.
The simple outline of CBA in the previous section emphasized the notion
of selecting investments that maximize social surplus rather than just pecuniary
returns. One of the major problems in this is that of inter-personal comparisons
of welfare. Is it possible to say social welfare has risen if one group becomes better
off at the expense of another? This represents a common situation in transport
where users tend to benefit at the expense of non-users, or amongst users, travel-
ers using one mode gain at the expense of another. CBA attempts to circumvent
this conceptual problem by making use of ‘hypothetical compensation tests’.
Strictly, since we only have a notion of the ordinal ranking of individuals’ priori-
ties, inter-personal welfare comparisons can only be made in very limited circum-
stances. The Pareto criterion, which underlies most modern welfare economics,
states that an action can only definitely be said to be socially desirable if at least
one agent benefits and no one suffers any diminution of welfare.

BUTTON 9781786435668 PRINT.indd 412 27/04/2022 11:07

 413
INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­

Diagrammatically, in the upper portion of Figure 11.3 we have two individu-

als, A and B, who enjoy various levels of welfare recorded on the horizontal and
vertical axes as UA and UB. If we have a finite collection of goods and services
available (including transport services), together with a fixed level of costs, then
the well-being of A and B will depend upon how these goods, services, and costs
are distributed between them.
The utility possibility frontier represents the maximum possible welfare they
could enjoy given different distributions of the goods, services, and costs. Initially,
the goods etc. are distributed so that point X on the pre-investment frontier is
achieved (that is, A enjoys a utility of 0UA* and B of 0UB*). The goods, services,
and costs could be redistributed so that any other point on pre-investment could
be obtained but no point outside of it. Suppose now that a transport investment
results in the bundle of goods, services, and costs changing and that position Y
beyond the pre-investment frontier is reached. This would be deemed a Pareto
improvement because both A and B are better off. Indeed, any ex post position
within the 90° zone marked (such as Y) would be Pareto-superior to X since, even
on one of the limits, at least one person is better off while the other, at worst,
retains their original utility level.
The strict Pareto criterion is of little use in practice as most transport
schemes have either direct or indirect net adverse effects on some members of
the community. The suggested method of allowing for this, developed in the late
1930s, is to adopt a hypothetical Pareto criterion and decide whether, after the
investment, it would be possible to redistribute the impacts in such a way that
no one is worse off but there is still some residual gain to others. Whether such
redistribution occurs is felt to be a normative issue and should be treated as being
within the domain of politics rather than economics.
There are two broad approaches to the hypothetical compensation criteria.
The first, initially advanced by Nicholas Kaldor (1939), forms the basis for most
CBA studies and suggests that a scheme is socially desirable if the beneficiaries
could compensate the losers and remain better off. In the upper part of Figure
11.3, Y is Kaldor-superior to X (although not strictly Pareto-superior) because
Y is on the post-investment utility possibility frontier that passes outside of
X. Hence, the bundle of goods, services, and costs which generate the post-­
investment frontier could be redistributed from Y to Y*, which is Pareto-superior
to X.
Whether the government instigates a tax/subsidy scheme to cause such a
movement along the post-investment frontier is considered a political issue;
Kaldor (in a discussion of the repeal of the nineteenth-century Corn Laws!)
simply argues that it could be done. John Hicks (1940) favors a similar approach
but adopts pre-investment weights for his criteria. Specifically, he argues for a pro-
ject’s acceptance if those who lose as a result of its implementation cannot bribe
the gainers not to do it without becoming worse off themselves. In the figure, Y is
Hicks-superior to X because, with the pre-investment package of goods, services,
and costs that permitted the post-investment frontier to be attained, it is impos-
sible for A (the loser) to bribe B (the beneficiary) not to support the investment.

BUTTON 9781786435668 PRINT.indd 413 27/04/2022 11:07

 414 TRANSPORT ECONOMICS, 4TH EDITION

UA

U*A X Post-investment

Pre-investment

0 U*B UB

UA Z

Y*
W

X Post-investment

Pre-investment

0 U*B UB

Figure 11.3 The Hicks–Kaldor compensation criteria

The problem with the Kaldor and Hicks approaches is that they may, in some
circumstances, contradict one another. For example, in Figure 11.3, although
combination W is Kaldor-superior to X, we can see that X is Hicks-superior
to W. Further, even if one only used the Kaldor test and the investment was
completed and position W attained, it then becomes possible to show that it is,
again following the Kaldor criteria, socially beneficial to disinvest and return
to X. (Similar types of problems exist with the Hicks test.) Paul Samuelson
(1961) argues that the problem will always exist if the pre- and post-investment
utility possibility frontiers cross and that comparisons in such circumstances are
invalid.
This is an extremely restrictive view. Provided the two positions being com-
pared are on the same side of any intersection of the frontiers, then the two
criteria give consistent assessments and there are no problems of ‘reversibility’
(Skitovsky, 1941). Figure 11.3, for instance, shows a situation where Y meets both
the Hicks and the Kaldor hypothetical compensation tests and may, therefore,
be considered socially superior to X. There is also no question of advocating

BUTTON 9781786435668 PRINT.indd 414 27/04/2022 11:07

 415
INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­

reversibility once the investment leading to Y has been undertaken. Whether,

in practice, transport analysts need to test for these problems is an empirical
question that has, to date, been inadequately explored. At least one experienced
economist in the CBA field suggests that the Skitovsky criteria may be violated on
more occasions than is sometimes supposed.
A different problem may arise when appraising a series of piecemeal
investments. It is possible, because of the relative nature of consumer surplus
that underlies all these tests, for a series of small investments each to pass the
Skitovsky test but for the eventual outcome to be socially inferior to the initial
position. In Figure 11.4, if we have an initial position of E on the pre-investment
frontier, then an investment that permits F to be reached satisfies Skitovsky’s
criterion (and, indeed, Samuelson’s); further, a move from F to G, following addi-
tional investment, may be approved and, likewise, a subsequent move from G to
H, which again meets the Skitovsky test. However, despite the fact that we have
seen that E < F < G < H in terms of hypothetical compensation tests, it is clear
in Figure 11.4 that E is preferable, on Skitovsky grounds, to H. Each of the series
of small changes appears desirable, but the final, overall outcome leaves society
worse off.
If this problem was only a theoretical curio there would be no need for
concern, but unfortunately this does not seem to be the case. Numerous exam-
ples of piecemeal decision-making leading to a subsequent diminution of social
welfare can be cited in urban planning, but in the context of transport perhaps
the most worrying problem concerns the growth in car use since the Second
World War at the expense of urban public transport (Mishan, 1967). For sim-
plicity, we take a concentric-shaped city with employment concentrated in the

UA

IV

III

II
I
E
H

0 UB

Figure 11.4 The problem of assessing a series of small investments

BUTTON 9781786435668 PRINT.indd 415 27/04/2022 11:07

 416 TRANSPORT ECONOMICS, 4TH EDITION

center. The urban core is surrounded by a ring of residential estates. The analysis
is short-term and assumes that this land-use pattern is fixed. It is now possible to
define three phases:

• Phase I. All commuters have only one mode of transport available to them
and travel to work by means of public transport – taking ten minutes is the
norm.
• Phase II. One commuter buys a car and drives to work, taking five minutes,
leaving the other travelers unaffected by his/her action and still taking ten
minutes to reach work by public transport
• Phase III. Many commuters, observing the advantage enjoyed by the car
driver, begin to buy and use cars that, with the congestion they generate,
increase driving time to work to 15 minutes and, due to the impedance
caused by the cars, slows public transport so that commuters using this mode
now suffer a 25-minute journey. In the longer term, because of the techno-
logically unprogressive nature of public transport, the service may be with-
drawn (following the syndrome of few passengers ⇒ higher fares and poorer
service ⇒ even fewer passengers, etc.), leaving a choice of car purchase
or walking to work. The result is a ‘prisoner’s dilemma’ type of situation,
where individually each commuter would prefer the original situation, rather
than the new undesirable equilibrium, but cannot attain it by uni-lateral
action.

The example illustrates the difficulties which may arise as the result of deci-
sions based upon relative welfare measures: each commuter thought that he or
she would benefit by investing in a car because they did not take cognizance of
the whole set of decisions being made. Ideally, a CBA study should appraise all
the systems or sequences of potential investments other than assess individual
components of a program of events. The urban planning process, discussed in
Chapter 12, is an area where this is relevant.
While most CBA studies of transport projects have concentrated on effi-
ciency considerations, relying upon the hypothetical compensation criteria, it
may be seen as important that some allowance for distributional impact should be
incorporated. Specifically, it is argued that a project should only be accepted using
the hypothetical type of criteria if the outcome improves the income distribu-
tion. For example, in the lower part of Figure 11.1, we assume that an improved
income distribution means greater equality of welfare and thus corresponds to a
movement closer to the 45° line depicted. Thus, Y, on the post-investment policy
frontier, is both Skitovsky-superior to X and offers more equality because it is
closer to the 45°, equal utility, line. It is important to note that it is the outcome
that is being considered and not potential redistributed packages of the post-
investment collection of goods, services, and costs. As we see in Section 11.5,
there are several ways in which this distributional element may be incorporated
within CBA studies.

BUTTON 9781786435668 PRINT.indd 416 27/04/2022 11:07

 417
INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­

11.6 Coping with Network Effects

The discussion above applies to all CBA applications, but a difficulty of apply-
ing the technique to transport investment decisions is the need to incorporate
adequately the wide-ranging effects that a change in one part of the transport
system has on the rest of the network. Most transport infrastructure forms a
link in a much larger, interacting network and, consequently, changes in any one
link tend to affect demand on competitive and complementary links. Although
this sort of complexity exists for virtually all forms of transport, the problem
of assessing the overall effect on road transport of improving a single link has,
because of the dominance of this form of transport in modern society, attracted
most attention.
If there are two roads, one from X to Y and the other from X to Z, where Y
and Z are to some degree substitute destinations, then an improvement in route
XY will affect three groups. We will assume for simplicity that all demand curves
are linear and that the pre-investment traffic flows on XY and XZ are TXY and
(TXZ + R) respectively. The three groups of users to consider are then:

1 Existing users who remain on their original routes (that is, TXY and TXZ).
These will enjoy a gain in consumers’ surplus because those on route XY
will now be using a higher-quality facility while those on XZ will benefit
from reductions in demand for this route as some former users switch to
the improved XY. If this latter traffic, which has diverted from XZ to XY, is
denoted as R, then the total benefit to those remaining loyal to their initial
routes may be represented as:

TXY(C1 – C2) + TXZ(C*1 – C*2) (11.3)

where C1, C2, C*1, and C*2 are the pre- and post-investment costs by roads XY
and XZ respectively.
2 Generated traffic consisting of people who did not previously travel (that
is, GXY and GXZ). On average (given the linear demand curves), each of
these groups of new road users will benefit by half as much as existing, non-­
switching traffic. (Some will obviously be marginal trip-makers and only just
gain by making a trip, while others are intra-marginal and enjoy nearly as
much additional consumer surplus as the non-switchers.) The total benefit
of the investment to this group will thus be:

0.5GXY(C1 – C2) + 0.5GXZ(C*1 – C*2) (11.4)

3 Diverted traffic that switches from route XZ to route XY because of the

investment (that is, R). The switch, given the free choice situation open to
travelers, must leave this group better off – they would not have switched oth-
erwise – and the additional welfare they enjoy can be seen to equal half of the
difference in benefit between the cost reductions on the two routes, that is,

BUTTON 9781786435668 PRINT.indd 417 27/04/2022 11:07

 418 TRANSPORT ECONOMICS, 4TH EDITION

R{(C*1 – C*2) + 0.5[(C1 – C2) – (C*1 – C*2)]}

which may be reduced to:

0.5R[(C1 – C2) + (C*1 – C*2)] (11.5)

The total benefit of the investment is the summation of these three elements,
namely:

TB = TXY(C1 – C2) + TXZ(C*1 – C*2) + 0.5GXY(C1 – C2)

+ 0.5GXZ(C*1 – C*2) + 0.5R[(C1 – C2) + (C*1 – C*2)] (11.6)

Figure 11.5 shows this diagrammatically. The left-hand part of the figure
represents the supply and demand situations on route XY and the right-hand part
represents those on route XZ. On route XY we see that demand has increased the
supply of road space but that demand for its use has declined because the relative
generalized cost of using XZ changes as traffic diverts from it to the improved
facility.
Using the notation in the diagram, we know that QXY = TXY and that
Q'XY = (TXY + GXY + R), therefore:

0.5(QXY + Q′XY) = (TXY + 0.5GXY + 0.5R)

Similarly, since QXZ = (TXZ + R) and Q′XZ = (TXZ + GXZ) we know that:

0.5(QXZ + Q′XZ) = (TXY + 0.5GXY + 0.5R)

Substituting this into equation (11.6) we discover:

TB = [0.5(QXZ + Q′XZ)(C1 – C2) + 0.5(QXZ + Q′XZ)(C*1 – C*2)]

$ $
S1
W A S2
C1
S*1 = S*2
C*1 Y E
Z
C2
B C*2
D1 F D*1
D2 D*2

0 Q'XY QXY Traffic 0 Q'XZ QXZ Traffic

Traffic X => Y Traffic X => Z
Figure 11.5 Social benefits over a transport network

BUTTON 9781786435668 PRINT.indd 418 27/04/2022 11:07

 INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­ 419

or more generally, this is the oft-cited ‘rule of half’:

TB = 0.5 ∑ (Qn + Qn* )(Cn + Cn* )

(11.7)
N

This net benefit is equivalent to the shaded areas seen in Figure 11.5. The rule
of half can be applied to all transport schemes that interact with other compo-
nents of the transport system where demand curves are linear. (It must, however,
be used with a degree of circumspection when routes are complementary, where
demand for the non-improved links may shift to the right, but the broad principle
applies.)
The method of handling interdependencies outlined above was initially
developed in the late 1960s as part of the London Transportation Study, but it
does rely upon a rather strong implicit assumption that the income elasticities
of demand for routes XY and XZ are equal. The problem is that there are many
possible sequences in which the price changes on routes XY and XZ could follow;
each would yield a different level of aggregate social welfare. For instance, if the
chain of price changes is (C1, C*1) => (C2, C*1) => (C2, C*2) then the consumer
surplus gain in the diagram would be {(C1AXC2) + (C*1YFC*2)}. But if the
sequence is (C1, C*1) => (C1, C*2) => (C2, C*2) then the aggregate benefit would be
{(C1WBC2) + (C*1EXC*2)}.
The general measure set out in equation (11.7) assumes that the demand
fluctuations are linear in their own prices and with respect to cross-price effects; if
this is so, then the measure would give identical results to both the sequences out-
lined above (which would themselves yield identical benefit estimates). Whether
such assumptions are valid is debatable, but Foster and Neuberger (1974) argue
that any deviation is unlikely to be of any practical significance in practical
evaluation exercises. Certainly, given the other major difficulties of evaluation
and measurement, the ‘rule of half’ provides a robust and useful guide to the user
benefits of transport schemes.

11.7 Cost–Benefit Analysis in Practice and Variations on the

Theme

Our attention, to this point, has focused on the theoretical ideas and con-
cepts underlying CBA. We now turn to look more directly at its application
in transport fields. The equation set out above (equation (11.2)) gives a formal
mathematical definition of CBA, while Prest and Turvey (1965) give the verbal
counterpart:

CBA is a practical way of assessing the desirability of projects, where it is important

to take a long view (in the sense of looking at repercussions in the distant as well
as the immediate future) and a wide view (in the sense of allowing for side effects
of many kinds on many persons, industries, regions, etc.) that is, it implies the
­enumeration and evaluation of all the relevant costs and benefits.

BUTTON 9781786435668 PRINT.indd 419 27/04/2022 11:07

 420 TRANSPORT ECONOMICS, 4TH EDITION

CBA has, over the years, formed the basis of investment appraisal of many
major transport schemes in the United Kingdom (for example, the M1 motorway,
the Victoria Line underground railway, the Channel Tunnel, London’s system of
ringway or beltway urban motorways, and the siting of a third London airport)
and elsewhere. It has also become a tool in more routine decision-making, for
example, to assess railway social service subsidies in the late 1960s and as a com-
ponent of inter-urban road investment appraisal in the form of COBA and its
subsequent modification, New Approach to Appraisal (NATA).
The United Kingdom situation, however, is a specific case. Elsewhere, there
can be important variations in the forms of appraisal used to assess transport
projects, some more comprehensive and others less so. Table 11.4 offers a broad
overview of the differing nature of decision-making processes in several major
European countries adopted from the 1990s. Over-riding these when European
Union financing is involved are standard Union appraisal frameworks that have
to be applied. The United States, being a federal system, has both a manual
for conducting CBA for national highway investments as well as each state
having its own approach for allocating its own resources (US Federal Highway
Administration, 2020).
Major transport infrastructure appraisals, such as that conducted by the
Roskill Commission in the late 1960s when looking at possible sites for a new
London airport, have generally sought to provide a comprehensive economic
assessment. But many CBA studies are less ambitious. The COBA computer
program that was for many years employed as part of the inter-urban road
appraisal process in the United Kingdom, for example, emphasised consistency
of methodology over the comprehensive coverage of all costs and benefits. COBA
made use of traffic forecasts derived from a standardized procedure and used
them to compare discounted monetary valuations of travel time changes, varia-
tions in vehicle operating costs, and impacts on accident rates with the capital and
maintenance costs of a project. While this provided an indication of the costs and
benefits to traffic and the Exchequer, information on third-party effects, such as
environmental impacts, were, until the early 1990s, treated separately. They were

Table 11.4 The nature of transport infrastructure decision-making in European countries

Country Systems Master Inter-modal Time Private

approach plan cooperation horizon financing

France S M S Long M
Germany M W M Long S
Italy S M S Short S
Belgium S S S Short S
Sweden M M S Medium S
Denmark M M M Short S
Norway M M S Medium S
Finland S S S Medium S
Switzerland W W M Long S

Notes: S = scarcely developed. M = partially developed. W = well developed.

BUTTON 9781786435668 PRINT.indd 420 27/04/2022 11:07

 INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­ 421

Exhibit   The Third London Airport Study

The Third London Airport Study, The Roskill Commission, which reported in 1971,
established the modern textbook CBA approach to large-scale investment appraisal. In this
case, however, it was not a full CBA study because it was assumed that a new airport was
necessary for London anyway and that the economic benefits were deemed virtually equal
for all possible sites.
The growth in United Kingdom air traffic grew rapidly throughout the 1950s, and by the
early 1960s the question of a possible third London airport began to emerge. By 1961,
it was forecast that the two London airports, Heathrow and Gatwick, when considered
together, would exceed capacity from about 1973 and there was a need for a new, two-
runway airport by 1980. After several attempts to look at alternatives in essentially a
piecemeal way, a Commission was established in 1968. This, under Mr Justice Roskill, was
‘[t]o enquire into the timing of the need for a four-runway airport to cater for the growth
of traffic at existing airports serving the London Area, to consider the various alternative
sites, and to recommend which site should be selected’.

Cublington Foulness Nuthampstead Thurleigh

Capital costs
   Construction of airport 184.0 179.0 178.0 166.0
   Airport services 14.3 9.8 14.5 11.6
   Extension/closure of Luton Airport –1.3 10.0 –1.3 –1.3
   Road & rail development 11.8 23.4 15.5 6.5
   Relocation of defense & public
  scientific establishments 67.4 21.0 57.9 84.2
   Loss of agricultural land 3.1 4.2 7.2 4.6
   Impact on residential conditions 3.5 4.0 2.1 1.6
   Impact on schools, hospitals, etc. 2.5 0.8 4.1 4.9
  Other 3.5 0.5 6.7 10.2
Total 288.8 252.7 284.7 288.3
Current costs
   Aircraft movements 960.0 973.0 987.0 972.0
  Passenger users 931.0 1,041.0 987.0 931.0
  Freight users 13.4 23.1 17.0 13.9
   Airport services & operations 60.3 53.1 56.2 55.6
  Travel costs to/from airport 26.2 26.5 24.4 25.4
  Other 12.4 7.5 8.5 7.2
Total 2,003.3 2,124.2 1,988.1 2,005.1
Benefits (relative to Foulness)
  To common/diverted traffic
    (net of costs) – n.a. – –
  To generated traffic 44.0 n.a. 27.0 42.0
Total (cost minus differential benefits) 2,248.1 2,376.9 2,245.8 2,251.4

Note: The table is only a partial reflection of the results obtained and does not, for example, include the
sensitivity analysis conducted.
n.a. = not available.

BUTTON 9781786435668 PRINT.indd 421 27/04/2022 11:07

 422 TRANSPORT ECONOMICS, 4TH EDITION

The implication of the question posed involved considering where and when an airport
should be built – not whether one should be built. This meant that in some ways it became
a social cost-effectiveness study: finding the site with the lowest social costs attached to
it. The initial list of 78 sites was reduced to an intermediate list of 29, before detailed
consideration of Cublington, Foulness, Nuthampstead, and Thurleigh.
While the study team favored Cublington as marginally superior to the other sites,
subsequent parliamentary debate over-ruled this in favor of Foulness. Although even this
revised proposal was later abandoned, the study proved useful in showing up some of the
practical difficulties in conducting a CBA study of a scheme which has extremely wide-
ranging and diverse impacts – many of them posing serious problems of evaluation.
The present values of the various cost and benefit items for each alternative discounted
from 2006 back to 1975 values and in millions at 1970 prices are given in the table. In all,
the final analysis comprised 20 factors for consideration. All the estimations made were
essentially based upon informed judgements.
Many of the items had shadow values placed upon them to allow them to be represented
in a common unit. Noise nuisance, for example, was valued using hedonic indices. These
considered the market value of housing at various distances from the noise envelopes
created by aircraft taking off and landing at airports. These values were broken down by
such things as the sizes of plots, the sizes of the housing units, the number of features of
each house – bathrooms, bedrooms, etc., – whether there was a garage, distance from
shopping facilities, etc., as well as the noise impact they were subjected to. Normalizing for
all features but noise nuisance and setting this off against house values provided an economic
cost of the noise.
Regarding travel time costs to/from the various locations, values were placed on the loss of
work due to travel to and from an airport and on leisure time adjusting for the average
incomes of business and leisure travelers. These were further broken down by mode of
transport used (cars, light vans, and public transport), making allowances for occupancy
levels. The values used were based upon theoretical considerations that work time costs
were ultimately borne by the employer and thus the extra wages and supplementary
payments to workers was used. For leisure, a fraction of this derived from empirical studies
was adopted. Similarly, the costs at the airport of boarding, taxing, etc., were included,
together with the costs to airlines of the different routings they would have to fly.
The site at Cublington was recommended by the Commission but the government
rejected this and accepted a dissenting report by Colin Buchanan, a member of the
Commission, which recommended that a new airport should be developed at Foulness
(later known as Maplin Sands) in Essex. Subsequently, plans for a new airport were replaced
by a smaller-scale redevelopment of Stansted Airport, a site not short-listed by the Roskill
Commission.
See also: E.J. Mishan (1970) What is wrong with Roskill, Journal of Transport Economics and
Policy, 4, 221–34; and Roskill (Commission on the Third London Airport) (1971) Report,
HMSO.

BUTTON 9781786435668 PRINT.indd 422 27/04/2022 11:07

 INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­ 423

considered, and largely in qualitative terms, in public inquiries into which COBA
quantitative outputs were also fed.
Despite the widespread adoption of CBA by the transport sector, there has
been a gradual disillusionment with the all-embracing, stereotypical appraisal
implied by Prest and Turvey. This has manifested itself most strongly since the
rejection of the Roskill Committee’s recommendation regarding the siting of a
third London airport and became noticeable at public inquiries into new road
proposals in the late 1970s and early 1980s. While the criticisms of CBA as a
method of socially evaluating transport investments have been extensive, they
are perhaps most adequately summed up by Aaron Wildavsky (1966): ‘Although
cost–benefit analysis presumably results in efficiency by adding the most to
national income, it is shot through with political and social value choices and sur-
rounded by uncertainties and difficulties of computation’. A former Chairman
of British Rail, Peter Parker, summarized the attitude evolving in the United
Kingdom when he argued that there is a need for an approach that

can be understood by ordinary intelligent people [and] incorporates the methods of

analysis developed by welfare economists over the last decade or so [and] gets away
from the naive position adopted by the early cost–benefit men which seemed to imply
that every consideration could be perfectly weighted and that, therefore, there was a
single best solution.

The responses of analysts to these dissatisfactions with mechanical CBA pro-

cedures took two broad lines. The first was an attempt to modify the original
CBA framework (as exemplified by equation 11.2) with some allowances made
to meet the major criticisms. Effort was put into evaluating externality items
included in a CBA account and to placing more reliable values on time-saving
attributes of schemes. Advances on this front were more rapid in some countries
than in others. Sweden, for example, used a variety of techniques to enable a
diverse range of effects, often expressed in monetary terms, to be brought into the
appraisal process. The position in the United Kingdom was to seek a consistent
process following experimentation in embracing a limited number of environ-
mental factors within COBA (UK Department of Transport, 1992).
Additionally, techniques have evolved that introduce allowances for the dis-
tributional effects of schemes – an area neglected in earlier work which concen-
trated on overall impact – and for the risk and uncertainty that the predicted cost
and benefit streams will diverge from that forecast. Theoretically, distributional
effects can be allowed for by weighting the costs and benefits according to the dif-
ferent groups affected. Unfortunately, it has been demonstrated at the theoretical
level that in many investment situations the applications of such weights (which
may be based upon measures reflecting income tax liability) to cost and benefit
items can still lead to the acceptance of projects which benefit the rich to the
detriment of the poor. Consequently, there is a case for treating distributional
considerations independently of efficiency.
Risk and uncertainty about probable outcomes pose even more difficult
problems. With risks there is some knowledge about the likelihood of errors in

BUTTON 9781786435668 PRINT.indd 423 27/04/2022 11:07

 424 TRANSPORT ECONOMICS, 4TH EDITION

forecasts and this can be incorporated in the analysis by indicating the range
of probable long-term effects of investment, together with an indication of the
probabilities of different levels of costs and benefits occurring. Unfortunately,
there is no such knowledge of possible error with uncertainty, and consequently
­adjustments tend to be made according to intuition or ‘skilled judgment’. With
many transport projects, the costs of under-engineering are likely to be higher than
those of comparable over-engineering (the ‘premature’ physical d ­ isintegration of
the United Kingdom motorway system being a good example) and thus there is a
tendency to over-react to the possibility of uncertain outcomes.
While these advances in traditional CBA techniques went some way towards
meeting criticism of early studies in the field, they tend to complicate the estima-
tion and decision-making frameworks and, hence, to move even further from
the openness sought by the Minister for Transport and the Leitch Committee
on Trunk Road Investment appraisal in the 1970s. One offshoot of CBA which
retains the notion of social welfare maximization but also makes the CBA account
accessible to the proverbial ‘educated layman’ is the planning balance sheet (PBS),
which was initially devised and developed over a series of case studies to help
urban planners. We discuss the PBSs in more detail below.
The second response to the critics is to move entirely away from the notion
of a social welfare maximization CBA approach and to adopt a lower-level, but
possibly more operational and manageable, approach to investment appraisal.
This, for example, is the approach that has increasingly been followed in France
since the late 1960s.
Broadly it is argued that, like most large private companies, public transport
undertakings have insufficient information about the stream of costs and benefits
(including social items) associated with the different policy options open to them
and should, therefore, attempt to meet broad minimum levels of achievement
rather than to maximize net benefits. This notion of ‘satisficing’ fits in with the
attitude of most mature industrial concerns towards managerial decision-making
(Simon, 1959). Although this second type of response to the critics of CBA is,
to date, still comparatively under-researched in the transport field, several multi-
criteria investment appraisal techniques have been developed, often only at an
abstract level, in related areas of study such as regional and national resource
planning (Dimitriou et al., 2016).
The PBS approach mentioned above, although firmly founded in the CBA
tradition, offers a methodology that is sufficiently flexible to adaptation for both
maximizing and satisficing frameworks. It has two main merits: first, it shifts the
emphasis of analysis away from the total measure of net benefit to the distribution
of the costs and benefits among affected groups; and, second, it circumvents many
of the problems associated with expressing all costs and benefits in money terms.
The technique involves setting down, in tabular form, all the pros and cons
associated with alternative investment options. These socioeconomic accounts are
expressed in monetary values wherever possible but should this prove impracti-
cable then physical values are used and, if quantification is not possible, ordinal
indices or scales. The accounts are subdivided to show the effect of different

BUTTON 9781786435668 PRINT.indd 424 27/04/2022 11:07

 425
INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­

schemes on the groups affected, and this offers guidance to distributional implica-
tions. The accounts are compared with pre-determined planning goals (and these
instrumental objectives may imply either maximization or satisficing objectives)
which are selected as reflective of community preferences. Alternative investment
plans are ranked under each objective heading using ordinal ranking procedures
and the ranks are then added together to produce a ranking of the investments
with respect to the objectives taken as a whole.
A technique of this general kind met with approval from the Leitch
Committee in the United Kingdom as a tool in inter-urban road investment
appraisal. The Committee felt ‘the right approach is through a comprehensive
framework which embraces all the factors and groups of people involved in
scheme assessment’ (UK Department of Transport, 1978b). The project impact
matrix, as the Leitch Committee called their variation, sets out a ‘general frame-
work’ of about 80 relevant measures of the effects of transport schemes. As
with most PBS studies the final account produced was extensive but Table 11.5
provides a summary. The intention is to use such an account to make pair-wise
comparisons between the magnitude of the effects associated with different
investment alternatives, or, where the problem is deciding upon a specific project
in isolation, to compare them with some instrumental objectives.
The PBS-type approach has, despite its attractions, some inherent limita-
tions. It depends upon crude ranking criteria and scaling methods. The selec-
tion of instrumental objectives is itself highly subjective and, although it does
force the decision-maker to make his/her underlying value judgments explicit, it
can result in some conflict between interested parties. There is also the danger
that the subjectivity of these objectives and trade-offs is forgotten in the mass
of data incorporated in the accounts. The PBS has the advantage over some
of the more mechanical approaches where numbers are simply fed into some
computer program (such as COBA, developed in the United Kingdom for trunk
road investment appraisal) in that the construction of the initial socioeconomic
account can often be educational and can shed considerable light on salient ques-
tions the decision-maker should be asking.
While PBS can be considered as an extension of CBA, it may also be viewed
as a primitive form of a ‘multi-criteria decision-making technique’. Multi-criteria
decision-making techniques fall into the second category of advances outlined
previously, in that they are concerned more with the meeting of certain low-level
aims than with maximizing social welfare. They involve weighting the different
effects of an investment to reflect social priorities, but the weights reflect success
at attaining certain objectives rather than maximizing an output.
Several multi-criteria approaches have been devised, each attempting to
achieve a multi-dimensional compromise between the wide diversity of goals and
objectives that are embodied in any form of public choice. Approaches differ in
their methods of presentation, the level of mathematical sophistication involved,
and the amount of data input required. Several of the techniques rely upon
geometrical representation to produce multi-dimensional scalings, while others
involve a considerable degree of intuition. Of greatest practical value in transport

BUTTON 9781786435668 PRINT.indd 425 27/04/2022 11:07

 426 TRANSPORT ECONOMICS, 4TH EDITION

Table 11.5 
The project impact matrix suggested by the Leitch Committee in the United
Kingdom

Incidence group Nature of effect Number of measures

Financial Other

Road users directly Accident savings 1 3

affected Comfort and convenience – –
Time savings 6 –
Vehicle operating cost savings 5 –
Amenity – 2
Non-road users Demolition or disamenity to owners – 22
directly affected of residential, commercial, &
industrial properties
Demolition or disamenity to users of – 15
schools, churches, public open spaces
Land-take, severance, & disamenity – 7
to farmers
Those concerned with Landscape, scientific, & historical 3 –
the intrinsic value value
of an area
(+ verbal description)
Those indirectly Sterilization of natural resources, – 6
affected land-use planning effects on other
transport operators
(+ verbal description)
Financial authority Cost & financial benefits 19 7

Source: Adapted from UK Department of Transport (1978a).

are some of the weighting techniques, of which there are numerous variations.
Maurice Hill’s (1968) goal achievement matrix, for example, offers an explicit
treatment of various goals and applies a set of pre-determined weights to them
so that each option can be assessed in terms of goals achieved. To facilitate this,
the goals are related to physical measures (for example, minimum traffic speeds,
acceptable accident rates, reduced levels of specified toxic exhaust emissions, etc.)
to reflect the extent to which they have been achieved. The final goal achievement
account employs the weighted index of goal achievement to determine the pre-
ferred course of action.
The problem with all useful multi-criteria procedures is the derivation of
weighting schemes that reflect the relative importance of physical ‘goals’ or
‘objective instruments’; seldom will a public sector transport scheme do all that is
hoped for. The traditional CBA approach, albeit in a maximizing context, avoids
this problem by using monetary values as weights. While there is evidence that
those responsible for decision-making in the publicly controlled sectors of trans-
port favor movement towards multi-criteria appraisal techniques, the practical
problems are unlikely to permit the widespread use of such approaches – beyond
the project impact matrix type of analysis – soon.

BUTTON 9781786435668 PRINT.indd 426 27/04/2022 11:07

 INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­ 427

These types of modifications to the cost–benefit approach are likely to lead

to a more consistent treatment of environmental effects, but there is a more
fundamental point about infrastructure provision. It has been argued that the
conventional CBA approach to appraisal, because of its inherent assumptions
regarding pricing, can lead to over-investment in transport infrastructure and
excessive transport use. In Figure 11.6 we show various combinations of prices
charged for road use, and, on the vertical, various levels (in money terms) of
investment. There will be some optimal price–investment mix, such as P*/S*,
which is optimal. This would be the socially efficient outcome if road users
were charged optimal prices (embracing all external considerations) for their
journeys.
Suppose, however, that the price in effect does not fully reflect costs; then,
since there would be a heavy demand for road use, the conventional CBA
approach would imply that more investment than S* is required. Curve I traces
out the relevant optimal price–investment combinations that would emerge. In
fact, the low price is generating demand beyond the optimal level and thus the
overall amount of traffic on curve I at points to the left will be suboptimally large.
William Wheaton (1978) argues, therefore, that investment should be limited
along a P/S curve such as II. The additional congestion occurring at any price
below P*, because of the limited additional investment in capacity, will in effect
constrain traffic flows to the optimal level. The reasoning behind this conclusion
is summarized by Wheaton in the following way:

Such a reduction will increase congestion, and this helps to discourage the demand
that has been ‘artificially’ induced by underpricing. It is important to remember that
second-best investment does not call for building fewer roads as the price of driving is
lowered. That would result in ‘excessive’ congestion. Rather it requires accommodat-
ing less of the induced demand than would be met if a simple cost–benefit analysis
were applied.

Investment ($)

S* II

0 P* Price ($)
Figure 11.6 Second-best investment with suboptimal congestion pricing

BUTTON 9781786435668 PRINT.indd 427 27/04/2022 11:07

 428 TRANSPORT ECONOMICS, 4TH EDITION

Applying this to the question of environmental policy, the relevance is mainly in

terms of just how much investment in new infrastructure is economically justi-
fied when transport prices do not reflect true costs. The standard methodology
tends to ignore the imperfections which exist in terms of transport users not
paying for the full costs of their activities and thus to favor high levels of invest-
ment. Wheaton’s analysis essentially implies that in these conditions, where money
prices are ineffective, second-best criteria determined by travelers’ time costs (that
is, congestion) can be used to limit travel to a level closer to the optimum.
The conditions for achieving this second-best situation may, however, prove
to be complex. In terms of equity there may be a further argument in favor of
such an approach, in that time is allocated even across individuals. Practically,
there are difficulties in working out the optimal second-best strategy and, in
overall environmental terms, given the proportionately higher pollution, noise,
and other costs associated with congested roads, it is not altogether clear what the
ultimate overall social outcome would be.

11.8 Comparability Between Appraisal Techniques

If scarce investment funds are to be allocated to best effect within the overall
transport sector and between it and the rest of the economy, it is clearly impor-
tant that in some way comparisons are made between the potential effect of using
funds in projects evaluated on commercial criteria and using them where social
evaluation techniques such as CBA are employed.
Accepting that for institutional or administrative reasons there is no hope
of a common method of assessment being employed in practice, then one pos-
sible method of comparing projects between sectors is to develop comparability
criteria that reduce social and financial costs and benefits to a common denomi-
nator. Essentially a mathematical relationship between net social and net financial
returns must be found. In certain highly restrictive situations this may prove to
be feasible. It is theoretically possible, under simplistic assumptions, to reduce
everything to either a common financial or common social basis; we will assume,
however, that we are assessing a potential investment aimed at improving an inter-
city rail service (where profit-maximizing levies are charged) and wish to convert
the net reserves obtained into social welfare terms. Social welfare is assumed here
to refer to social surplus (that is, combined consumer and producer surpluses), as
is standard practice in welfare economics.
Figure 11.7 shows the demand curve for the existing rail service to be linear
(D1 with marginal revenue curve MR1) and that the improvement will result in
a parallel shift of this curve to D2. The average and marginal costs of using the
service are assumed to be constant, irrespective of custom with MC1 (= AC1)
being the relevant curve prior to improvement, and MC2 (= AC2) being operative
afterwards.
With these assumptions, the demand curves are easily represented: D1 as
P = a – kQ and D2 as P = b – kQ, where P is price, Q is the level of traffic flow,

BUTTON 9781786435668 PRINT.indd 428 27/04/2022 11:07

 INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­ 429

a
P2 g h
P1

c
d MC1 = AC1

e
f MC2 = AC2

MR2 MR1 D2 D1
0 Q1 Q2 Traffic flow

Figure 11.7 Comparability between commercial and social investment criteria

and k is the slope of the parallel demand curves. With profit maximization (that is,
output at the point where MC = MR), it is seen from Figure 11.7 that Q1 = (a – d)/
(2k) and Q2 = (b – f)/(2k). Integrating under the relevant marginal revenue curves
shows that the improved rail service would increase profits by {Area (bef) – Area
(acd)}, which is:
1
0.5 ( b − f ) Q2 − 0.5 ( a − d ) Q1 = (b − f )2 − (a − d )2  . (11.8)
4k 

The increase in consumers’ surplus associated with the improved rail service
is obtained from integrating under the relevant demand curves but above price. In
this case the integration yields Area (bhP2) – Area (agP1), which equals:
1
0.5 ( b − P2 ) Q2 − 0.5 ( a − P1 ) Q1 =
(b − f )2 − (a − d )2  . (11.9)
8k 
Since social surplus is composed of producers’ surplus (that is, profit) plus
consumers’ surplus, it is apparent from adding equation (11.8) to equation (11.9)
(and then comparing back to equation (11.8)) that the gain in social welfare,
resulting from rail investment in this profit-maximizing situation, is 1.5 times the
profit which would be earned. It seems possible, therefore, in the circumstances,
to be able to convert profits earned into a comparable social surplus by multiply-
ing by 1.5.
How useful is this conversion factor likely to be in practice? This only
applies to user costs and does not permit the inclusion of external factors,
either in terms of pollution or congestion, which limits its usefulness in urban
transport appraisal or for certain types of infrastructure, such as airports,
which are environmentally intrusive. Further, even within the strict confines of
user-benefit analysis the conversion factor crucially depends upon a series of

BUTTON 9781786435668 PRINT.indd 429 27/04/2022 11:07

 430 TRANSPORT ECONOMICS, 4TH EDITION

limiting assumptions. The factors which have been found to influence the ratio
include:

• the shape of the MC curve before and after the investment;

• the shapes of the demand curves before and after the investment;
• the extent of price discrimination;
• the pricing policy pursued and the nature (if any) of its deviation from profit
maximization;
• the consistency of the pricing policy employed as investment alters the cost
and demand conditions;
• the incidence of externalities including network effects; and
• the extent to which revenue and benefit streams differ in their availability for
reinvestment.

Given the sensitivity of the ‘1.5 rule’ to these various factors, it is hardly
a practicable method of introducing comparability into transport investment
decision-making.
The comparability ratio approach assumes an investment is undertaken and
then prices set to achieve some economic objective, usually profit maximization.
Financial returns are then compared to social returns. Starkie (1979) argues that
a more practicable approach is to determine a common basis for pricing first and
then adjust capacity accordingly. The basic idea stems from work on the railways
by Stewart Joy (1964), who looked exclusively at freight investment, but Starkie
generalizes the approach to all forms of transport investment. It is assumed that
the correct economic price for each mode is determined along the second-best
pricing lines formalized by William Baumol and David Bradford (1970), and
that investment, or disinvestment, should be adjusted until LRMCs are equated
with the revenues obtained. This means that prices are set to cover SRMC with
a mark-up in proportion to the inverse of the price elasticity of demand for each
mode. The mark-up then reflects ‘what the user will bear’ towards the cost of
capacity provision (that is, consumer surplus above SRMC). If this mark-up,
combined with the revenue covering SRMC, does not meet the full LRMC then
capacity should be reduced until an equality is established. If such a pricing
regime produces a surplus in excess of LRMC then, ceteris paribus, there is a case
for expanding capacity.
The fundamental idea is that if sufficient price discrimination is applied then
all potential consumers’ surplus is transferred to the supplier and, ipso facto, net
revenue can be equated with social surplus. Its main advance is the importance
which it places on pricing and the recognition that in the long run it is possible
to fine-tune investment at the margin. Such an approach obviously removes the
need to conduct comparability studies but it has its limitations. The main dif-
ficulty is that, while it may be possible in some areas to apply the discriminate
pricing Starkie advocates, and mainly those undertakings directly controlled
by government, in other cases private provision of transport facilities makes
it rather difficult to ensure that the Baumol/Bradford rules are being applied.

BUTTON 9781786435668 PRINT.indd 430 27/04/2022 11:07

 431
INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­

Consequently, it is difficult to see how it could be decided what is the correct

level of overall investment in the publicly owned sector of transport vis-à-vis
the aggregate for the private sector. Because much public investment is assessed
on social criteria and virtually all private sector investment is assessed on com-
mercial criteria, many of the problems of comparability remain. (This problem
is avoided in Starkie’s empirical work, which focuses on road and railway track
investment.) Further, the approach once again emphasizes user benefits but does
not allow for external factors, especially the environmental effects of transport
on non-users.

11.9 Assessing the Effect on National Income

It has been suggested that, rather than expand financial surplus by a comparabil-
ity ratio, or force some form of common pricing on all sectors of transport, an
entirely different measure of the net value of investment, applicable to all forms
of transport project irrespective of ownership, may be preferable. The effect of
transport on national income, for example, could be used as a substitute for the
combined consumers’ and producers’ surplus generated. However, besides the
practical difficulties involved in estimating the change in national income associ-
ated with alternative transport investments, the measure throws up an additional
problem that involves the more fundamental question of whether the national
income approach really does offer a reasonable and acceptable guide to the rela-
tive desirability of alternative investments. (We should perhaps note at this stage
that national income, in this context, refers to the accountancy concept used in
macroeconomics rather than the wider notion of national income referred to by
Wildavsky earlier in the chapter.)
We can consider Figure 11.8 and assume that demand will not shift following
a change in capacity. Further, if we assume the transport undertaking acts as a
monopolist in its pricing policy, then we can see that an investment that reduces
marginal costs from MC1 to MC2 will increase social surplus by (hgcd) in the
diagram and profits by (hdef) – (gcba). The reduction in costs will also produce
a higher national income. If the Laspeyres index is used (that is, the change in
output valued at the pre-investment price), the rise will be measured as Q1cjQ2.
If the Paasche approach is favored (that is, the change in output is valued at the
post-investment price), however, the addition to national income is found to be
only Q1idQ2.
There is no reason for the social surplus measure to correspond to the
national income measure (or for different estimates of the latter to correspond)
except in rather unrealistic circumstances. Nor, indeed, need it correspond to the
profit generated. Of more practical importance, there is no reason why alternative
investment possibilities will be ranked consistently by the different methods.
The reason why social surplus and national income measures (and, indeed,
financial measures) need not correspond, nor rank consistently, stems from the
fact that they measure entirely different things. Social surplus both includes

BUTTON 9781786435668 PRINT.indd 431 27/04/2022 11:07

 432 TRANSPORT ECONOMICS, 4TH EDITION

g c j
P1
P2 d
h i

a b
MC1

f e
MC2
MR Demand

0 Q1 Q2 Traffic flow
Figure 11.8 The national income change approach to transport investment appraisal

leisure benefits emanating from a project and explicitly incorporates allowances

for the diminishing marginal utility associated with the increased consumption
of travel. The national income measure does neither: it concentrates exclusively
on goods and services traded in conventional markets, and assumes either a fixed
pre- or post-investment price level.
Herbert Mohring (1976) demonstrated that the only time the two types
of measure yield identical numerical results is ‘If a change takes place which
increases the output obtained from a given set of primary resources, and if the
primary resources allocated to market activities do not themselves change, and if
the same pricing rules are used in consumers’ surplus as in national income change
benefit calculations’ (emphasis original). Clearly, this means that in practice, the
national income measure is likely to differ from either the financial or social
surplus measure of benefit and hence is simply an alternative method of ranking
and appraisal. It is different but not necessarily superior. Given the practical dif-
ficulties of estimation in most advanced economies its usefulness seems limited.
In less developed countries, though, it may be a more viable appraisal technique if
distributional and welfare considerations are felt to be of secondary importance
to boosting the national product. Transport projects that help stimulate national
income growth in these circumstances may be given priority over others.
The practical problems of incomparability between investments in different
transport sectors has not gone unnoticed by United Kingdom and other trans-
port authorities. The problem is particularly acute in publicly owned inland inter-
urban transport where government policy in countries like the United Kingdom
have distinguished between different types of service, some of which must show
a financial return and others a social return. While the moves towards privatiza-
tion and deregulation from the 1970s (see Chapter 14) have rather blurred the
distinction, the problem of allocating resources between commercial and social
transport services remains.

BUTTON 9781786435668 PRINT.indd 432 27/04/2022 11:07

 433
INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­

Certain measures have been introduced in the United Kingdom that

attempted to standardize some elements of appraisal procedures across the two
groups. Since the late 1960s, for example, discounting techniques were employed
in all sectors using a nationally stipulated rate of discount. Some minor changes
in the basis of calculation took place subsequently but such techniques remained
the main tool of appraisal for a long period. The differences between sectors are
just the items discounted; in one case it is simply the fare revenues, while in the
other it is user benefits (usually calculated in terms of the financial value of the
money, accident rate, and travel time changes associated with the investment).
The official position, until relatively recently, was that, although there may be
a need to take distributional or environmental factors into account when doing
either type of calculation, the pricing policies pursued mean that revenue may be
equated with social surplus in the financially based calculation. Comparability
then becomes unnecessary. The rationale behind this view was summarized by
the Department of the Environment, the argument being that although ‘the
mechanics of assessment differ … in ordinary circumstances British Railways
pursue a policy of market pricing which ensures that most benefits to users are
fully reflected in the revenue collected’ (emphasis added). A Select Committee was
not convinced, however, that inconsistencies did not remain and advocated the
establishment of an Interurban Directorate to tackle the problem.
The Leitch Committee in the United Kingdom, when looking at trunk
road investment, examined comparability in more detail and recommended that
consistency could only be achieved if those undertakings employing financial
techniques extended their analysis to embrace the wider social-surplus-based
procedures of appraisal. As we have seen, the type of CBA envisaged, however,
is somewhat more comprehensive and transparent than that currently used for
road investment appraisal in the United Kingdom, which concentrates solely on
user benefits. Further, to gain an even greater degree of consistency, there is a
need to bring port and airport investment into the general CBA framework and
widen the scope of appraisal away from the purely financial aspects which tend
currently (with some significant exceptions) to dominate decision-making in
these spheres.
While the notion of common CBA techniques being applied throughout
the publicly controlled inter-urban transport system may have certain appar-
ent advantages – in Germany and some other European countries it is used for
all inter-urban road, rail, and waterway analysis – the Leitch study recognized
that the private ownership of coastal shipping and pipeline limits the degree of
comparability. Essentially, the widespread use of CBA within the publicly owned
sector (which, in fact, goes beyond the road/rail situation considered by Leitch)
may produce consistent investment decisions within this sector but does not
determine the optimal size of public vis-à-vis private sector investment. In effect,
very limited progress has been made towards a common method of economic
assessment for road and rail investments in the United Kingdom, and some other
countries.

BUTTON 9781786435668 PRINT.indd 433 27/04/2022 11:07

 434 TRANSPORT ECONOMICS, 4TH EDITION

11.10 Some Institutional Considerations

From a policy perspective, while there are reasons to ensure that adequate trans-
port is available to facilitate economic development, and thus infrastructure
investments are made efficiently, there are market imperfections that can prevent
this from happening. One set of constraints encountered in most countries
is associated with planning requirements (Plotch, 2015). These can delay the
construction of a facility and affect its design and operations. We will not deal
directly with these issues here, because they vary considerably between countries,
but will rather focus on issues of finance and efficient operations. Financing
can pose particular problems in poorer countries, or even regions within more
prosperous nations, if they do not have an adequate tax base or access to more
sophisticated money markets. The scale of much transport investment also means
that they often get entangled in macroeconomic policy and the cycles of public
expenditure that accompany this.
Just taking the example of airports (although roads or seaports would serve
equally well): airports are relatively expensive to build, and, because of their
monopoly status in many locations, are often not operated efficiently. To reduce
the X-inefficiency, there have been moves in many countries to make airport
operators more commercially oriented. The national approaches aimed at inject-
ing more commercial pressures into the provision of airport services have varied.
Much of the interest that is now emerging is either in the privatization of some
aspects of airport activity or in engaging the private sector in some partner-
ship arrangement with the state. In part this is because an airport is effectively
a composite entity consisting of units offering a variety of services: land access,
parking, concessions, terminals, runways, ground handling, fire and response
units, security, etc. Commercialization does not have to be applied to all these
activities, and may be pursued in a piecemeal way if politics or economics dictate.
Approaches differ according to the state of the local air transport market,
which is, in turn, often linked to the stage in economic development of the region
concerned. For example, airports can vary in terms of their potential revenue
flow from different sources – for example, slot fees and concessions – and this can
affect the degree of privatization or deregulation that is possible and the form it
is most likely to take.
Much depends on the state of the regional air transport market. The
forecasts of relatively slow longer-term growth in air traffic in and between the
developed countries (for example, estimated at about 2.6 percent a year to 2039
within North America, 3.3 percent within Europe, and 2.5 percent between North
America and Europe) means that their major airports will increasingly become
dependent on commercial or non-aeronautical revenues to enhance their revenue
stream. This in turn can pose problems in terms of regulation, as has already been
seen in the debates over the imposition of the price-capping regime used in the
United Kingdom.
While still relatively small, the protected growth of many air markets involv-
ing regions of developing countries offers the potential for increased airside

BUTTON 9781786435668 PRINT.indd 434 27/04/2022 11:07

 435
INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­

revenue in situations where there are potentially fewer social constraints involving
such things as noise and land-take on building additional capacity or where there
is already adequate capacity. The scope for raising significant commercial income
is much less, however, because of lower initial traffic bases. It also suggests,
though, that the regulatory regime overseeing a privatized airport system needs
to be less sophisticated because it must only deal with airside issues, making the
potential for various forms of regulatory capture smaller.
The problem for regions of the poorest developing countries, however, is
that even though their air traffic flows may in aggregate be growing, this is from
a low base and they still seldom generate sufficient revenue to cover the full costs
of operations. Airports are essentially decreasing cost entities for which cost
recovery can be difficult, especially in a situation where there is competition from
other airports. This makes pure privatization options less tenable and the need
for outside assistance from aid agencies or from government more relevant. PPPs
offer another alternative (Button, 2008).
The wide variety of circumstances around the world has led to a diversity
of approaches to commercialization of airports. Some have involved complete
divestiture of former state assets to the private sector, albeit with some oversight
of how the airport is operated, but in other instances the withdrawal of the state
has been less complete.
The management contract approach retains government control, but
involves contracting out specified elements of airport services – parking, hotels,
retail concessions, etc. – for periods of time. This normally involves some form
of auction. The system is in line with the notion of ‘competition for the market’.
Long-term contracting involves giving over the operational side of an airport,
sometimes including investment commitments in additional capacity, for an
extended period with the authorities retaining a degree of strategic control. The
financing required normally entails bringing specialized international companies
with the expertise to manage an airport, or system of airports, together with
finance houses that can provide the necessary support for large-scale service
activities. These types of concessions are widespread in South America, where
local expertise and finance is limited, but there is reluctance for the state to divest
itself of aviation assets.
Many developed countries have also pursued similar philosophies when
expanding the involvement of private enterprise, but falling short of complete
divestiture (Button, 2007). Many United States airports have adopted various
concessionary schemes – for example, Boston, Pittsburgh, and Reagan National
Airport in Washington DC have entered into concessionaire agreements, for the
entire operations at Pittsburgh and for specific terminal buildings at the other
airports. At Chicago O’Hare airport, parking has been contracted out, and the
Port Authority of New York and New Jersey, which owns several airports, has a
variety of agreements covering such things as the operation of terminal buildings
and the supply of heating and cooling at some of its facilities. There is also a tra-
dition in the United States of significant airline involvement in providing check-in
facilities and baggage systems.

BUTTON 9781786435668 PRINT.indd 435 27/04/2022 11:07

 436 TRANSPORT ECONOMICS, 4TH EDITION

Complete privatization of major airports is uncommon, although increas-

ing. In most cases there is concern that a privately owned airport will exercise its
monopoly power to extract rent from customers. The challenges are to devise and
operationalize appropriate regulatory regimes to monitor and direct these large
companies in the public interest; often price-capping is deployed. The ongoing
debates about full privatization concern such things as whether single airports,
or as with Argentina, systems of airports, should be privatized, and when they
are privatized, what should be regulated? Should it be all airport activities or just
those directly aviation-related?
In their overview of various governance options, Michael Carney and Keith
Mew (2003) focus correctly on the government being involved in both seeking
to improve the efficiency of their airports and at the same time trying to direct
the gains to specific groups rather than leaving management with full autonomy
in their actions. This involves complexities that, while common to businesses in
developed countries, are unfamiliar in many parts of the world. As a result, this
has added to the growth in international firms specializing in airport manage-
ment, including ownership, to allow the development of common, best-practice
methods of operation while at the same time being innovative in creating bespoke
models for different circumstances.

References

Atkinson, A.B. (2011) The restoration of welfare economics, American Economic Review,
101, 157–61.
Baumol, W.J. and Bradford, D.F. (1970) Optimal departures from marginal cost pricing,
American Economic Review, 60, 215–43.
Button, K.J. (2007) The implications of the commercialization of air transport infrastruc-
ture, in D. Lee (ed.), The Economics of Airline Institutions, Operations and Marketing
2, Elsevier.
Button, K.J. (2008) Air transport infrastructure in developing countries: privatization and
deregulation, in C. Winston and G. de Rus (eds), Aviation Infrastructure Performance:
A Study in Comparative Political Economy, Brookings Institution.
Button, K.J. (2016) Public–private partnerships: a review of economic considerations
with particular reference to transportation projects, Transportation Planning and
Technology, 39, 136–61.
Button, K.J. and McDougall, G. (2006) Institutional and structural changes in air
navigation service-providing organizations, Journal of Air Transport Management, 12,
236–52.
Carney, M. and Mew, K. (2003) Airport governance reform: a strategic management per-
spective, Journal of Air Transport Management, 9, 221–32.
Dimitriou, H.T., Ward, F.J., and Dean, M. (2016) Presenting the case for the applica-
tion of multi-criteria analysis to megaproject infrastructure appraisal, Research in
Transportation Economics, 58, 7–20.
Elhorst, J.P. and Oosterhaven, J. (2008) Integral cost–benefit analysis of Maglev rail pro-
jects under market imperfections, Journal of Transport and Land Use, 1, 65–87.
Foster, C.D. and Neuberger, H.L.I. (1974) The ambiguity of the consumer’s surplus
measure of welfare change, Oxford Economic Papers, 26, 66–77.

BUTTON 9781786435668 PRINT.indd 436 27/04/2022 11:07

 INVESTMENT CRITERIA: PRIVATE AND PUBLIC SECTOR ANALYSIS ­ 437

Guasch, J., Laffont, J.-J., and Straub, S. (2008) Renegotiation of concession contracts in
Latin America: evidence from the water and transports sector, International Journal of
Industrial Organization, 26, 421–42.
Hanson, I.G. (2011) Review of public–private partnerships in heavy railway infrastruc-
ture projects, COST Symposium Public Private Partnerships in Transport: Trends and
Theory, Lisbon.
Hart, O. (2003) Incomplete contracts and public ownership: remarks, and an application
to public–private partnerships, Economic Journal, 113, C69–C76.
Hicks, J.R. (1940) The valuation of social income, Economics, 7, 105–24.
Hill, M. (1968) A goal achievement matrix for evaluating alternative plans, Journal of the
American Institute of Planners, 34, 19–29.
Joy, S. (1964) British Railways’ back costs, Journal of Industrial Economics, 13, 74–89.
Kaldor, N. (1939) Welfare proposition and interpersonal comparisons of utility, Economic
Journal, 49, 549–52.
Lindblom, C.E. (1959) The science of muddling through, Public Administration, 19, 79–88.
Mackie, P.J. and Simon, D. (1998) Twenty-one sources of error and bias in transport
project appraisal, Transport Policy, 5, 1–7.
Mishan, E.J. (1967) Interpretation of the benefits of private transport, Journal of Transport
Economics and Policy, 1, 184–9.
Mohring, H. (1976) Transportation Economics, Ballinger.
Plotch, P.M. (2015) What’s taking so long? Identifying the underlying causes of delays
in planning transportation megaprojects in the United States, Journal of Planning
Literature, 30, 282–95.
Prest, A.R. and Turvey, R. (1965) Cost–benefit analysis: a survey, Economic Journal, 75,
683–735.
Samuelson, P.A. (1961) Evaluation of social income, capital formation and wealth, in P.A.
Lutz and D.C. Hague (eds), The Theory of Capital, St Martins.
Simon, H.A. (1959) Theories of decision-making in economics and behavioral science,
American Economic Review, 49, 253–83.
Skitovsky, T. (1941) A note on welfare propositions in economics, Review of Economic
Studies, 9, 77–88.
Starkie, D.N.M. (1979) Allocation of investment in inter-urban road and rail, Regional
Studies, 13, 323–36.
UK Department of Transport (1978a) Report of the Advisory Committee on Trunk Road
Assessment, HMSO.
UK Department of Transport (1978b) COBA 9 Manual, Department of Transport.
UK Department of Transport (1992) Assessing the environmental impact of road
schemes: standing advisory committee on trunk road appraisal, Report, HMSO.
US Congressional Budget Office (1998) Innovative financing of highways: an analysis of
proposals, Report, US Congress.
US Federal Highway Administration (2020) Operations Benefit-Cost Analysis, TOPS-BC
User’s Manual FHWA.
Vining, A.R., Boardman, A.E., and Poschmann, F. (2005) Public–private partnerships
in the US and Canada: ‘There are no free lunches’, Journal of Comparative Policy
Analysis: Research and Practice, 7, 199–220.
Wheaton, W.C. (1978) Price-induced distortions in urban highway investment, Bell Journal
of Economics, 9, 622–32.
Wildavsky, A. (1966) The political economy of efficiency: cost–benefit analysis, systems
analysis and program budgeting, Public Administration, 26, 292–310.

BUTTON 9781786435668 PRINT.indd 437 27/04/2022 11:07

 12 Transport Planning and Forecasting

12.1 The Development of Transport Planning

The preceding chapters have concentrated primarily upon pricing and investment
decisions for individual modes of transport in isolation: the pricing of public
transport, urban car users, etc. Little has been said about the coordination of
pricing and investment decisions across whole sectors of transport. Coordination,
as Adam Smith pointed out, will come about automatically in a perfectly com-
petitive market framework where marginal cost pricing principles are universally
applied. Indeed, there has been considerable emphasis on coordination through
the market in the development of inter-urban transport policy in countries such
as the United Kingdom and the United States, but extending more broadly, from
the late 1970s as we shall see in Chapter 14. Concern over safety and the envi-
ronment is now often considered to be the main interest of the authorities and,
within a quality licensing and quality regulatory framework, competition both
within and between inter-urban modes is often encouraged.
There is, however, generally some planning of infrastructure provision
in most countries, either by central or local government. At the international
level there are coordinated plans for strategic highway, air traffic control, and
high-speed rail developments. Some involve integrated planning of all transport
systems, as in the Trans-European Networks program of the European Union,
while others, including the Pan-American Highway linking almost all of the
Pacific coastal countries of the Americas, are mode-specific. The extent and
nature of this planning varies considerably between countries, as well as between
modes and regions within them. The main objective in these cases is often the
avoidance of duplication, especially in relation to major development projects
with a long period of gestation, and the adoption of a high degree of technical
consistency.
In other areas of transport activity, and especially in the urban context,
this feeling has declined somewhat in recent years, necessary to introduce a high
degree of planning, with central and local government intervention to improve
the overall efficiency of local transport provision. The role that transport may
play in other spheres of economic activity is one reason for this (improved trans-
port, for example, may form part of a social welfare policy and is currently seen as
central to the revitalization of local economies in depressed inner-city areas), but
a more general explanation may be found in the magnitude of the imperfections
of the urban transport markets. The justification for urban transport planning
in this context was usefully summarized some years ago in a United Kingdom
Department of Transport (1977) policy document, Transport Policy:

BUTTON 9781786435668 PRINT.indd 438 27/04/2022 11:07

 TRANSPORT PLANNING AND FORECASTING ­ 439

The many activities concentrated in urban areas must be accessible to people and the
economic and social life of cities depends on enormously diverse and complex pat-
terns of travel and destination. Yet there is not enough road space in large towns and
cities for people to travel as much as they like and how and when they like. This can be
one source of grievance. Another is that intrusion of dense traffic brings objection-
able and sometimes intolerable noise, fumes and vibration.

We have already seen, in Chapter 6, the extent of the external effects of transport
and have subsequently considered ways in which they may be tackled individually.
A comprehensive marginal pollution pricing regime combined with comparable
social investment criteria could, for example, ensure optimality. Political resist-
ance to such an approach, combined with disquiet over the possible distributional
repercussions and the practical problems of implementation, have so far tended
to rule out the full-scale use of market mechanisms to regulate the urban trans-
port sector. This does not mean that the pricing mechanism is not used, but rather
that it forms part of a much larger package of policy instruments which are com-
bined with the intention of improving the overall efficiency of urban transport in
meeting the objectives of society.
Planning history and philosophies differ between countries, but there
are some general themes that are common (Lay, 2005). Technology changes
have often been a driving force; for example, the advent of the steam locomotive
brought with it the need for strategic planning as fixed networks of rail had to
be laid, which led to early traffic surveys and forecasts of traffic flows to enable
priorities to be developed and costs to be calculated. The emergence of street
cars and underground transport systems from the late 1850s meant that urban
transport planning began to emerge, with government acting as facilitator for the
private sector.

Exhibit   Blue-print planning: L’Enfant’s transport plan for Washington DC

Pierre Charles L’Enfant developed the plan for Washington DC, the new United States
capital, in 1791. The underlying legislation specified that the capital should be situated on the
Potomac River between the Eastern Branch (the Anacostia River) and the Conococheague
Creek near Hagerstown, Maryland. Thomas Jefferson, the secretary of state, had modest
ideas for the capital. L’Enfant saw the task as far more grandiose, believing that he was also
devising the city plan and designing the buildings. There was limited economic thought in
L’Enfant’s subsequent blue-print. It was design-driven.
L’Enfant developed a Baroque plan featuring ceremonial spaces and grand radial avenues,
while respecting natural contours of the land. There was a system of intersecting
diagonal avenues superimposed over a grid system. The avenues radiated from the two
most significant building sites that were to be occupied by houses for Congress and the
Presidency. L’Enfant stated that the avenues were to be wide, grand, lined with trees,
and situated in a manner that would visually connect ideal topographical sites throughout
the city, where important structures, monuments, and fountains were to be erected.

BUTTON 9781786435668 PRINT.indd 439 27/04/2022 11:07

 440 TRANSPORT ECONOMICS, 4TH EDITION

Open spaces were as integral to the capital as the buildings to be erected around them.
Washington’s current road network largely follows L’Enfant’s plan.
Little economic thinking went into the blue-print plan. Ideas of even rudimentary cost–
benefit analysis were absent. There were no serious comparisons with alternative road
layouts. The grid street pattern offers efficiency in moving traffic, and modern road use
makes use of uni-directional traffic flows over much of the system, with movements being
in opposite directions along parallel streets. The intersecting diagonals, however, produce
bottle-necks.

Little thought was given, either, to the management of traffic flows. Lights and traffic
circles (roundabouts) now offer some increased capacity at major junctions, and there are
parking controls. Nevertheless, Washington suffers from acute traffic congestion problems
as highlighted by former city mayor, Adrian Fenty, in his comment that ‘roughly 250,000
commuters drive solo into Washington daily, making them both polluters of our air and
freeloaders of our infrastructure’.
See also: C.M. Harris (1999) Washington’s gamble, L’Enfant’s dream: politics, design, and the
founding of the national capital, The William and Mary Quarterly, 56, 527–64.

By the end of the nineteenth century, transport planning had reached a stage that would be
recognized today. It had become a sequential process (Weiner, 2008):

• Political. This is planning at the conceptual level and considers broad social and economic
goals.

BUTTON 9781786435668 PRINT.indd 440 27/04/2022 11:07

 TRANSPORT PLANNING AND FORECASTING ­ 441

• Development. This is planning at the strategic and systems level, embracing

factors such as land-use and service levels.
• Transport. This is conducted at the corridor level, considering transport
demand management and infrastructure provision.
• Facility. This entails the implementation of the plan, with the provision of
track capacity and its design.

In more detail, the history of urban transport planning in the United

Kingdom, as a case study, is comparatively short, originating in the recogni-
tion that urban life-styles and cities themselves would change once ownership
of private cars became widespread. The physical planners of the immediate
pre- and post-Second World War periods were concerned with redesigning cities
and transport infrastructure to meet the requirements of an automobile age.
Abercrombie, in the County of London Plan of 1943, for example, typified much
of the philosophy of the period in his proposals for the university quarter of
Bloomsbury and the area around Westminster, by advocating the application of
the precinctual principle. Traffic was to be diverted around these areas on good-
quality arterial roads, with the precincts served by a limited number of local
access routes. While this idea, which was American in origin, was never applied
to the two sites studied by Abercrombie, the broad principle was employed in the
post-war reconstruction of Coventry.
The 1947 Town and Country Planning Act institutionalized planning by
making anyone wishing to develop land seek permission from the local authority.
The basis for decision-making regarding land-use changes were a series of pro-
posal maps which traced out proposed land uses for a horizon of ten years or
more. The ‘Town Map’ was a legal document that both took time to prepare
and required cumbersome procedures to be gone through prior to changes being
accepted.
The system had the advantage that it allowed land to be set aside for
schools, residential development, and so on, but only limited land was allocated
to road construction. Further, since funds were not always available for rapid
construction programs, in many places where land was allocated for transport
corridors it led to blight and lack of maintenance of buildings that might at
some unspecified future date be taken for road building. These were problems
since a central theme of the physical planning approach to the urban transport
problem was that improving the local transport network could alleviate excessive
traffic congestion.
As traffic grew, alternative solutions were sought. While the initial approach
to urban planning had been a narrow one, two important developments occurred
in the 1950s and l960s which strengthened the concept. First, although some plan-
ners, such as Abercrombie, took a broad geographical view of urban problems,
the focus of most early planners tended to be local, seeking piecemeal solutions
to specific traffic problems. In the 1950s there was a widening-out. Local highway
authorities were encouraged by central government to produce joint plans
for local road networks and the joint plans produced by the authorities in the

BUTTON 9781786435668 PRINT.indd 441 27/04/2022 11:07

 442 TRANSPORT ECONOMICS, 4TH EDITION

Manchester area (the SELNEC Highway Plan of 1962) and by the Merseyside
area authorities (in 1965) bear witness to the success of this policy. Second, and
not entirely independent of coordinated highway planning, came the recognition
of the strong links between land use and transport planning.
The Buchanan Report (UK Ministry of Transport, 1963) provided firm
evidence of the need to coordinate the two areas of planning. Buchanan was
concerned about the environmental cost of traffic, and argued that urban road
networks should not be expanded to the extent needed to reduce congestion to
some pre-defined levels, as was then the accepted practice, but rather changes
in transport and urban land-use systems should be assessed in terms of the
costs of reducing congestion while maintaining some pre-defined environmental
standard. If the costs of expanding the transport network without violating the
environmental standard proved excessive to the community then traffic should be
restrained until the environmental limit was attained.
Once it became recognized that not only was there a need to consider
objectives other than simply congestion in transport planning, especially since
transport is an integral part of a much wider urban economic system, it became
apparent that physical planning needed to be replaced by a more comprehensive
planning framework. The coordinated approach which resulted embodied ‘struc-
ture planning’, which sets out policies for the development of land, transport, and
the local environment.
The United Kingdom’s Town and Country Planning Act of 1968 embod-
ied the idea of structure planning while the creation of the Department of the
Environment in 1970 integrated overall responsibility for urban and transport
planning into one organization. The 1968 Transport Act created several pas-
senger transport authorities (PTAs) in major conurbations that were given
responsibility for public transport operations within their areas. The PTAs were
themselves committed to drawing up policies (within a year) and plans (within
two years) to provide ‘a properly integrated and efficient system of public trans-
port to meet the needs of [the] area’. While the commitment to draw up structure
plans necessitated liaison and coordination between local planning and highway
departments and the PTAs, in practice land-use, road building, and public trans-
port responsibilities remained separate – indeed in some cases the agencies had
different boundaries.
The Local Government Act, 1972, integrated the existing PTAs, plus three
newly created ones, into a reformed local government structure. The Act placed
further emphasis on the need for coordinated transport planning in urban areas
and each urban authority was compelled to produce a Transport Policy and
Programme (TPP) setting down the strategy and objectives that were being fol-
lowed. The TPPs have been used since 1975 as a means of assessing the level of
central government financial aid to local urban transport undertakings (via the
Transport Supplementary Grant) and emphasis has been placed on integrating
transport planning with the wider issues of land-use planning and social policy
in the area. The aim of this structure, therefore, has been to distribute grants to
reflect local transport needs, and to reduce central government control.

BUTTON 9781786435668 PRINT.indd 442 27/04/2022 11:07

 TRANSPORT PLANNING AND FORECASTING ­ 443

The move to structure planning, which still forms the basis for local trans-
port policy today, resulted in two major changes in the types of approach adopted
by local transport agencies. First, there has been a trend towards more structured
and phased planning; the Tyneside Study of 1968, for instance, produced an
immediate action program, a transportation plan for a 15-year horizon and a
general urban strategy plan to the end of the twentieth century. The TPP frame-
work encourages a continual monitoring and updating of plans within a rolling
framework. Second, planning is no longer viewed simply in terms of investment
but now embraces the short-term management of existing resources. In part this
may be a reflection of the changing objectives of urban transport planning but
it is also the consequence of a greater economic awareness in planning that there
is an opportunity cost associated with all actions involving the employment of
scarce resources.

12.2 The Ethos of Transport Planning

The movement away from physical transport planning to structure planning in the
1960s increased the economic input into the transport planning process, although
the development of modern planning methods must be attributed mainly to civil
engineers, statisticians, and mathematicians rather than economists. One of the
main problems of urban land-use/transport planning is the enormous range of
possible options available, although this is much less of a difficulty in a country
such as Britain, where urban land-use patterns are long-established and not sus-
ceptible to rapid change; in such cases, one is seeking the optimal transport
system for the existing urban structure. A sequential approach to urban transport
planning may, therefore, be appropriate with the four different levels of planning
outlined in Section 12.1 taking place viewed as system designs:

1 design of broad land-use plan;

2 design of strategic transport plan;
3 design of detailed land-use plan; and
4 design of detailed transport plan.

(There may be some feedback between numbers 3 and 4 in the sequence.)

The transport planning process itself is complex and entails far more
complex institutional and technical issues than can be fully considered here. In
general, as we have seen, it can be broken down into several stages, each involv-
ing a certain amount of economic input to it. There is no firm or accepted best-
practice method of drawing up a transport plan, since different agencies favor
different detailed approaches. Broadly, however, the process may be typified by
the various stages set out in Figure 12.1, but this must be seen very much as a
stylization, intended rather more to show where economics can contribute to
the transport planning process, than as a representation of any actual planning
­procedure. A few comments are justified on each of the stages.

BUTTON 9781786435668 PRINT.indd 443 27/04/2022 11:07

 444 TRANSPORT ECONOMICS, 4TH EDITION

Decide upon goals and

objectives

Inventory of existing
patterns of travel

Develop mathematical models

of local supply and demand
relationships in transport

Formulate alternative
plans

Forecast physical effects of plans on

local travel patterns

Economic evaluation of
alternative plans

Execute plan

Figure 12.1 The urban transport planning process

Goals and Objectives

As we have seen above, the objectives of urban transport planning tend to change
over time, periods where there is an emphasis being placed upon social and environ-
mental considerations often being followed by a focus on improving the efficiency
of the system. The general objectives of policy need to be made specific to permit
trade-offs between alternative goals at a later stage of the planning process. The
economist’s contribution at this stage is that of a balancing agent, often counteract-
ing engineering pressures for the emphasis of the plan to be on improving traffic
flow or adopting capital-intensive solutions to environmental problems. The notion
of ‘opportunity cost’ is essential if resources are to be used efficiently. The econo-
mist may also act as a sieve, pointing out the incompatibility or contrary nature of
certain goals. Since goals and objectives are often formulated at the beginning of
the planning exercise, before full information on existing transport problems have
been obtained, they may be redefined after later stages in the sequence.

Inventory of Existing Transport System

Information is gathered about the nature of the local transport system and travel
patterns both by sampling and counting people in the act of transport and by

BUTTON 9781786435668 PRINT.indd 444 27/04/2022 11:07

 TRANSPORT PLANNING AND FORECASTING ­ 445

obtaining information from firms and households about their travel behavior
and requirements (Stopher, 2008). Additional information is often extracted
from official sources such as the Census, the National Travel Survey, and the
Family Expenditure Survey. The types of surveys conducted and the questions
asked have changed over the years as a result of important developments in
transport forecasting. As we see in later sections, much more emphasis is now
placed on understanding why people travel rather than on modeling flows of
traffic.
Consequently, more detailed information of household characteristics is
now sought, although the greater efficiency of modern measuring techniques
means that the actual sample size has tended to fall (from over 10,000 households
in the large studies of the late 1960s – the last substantial household survey in the
United Kingdom, that of the West Yorkshire Transportation Studies, sampled
12,322 addresses – to considerably fewer than 1,000 today).
The broad-brush approach has given way to seeking greater insights into
representative travel behavior. For example, while the US Federal Highways
Administration (1973) set out guidelines for surveying household trip-making
patterns in the 1970s, more recently there have been advances for collecting
broader information on household activity patterns to help link these with travel
behavior (Stopher, 1992).
One reason for this is that the types of planning issues under review have
changed with time, with large road building programs tending to be replaced by
traffic management and public transit policies. But modeling has also evolved.
There has been a shift away from the more mechanical types of approach to travel
patterns, and more of a focus on understanding factors that influence them. The
development of stated preference techniques in recent years, for example, has
furthered our understanding of the underlying nature of travel decisions. Quite
clearly, since the basic aim is to seek information on existing conditions of supply
and demand for urban transport services, economists can contribute to formulat-
ing the types of questions to ask as well as the forms of model in which informa-
tion will be fed.

Mathematical Simulations of the Local Transport Market

The following sections look at modeling and forecasting techniques, and their
economic underpinnings, but it is important to emphasize at this stage the basic
requirements of the simulation models. Transport markets are complex and to
produce models which replicate all their details is both difficult and, more impor-
tantly, likely to be too cumbersome for later forecasting work. Ian Heggie (1978)
suggests that the pre-requisites of a good transport model are that it:

• assists in understanding and explaining behavior;

• aids policy formulation; and
• provides robust predictions.

BUTTON 9781786435668 PRINT.indd 445 27/04/2022 11:07

 446 TRANSPORT ECONOMICS, 4TH EDITION

The models of travel and transport demand are used for forecasting; there-
fore, it is important that the explanatory factors can themselves be predicted with
some degree of certainty. The models are also used to assess the effects of differ-
ent planning options; thus, it is important that they are simple to use and permit
the effect of several alternative strategies to be explored. One of the major limita-
tions of early models was that they were cumbersome to manipulate, limiting the
possible policy options that could be assessed.

Formulating Alternative Plans

The complex and wide-ranging effects of any change in an urban transport

system makes it very difficult in practice for more than a limited number of
detailed plans to be fully formulated. A comprehensive plan consists of a package
of projects and schemes, and for a large city the possible combinations forming
such a package are immense. In some cases, only one planning alternative is
drawn up in detail after a preliminary sifting of other possibilities at an earlier
stage in the planning process.
Usually, the planning alternatives considered are a little more numerous
and, besides the ‘do nothing’ situation, which may act as a benchmark, generally
involve at least one public-transport-orientated package, one private-transport-
based proposal and probably one rather more central alternative, offering a mix of
private and public transport. Since the plans themselves are later evaluated, this
approach should not be as restrictive as it appears. Following the forecasting and/
or the evaluation stage new light may be shed on the detailed effect of alternative
plans, and new compromise packages of projects emerge. Without feedback of
this kind, it would be quite possible, given the vast range of alternative plans that
are feasible for most large cities, to miss the potentially most beneficial alternative.

Forecasting the Physical Effects of Alternative Plans

Transport forecasting is looked at in detail in Sections 12.3 to 12.6, the modeling

and forecasting stages being both central to the planning process and having a
very substantial economics component. The key aspect of forecasting is that it
provides the planner/decision-maker with useful information about the long-
term implications of plans, emphasizing those areas where there are significant
differences in the consequences of the alternative plans. Given the fact that trans-
port is going to exist in the urban setting, irrespective of the plan adopted, and
providing a ‘do nothing’ option is assessed, then it is the relative performances
of plans that are important. Additionally, it is important that the final decision-
makers are aware of the underlying assumptions of the forecasts, so that they
can assess the reliability and strengths of the traffic forecasts. To this end, most
forecasts are now presented as a range of likely outcomes under alternative
assumptions.

BUTTON 9781786435668 PRINT.indd 446 27/04/2022 11:07

 TRANSPORT PLANNING AND FORECASTING ­ 447

Economic Evaluation

The economic evaluation of plans normally involves employing some variant,

albeit in a less formalized format, of the cost–benefit analysis (CBA) approach
outlined earlier. The comprehensive nature and emphasis on distributional
effects which characterizes the planning balance sheet makes it an attractive
technique in the urban planning context. It is also sufficiently ‘open’ as to permit
the various consequences of plans to be related directly to the objectives of the
planners.
With traditional CBA, not only are there problems of placing monetary
values on many items, but also members of the urban community often view the
final output of a single net present value of benefits with suspicion. General prac-
tice involves the inclusion of public participation at the evaluation stage, often
in the form of public inquiries. In some cases – such as the Layfield Inquiry into
the Greater London Development Plan in the United Kingdom – such inquiries
can result in quite substantial revisions of the plan to be adopted. Evaluation of
something as complex as a transport plan is, by necessity, a political process, but
economists can assist the decision-maker by ruling out plans which are clearly
inferior to others while at the same time giving a systematic presentation of the
pros and cons of the final short-list of alternatives.
The inherent value judgments built into all forms of CBA make it a powerful
aid in evaluation, but it is unlikely ever to provide an automatic, purely techni-
cal decision-making calculus, indeed it is debatable whether such a mechanical
approach is to be desired.

Implementation

While it is often possible to implement some parts (usually traffic management

components) of a transport plan almost immediately upon acceptance, other com-
ponents (usually those involving infrastructure changes) are much longer-term in
nature. It is important, therefore, that actions are phased so that costs are kept
at a minimum. Additionally, over time objectives change and ‘errors’ in forecasts
become apparent, and this may necessitate revisions of the plan. Consequently,
there are feedbacks from the implementation stage back to the early stages of the
planning process; in other words, modern planning is seen as an ongoing, rolling
process of adaptation and change. It is often important, in this context, to ensure
the maximum flexibility in the implementation program.
The preceding paragraphs have painted a thumbnail sketch of the transport
planning process, highlighting the role economics can play. It is thin on detail and
hides many of the subtleties of the planning exercise: readers interested in this
specific aspect of transport economics are strongly advised to refer to one of the
excellent texts now available on the subject. The following sections consider the
economic problems of transport modeling and forecasting. These areas, together
with plan evaluation, have become increasingly the preserve of economists and
involve the application of modern microeconomic theory.

BUTTON 9781786435668 PRINT.indd 447 27/04/2022 11:07

 448 TRANSPORT ECONOMICS, 4TH EDITION

12.3 Traffic Modeling and Forecasting

At the outset it should be said that the objective here is not to provide anything
like a comprehensive account of how traffic is modeled for planning purposes;
this is a subject that extends well beyond our scope. What we do is provide an
indication of the types of models that have been adopted, their limitations from
an economic perspective, and some information on some of the more economi-
cally oriented approaches that are gradually coming to the fore. In the past, much
of the work in this area has been dominated by engineers who have been rather
neglectful of the ways in which travelers make decisions, relying instead on
notions of flows derived from the physical sciences.
To conduct successful planning exercises, or indeed many other forms of
policy development, it is, however, essential to have reliable forecasts of the prob-
able effect of different policy options. General qualitative assessments can often
provide useful insights into the effects of different policies, but good planning
decisions require that we have more exact information of the detailed quantified
relationship between travel and transport and the factors that influence them.
Engineers, for example, need projections of future traffic flows when designing
roads and other infrastructure. Recent years have witnessed a substantial growth
in work attempting to specify and calibrate econometric travel demand models.
Not everyone feels that economists are making great progress. As a general
comment on economic forecasts, John Kenneth Galbraith observed: ‘The only
function of economic forecasting is to make astrology look respectable’.
In the United States, the increased role of the federal authorities in funding
both inter-state transport facilities and local, urban transit systems had a similar
effect. The need to allocate large sums to durable transport infrastructure schemes
in the Third World motivated the World Bank to pursue a similar course. More
recently still, the general move towards more careful project appraisal at the
micro level has led to the development of stated preference techniques employing
market-research-style procedures to transport forecasting.

Accuracy of Transport Forecast

Having noted the growth of work in the field, it is only fair to deviate and point to
the quantitative limitations of what has been done before moving on to the tech-
nical details of methodology. In his classic paper on the methodology of positive
economics, Milton Friedman (1953) emphasized the need for a good model to
predict relatively well. In practice, however, very little retrospective work is done
assessing the predictive abilities of models in transport economics.
An early assessment of traffic forecasting in the United Kingdom by
Mackinder and Evans (1981) did not paint a very complementary picture of the
engineering-based methods used at the time: ‘[A]n assumption of zero change
from the base year would not have produced larger forecast errors … . [F]or trips
the errors would have been considerably less [than those obtained from a trans-
port model]’.

BUTTON 9781786435668 PRINT.indd 448 27/04/2022 11:07

 TRANSPORT PLANNING AND FORECASTING ­ 449

A study in the late 1980s of 41 road schemes in the United Kingdom con-
cluded from a comparison of actual and projected flows that only in 22 cases were
the actual flows within 20 percent of the original forecast. Of the remainder, flows
ranged from 50 percent below to 105 percent above the original estimate (UK
House of Commons Committee of Public Accounts, 1988). The forecasts for
the M25 London orbital road, for instance, were that on 21 of the 26 three-lane
sections the traffic flow would be between 50,000 and 79,000 vehicles per day in
the 15th year, whereas the flow within a very short time was between 81,400 and
129,000.
A noted exception is found in Dan McFadden’s (2001) forecasts. These
deploy a random utility mode split model to examine the construction of the
Bay Area Rapid Transit (BART) system in the San Francisco area. The quality
of his results may help explain why he was awarded the Nobel Prize in Economic
Science, although it has not stopped the continued use of engineering consult-
ants’ four-stage modeling sequence for transportation forecasting. Capture is as
endemic in transportation work as elsewhere. (While the conventional aggregate
gravity model forecasts a 15 percent mode share for BART, his disaggregate fore-
cast was 6.3 percent and the actuality was 6.2 percent. Despite this, BART has
never adopted disaggregate modeling as a policy tool.)
Pickrell’s (1989) study of grant programs funded by the United States federal
Urban Mass Transportation Administration found that all ten urban public
transport projects examined produced major under-estimates of costs per pas-
senger (for example, the costs for the Miami heavy rail transit were 872 percent
of those forecast, for Detroit’s downtown people mover they were 795 percent of
those forecast, and for Buffalo’s light rail transit project they were 392 percent
of those forecast). While inaccurate costing was one element of the problem, the
forecast patronage in all cases was over-optimistic. Indeed, only the Washington
heavy rail transit project experiences actual patronage that is more than half of
that which was forecast. Some of the differences can be explained by difficulties in
predicting future values of explanatory variables such as demographic changes,
automobile costs, and the service level which the public transport service would
offer, but Pickrell argues that important questions must also be raised over the
structure of the models employed, the ways in which they were used, and the
interpretation of output during the planning process.
Updating of this work by looking at cost and ridership forecasts for 47
United States transit systems indicates only limited improvements over time.
Figure 12.2 provides details of the forecast versus out-turn investment costs. A
positive value indicates over-estimation, while a negative value indicates under-
estimation. Visual inspection shows that most transit systems do not perform as
well as forecasted in terms of ridership, nor are they constructed consistent with
their estimated costs; there are significant over-runs. For example, in the case
of the Los Angeles Orange Line, ridership was under-estimated by more than
200 percent.
But there have been changes. During the period prior to Pickrell’s study, visual
inspection clearly suggests that ridership was over-estimated while capital costs

BUTTON 9781786435668 PRINT.indd 449 27/04/2022 11:07

 450 TRANSPORT ECONOMICS, 4TH EDITION

150
Ante-Pickrell Post-Pickrell
100

Difference in capital cost estimate

50

and ridership forecast (%) 0

–50

–100

–150

–200

–250
1970 1975 1980 1985 1990 1995 2000 2005 2010
Forecast year
Capital cost difference (%) Ridership difference (%)
Linear (ridership difference (%)) Linear (capital cost difference (%))

Figure 12.2 Ridership forecast and capital cost estimate trends

were under-estimated. However, during the post-Pickrell period, there has been a
slight reversal of this. Further, post-Pickrell values are smaller than ante-Pickrell
values (generally within 50 percent difference), implying an overall improvement
in the forecasting and estimation process. The linear trend lines showing the dif-
ferences in ridership forecasts and capital cost estimates reinforces this notion of
improved forecasts; the differences moved closer to 0 percent line. Of course, exter-
nal factors change with time (including land-use characteristics, transit system
types, and transit technology), and corrections are needed to reflect this.
More recently, Bent Flyvbjerg et al. (2002), looking solely at cost estimations
for 258 transport infrastructure projects (including rail, fixed-link, and roads)
found they are generally under-estimated and are systematically misleading. Rail
projects had the largest error where actual costs were 45 percent higher than esti-
mated. Flyvbjerg (2007) later argues for techniques including benchmarking or
reference class methods, to be developed to improve economic and financial risk
assessment and management in transport policy and planning.

The Challenges of Forecasting

It is quite clear from these findings that forecasting traffic is far from easy. While
these problems are diverse and some rather technical, several general comments
on the application of econometric analysis to transport forecasting highlight the
difficulties that have been encountered in constructing travel demand models.
First, transport is by its nature a derived demand but equally, as we have seen
in previous chapters, it interacts with land-use and location patterns. There is,
therefore, logic in developing a forecasting model that allows for these very close

BUTTON 9781786435668 PRINT.indd 450 27/04/2022 11:07

 451
TRANSPORT PLANNING AND FORECASTING ­

Exhibit   Accuracy in traffic demand forecasting

Considerable efforts go into the long-term demands for transportation and the costs of
constructing and maintaining the required infrastructure. As an extensive number of ex post
studies, and those of Flyvbjerg, have shown, however, the forecasts produced are generally
not very accurate. The evidence is that demand for infrastructure is often over-estimated,
and costs are under-estimated – for example, nine out of ten projects are subject to
significant cost over-runs.

35

30
Percentage of projects

25

20

15

10

0
–80 –60 –40 –20 0 20 40 60 80 100 120 140
Accuracy (%)

Railway projects

35

30
Percentage of projects

25

20

15

10

0
–80 –60 –40 –20 0 20 40 60 80 100 120 140 160 180
Accuracy (%)

Road projects
Differences in the extent of these biases depend on the nature of the investments
concerned. As we see in the figures, plotting the accuracy of passenger traffic forecast for
27 rail and 183 road projects from 14 countries across three continents, the most severe
errors are related to rail projects; on average, rail traffic is over-estimated by over 100
percent compared to an average under-estimation of about 9 percent for roads. Rail errors
also have a more bell-shaped error distribution. (Accuracy is the percentage by which the
out-turn traffic fails to meet that predicted; a lower value on the horizontal indicates greater

BUTTON 9781786435668 PRINT.indd 451 27/04/2022 11:07

 452 TRANSPORT ECONOMICS, 4TH EDITION

over-prediction.) The problem of misleading forecasts for road planners emerges as less
serious than for the railways, and less one-sided.
Looking at the data over time, there are no indications that forecasts have become more
accurate, indeed between 1969 and 1998 they may have become less accurate. Regarding
rail, there also seems to be no difference in accuracy between forecasts made when a project
commenced and when it was concluded. In the case of roads, the degree of under-estimation
is, with some variation, largest when forecasts are made nearer the completion of the project.
This latter situation seems to be due to ‘assumption lag’. The forecasters in some countries
seem to be slower to adjust their assumptions as planning and construction moves forward.
Inaccuracies also seem to grow with the costs involved in rail projects, but this is not the
case with roads, although they do increase with larger projects when measured in terms of
vehicles involved. The latter result has also been seen regarding aviation forecasting.
Those involved in traffic forecasting often provide specific reasons for deviations from the
outcome values. These ‘soft’ data indicate that, for rail, uncertainty over the distribution of
traffic and a deliberate ‘slanting of forecasts’ appear as dominant causes. The effect is to bias
subsequent cost–benefit analysis in favor of rail investment. With respect to rail, uncertainty
about overall traffic growth and over land-use development are major issues. There appears
to be much less political interference in road traffic forecasting, and in the ultimate forecasts
that are accepted, than is the case with railways.
See also: B. Flyvbjerg, M.K. Skamris Holm, and S.L. Buhl (2005) How (in)accurate are demand
forecasts in public works projects? The case of transportation, Journal of the American
Planning Association, 71, 131–46.

linkages. This is obviously not an easy thing to do but it is unlikely that reliable
long-term forecasts will be forthcoming if transport and land-use models are
treated in tandem rather than developed within an interactive framework. Indeed,
using the type of interlinked land-use and transport model set out in Figure 12.3,
it was shown that interactive effects over time account for a significant portion
of the economic benefits of a transport scheme (Berechman and Gordon, 1986).
Moving on to the actual transport modeling framework, traditional micro-
economic analysis specifies a demand relationship relating quantity demand to
price and assumes that this relationship only changes (that is, shifts) when factors
other than price vary. In transport demand analysis it is easy to incorporate
the ‘shift’ variables into a modeling framework because their values are essen-
tially determined outside the transport system (for example, income changes or
changes in taste). The ‘control price’ variable is much more difficult. As we have
seen, price is a broad concept in transport, embracing time, comfort, and other
factors, in addition to simply the monetary cost of a trip. While generalized cost
offers one method of reflecting the multi-dimensional nature of the price variable
there is a tendency in forecasting work to employ changes in the ease of access as
a proxy for price. Accessibility is nothing more complicated than an index that
reflects the ease with which people can achieve the various activities they wish.

BUTTON 9781786435668 PRINT.indd 452 27/04/2022 11:07

 453
TRANSPORT PLANNING AND FORECASTING ­

Transport costs Economic

activity location

Residential and
Car ownership employment Land utilization
location

Capacity of
Trips Housing location
transport system

Source: Webster et al. (1988).

Figure 12.3 Urban land-use and transport interaction model

Transport systems are extremely complicated, comprising many modes of travel,

varieties of routes, and combinations of different potential travel patterns. The
sheer number of links and possibilities in any non-trivial transport network has
inevitably resulted in simplifications having to be adopted by forecasters to permit
calculations to be reduced to manageable proportions. Further, the ‘product’
being offered by the transport system is also unique in that a passenger-mile at a
particular point on the network at a specific time may be performing a completely
different function to another passenger-kilometer at another time and at a differ-
ent point in the system.
While not always designed to handle the land-use/transport interaction
effects, these latter types of problems have resulted in the emergence of three
broad types of forecasting framework, each with its own characteristics and each
with its advantages and defects. Each of these has many variants, but they can
generally be thought of as sequential, disaggregate, and interactive frameworks.
They are looked at in turn. While the discussion is couched primarily in terms
of forecasting the demand for person movements, much of the traffic carried
by the transport system is freight. The demand to move commodities differs
quite substantially from person movements in the sense that it is, normally, uni-
directional, while person trips are usually circular (that is, from home to work
to home). The types of forecasting frameworks used to look at goods traffic are,
however, essentially the same as for person movements, modified usually only in
terms of the actual variables used. The three main approaches are now reviewed
in turn.

12.4 Sequential Travel Demand Forecasting

Sequential models attempt to reduce the complexity of travel demand forecasting

and making the handling of large databases easier by breaking the complicated
patterns of demand for travel into, usually, four submodels (McNally, 2008). This
approach has a long history, starting with the Detroit and Chicago Transportation

BUTTON 9781786435668 PRINT.indd 453 27/04/2022 11:07

 454 TRANSPORT ECONOMICS, 4TH EDITION

Studies of the 1950s, and is still, despite serious limitations, the dominant form of
modeling used today; for example, it is used within the US Government’s Urban
Public Transport Planning System (UPTPS). The stages are:

• First, the trip generation/attraction model (the trip-end model) is used to

forecast the number of trips originating and ending within pre-defined geo-
graphical zones of the study area (these might be countries in international
travel studies or smaller areas within a city in urban land-use studies).
• Second, the total zonal trips are ‘distributed’ between origin–destination
pairs of zones.
• Third, for each flow between different zones, a modal choice model is calcu-
lated to explain the split of traffic between the alternative forms of transport
available.
• Finally, the traffic flows between each pair of zones and by each mode of
transport are assigned to specific routes on the transport network.

In the context of freight modeling, which we discussed in Chapter 10, the

parallel submodels represent transactions, flows, mode changes, and network
assignments, with an industrial location model added to make the links between
industrial activities and freight vehicle movements explicit. A vehicle-loading
submodel is also sometimes added to shift the emphasis from the vehicle to the
consignment.
Figure 12.4 provides an example of the sequential approach as developed by
the regional highway traffic model (RHTM) team in the United Kingdom, and

Planning and Assemble planning data

network data and build network

Car ownership
Car ownership model
Disaggregation/ relationships
aggregation of
matrices
Trip-end model
Trip rates
Scheme matrices for
different time periods
Trip-distribution model Deterrence
functions

Local planning and Route choice

traffic data Traffic assignment criterion

Main RHTM model Input data

Supplementary elements used Computer program

for RHTM procedures Calibration function

Figure 12.4 The United Kingdom’s regional highway traffic model (RHTM)

BUTTON 9781786435668 PRINT.indd 454 27/04/2022 11:07

 TRANSPORT PLANNING AND FORECASTING ­ 455

used to assist specifically in national road planning. It is typical of the series of

sequences used in most other countries, including the United States. Mode choice
submodels are excluded because of the dominant position filled by motorcar
traffic. The approach set out is useful in distinguishing between the data analysis,
computer simulation, and calibration aspects of the overall modeling and fore-
casting process.
The sequential framework may, therefore, be seen as moving from an aggre-
gate overall traffic model to quasi-disaggregate forecasting with each successive
submodel in the sequence acting as a check on the one following. In economet-
ric terms the sequence is recursive. Attempts have been made (for example, in
work forming part of the appraisal of the Greater London Development Plan)
to introduce feedbacks from later to earlier submodels in the sequence on the
grounds that one cannot really forecast, say, mode choice without knowing prob-
able levels of congestion on each route, and this latter knowledge only becomes
available after the assignment stage. A land-use forecasting model that attempts
to describe the effect of changing land-use patterns on transport usually pre-
cedes the transport model sequence. Ideally, because land use is itself partly
affected by transport conditions, there should also be feedbacks from the assign-
ment submodel to land-use patterns, but to date this has proved to be practicably
impossible.
Trip-end submodels are usually estimated on a household basis using either
multi-variate regression techniques or category analysis. The former statistically
relate the number of trips made by households to the socioeconomic characteris-
tics of the households (for example, income, number of residents, car ­ownership,
social status, etc.) and the type of environment in which they are located. The
concentration on the household is now widely accepted as standard practice
to reduce statistical problems that appear when data are grouped excessively
(for example, by geographical zone). Category analysis is simpler, involving the
construction of a multi-dimensional matrix with each dimension representing a
socioeconomic variable stratified into several discrete classes or categories. For
example, households may be divided into four income classes, three car owner-
ship classes, and two location classes, giving in total 24 categories of household,
each with its own average trip-generation level. Forecasts are obtained by predict-
ing the number of households falling into each category at the target date, and
multiplying by the relevant average trip generation rate. The total zonal trips in
the example above would be predicted as:
24
T = ∑ nk rk
(12.1)
t =1

where n is the future number of households in category k, and r is the correspond-

ing trip rate.
Revealed preference approaches to trip distribution models are of two
broad types. Growth factor models involve extrapolating existing patterns of
trips between alternative origins and destinations with projected trip-end esti-
mates acting as constraints on the total number of trips leaving or entering any

BUTTON 9781786435668 PRINT.indd 455 27/04/2022 11:07

 456 TRANSPORT ECONOMICS, 4TH EDITION

individual zone. They are little more than mechanical procedures based upon past
behavior patterns and suffer from inabilities to allow for new zones being created.
They have also gone out of favor because they require a substantial amount of
data input.
The second group, simulation models, are more overtly economic in their
nature. The gravity model is the most commonly used member of this group and
has the attraction of having a precise economic interpretation. The underlying
concept is also widely used in other areas of transport analysis such as when
looking at the importance of transport costs in labor mobility, which will be dis-
cussed in Chapter 13.
Gravity models differ in form, but all exhibit terms reflecting the relative
attractiveness of destinations and terms that measure the effect of impedance
caused by the nature of the transport system. In early work, attractions were
specified simply in terms of the population sizes of the zones of attraction, but
in more recent studies a multiplicity of factors have been included, frequently
varying with the journey purpose under consideration. Similarly, the crude
notion that distance is a full reflection of impedance has given way to the incor-
poration of various forms of generalized cost measures.
The interactive version of the gravity model takes the general form:

Tij = TiTjAiAjBjf(Cij) (12.2)

−−11 −−11
   
Ai i ∑
subject to: Ai A
= ∑ andBBj j ∑
TTj jBBj j ff((CCijij)) and ∑ TTi iAAi i ff((CCijij)) where:
 j j   i i 

Tij is trips between zones i and j;

Ti is the total number of trips originating in zone i;
Tj is the total number of trips destined for zone j;
Cij is the generalized cost of travel between zones i and j;
A and B are ‘fuzz factors’ – sometimes justified as indices of inverse acces-
sibility – to ensure that total trips distributed across the whole study area
originating from i do not exceed Ti and those destined for j do not exceed
Tj.

This doubly constrained model assumes that trip-makers are competing for
a limited number of opportunities in any specific zone, and has clear applica-
tions to modeling the demand for work or school trips where job and educational
opportunities can be assumed to be independent of the transport system. In
many cases only one constraint (either Ti or Tj) is imposed; for example, with
inter-urban freight demand, one is often only interested either in the way move-
ments fan out from a city or depot or in the way they converge on it. Urban non-
work demand models are also often based upon origin-constrained models with
less concern about destinations. On other occasions it may prove necessary to
relax the constraints to facilitate easier fitting of the models, constrained versions
of the gravity model usually requiring specific computer software for calibration.

BUTTON 9781786435668 PRINT.indd 456 27/04/2022 11:07

 457
TRANSPORT PLANNING AND FORECASTING ­

An alternative simulation model considers the opportunities available in different

zones to meet the needs of travelers.
The intervening opportunities model assumes that people try to keep their
trips as short as possible and only lengthen them if nearer destinations do not
prove acceptable to their needs. Individual residents in each zone are assumed
to consider opportunities for the location of their conduct of specific activities
(residence, work, shopping, etc.) at various places, starting from the base zone
and fanning out to other zones in increasing order of difficulty in reaching them.
Each time an opportunity is considered, there is a given, constant chance that it
will be selected. The model takes the general form:

Tij = Ti [exp (– LVj) – exp(– LVj + 1)] (12.3)

where V is the possible destination just considered, and L is a constant represent-

ing the probability of the possible destination being accepted (if considered).
While this type of approach has an intuitive appeal, it does suffer from the
problem that, empirically, L seems to vary with V rather than remaining constant.
Adjustment techniques to correct for the ‘wandering’ of L values are available,
but their use seems to violate the notion of constancy. Empirical studies indicate
that the results obtained using the intervening opportunities model are no better
than those from gravity models. Attempts to develop the opportunities frame-
work by considering competing rather than intervening opportunities (by basing
the underlying probability function on the ratio between the trip opportunities in
a zone and its competing opportunities) offer no improvements in a forecasting
context but complicate the estimation process considerably.
In some instances, as for example with inter-urban road traffic forecasting in
the United Kingdom, there had been a tradition to employ a ‘fixed trip matrix’
which assumes that the total number of trips (that is, the trips generated and
attracted) is unaffected by the new transport scheme. The total traffic, which may
well grow due to changing socioeconomic conditions, is simply redistributed as a
result of the change in transport policy or infrastructure investment. If new traffic
is generated, however, this result may prove distortive.
In Figure 12.5, Demand represents the true aggregate relationship between
the cost of travel and the traffic flow. As a result of, say, a road investment, the
cost of trip-making falls from P to P*, with traffic increasing to F*. Assuming
there has been no shift in demand due to changing income or other socioeco-
nomic factors, the traditional model would assume that demand had remained
fixed at the all-traffic flow – in other words, that the implied demand curve is
Demand*. This means that the benefit enjoyed by the generated traffic (F* – F),
and represented by the shaded area in the diagram, would be excluded from the
subsequent appraisal stage. This is another reason why there are often feedback
links which, in this case, result in the trip-end models being re-run incorporating
some allowance for this generation effect.
Modal split models allocate traffic flows to types of vehicles. In some
cases, for example, with urban freight transport or long-distance international

BUTTON 9781786435668 PRINT.indd 457 27/04/2022 11:07

 458 TRANSPORT ECONOMICS, 4TH EDITION

Cost of travel

Demand*

P*
Demand

F F* Traffic flow
Figure 12.5 The fixed trip matrix

­ assenger transport, one mode so dominates a sphere of transport activity

p
that no mode choice submodel is required, although this is exceptional. The
traditional method of splitting origin–destination traffic flows by mode involves
the use of diversion curves. These show the proportion of traffic likely to favor
a particular mode given its relative cost (or other) advantage over alternative
modes. If we are concerned with two forms of transport, a and b, then a typical
model might be:

Tija 1
= (12.4)
Tij + Tij 1 + e
a b − λ(Cijb −Cija )

This yields a diversion curve of the general form seen in Figure 12.6.
Normally a series of curves are estimated by subdividing the traveling population
(for example, by income) and modes (for example, by service ratios).
In some studies mode choice is modeled prior to distribution using trip-
end or interchange models. The former are often used in highway-orientated
origin-and-destination studies where the emphasis is on car travel with public
transport being treated as little more than a residual to be subtracted from the
trip-end predictions prior to assignments being made. The emphasis on variables
such as car ownership and income, and the general neglect of public transport
characteristics, limits the usefulness of this approach in urban transport plan-
ning. Interchange models are more commonly used in public transport feasi-
bility studies and, consequently, concentrate on comparative time, cost, and
service differentials between competing modes. Models of this type take the
form:

Tijm = α0Xijα1 (FijM)α2 (Fijb)α3 (HijM)α4 (Hijb)α5 (12.5)

BUTTON 9781786435668 PRINT.indd 458 27/04/2022 11:07

 459
TRANSPORT PLANNING AND FORECASTING ­

100%
Mode’s share of i => j journeys by
households with mode available

50%

0 (Cija − Cijb )

Figure 12.6 Diversion curve of traffic between modes ‘a’ and ‘b’

where:

Xij is a matrix of exogenous economic and social variables;

FijM is the financial cost by mode M;
HijM is the time cost by mode M; and
the superscript b indicates the money/time cost by the best mode.

Models of this type have a foundation in economic theory but estimation of

parameters a1 – a5 poses serious statistical problems.
The final submodel, route assignment, compares the travelers’ preferences for
routes with the characteristics of the routes available. Early approaches relied upon
diversion curves like those used in modal choice work but the development of min-
imum-path algorithms combined with improved computing facilities has permitted
the introduction of more sophisticated techniques. A major problem in assignment
is the possible need to constrain traffic flows on each link in the transport network to
the capacity of the link. If the capacity constraint is omitted then an ‘all or nothing’
assignment results and no allowance is made for the congestion that accompanies
high traffic volumes. Further, travelers use all routes, both the initially cheap and the
not so cheap, especially when overall cost differentials are small. The introduction of
capacity constraints permits this to be reflected in the model by adjusting link speeds
(and hence costs) as the assignment proceeds and congestion levels rise.
In summary, the sequential method of forecasting broadly involves develop-
ing a series of mathematical models taking the general form:

Ti = f(Xi); Ti = f(Xj) (12.6a)

Tij = f(Ti, Tj, Cij) (12.6b)

Tijm = f(Tij, CijM, CijM′) (12.6c)

BUTTON 9781786435668 PRINT.indd 459 27/04/2022 11:07

 460 TRANSPORT ECONOMICS, 4TH EDITION

TijMP = f(TijMP, CijMP, CijMP′) (12.6d)

where the prime notation refers to alternative modes (M′) or routes (P′).
While widely used, the sequential modeling approach has serious limita-
tions. Clearly, the series of calculations required to calibrate this set of equations
places tremendous strains on the databases available and, in most studies, large
and expensive surveys are needed to gather the necessary information. Also,
as pointed out above, it is difficult to incorporate the desirable feedback from
assignment and other stages to trip generation so that quality of service vari-
ables are adequately and consistently reflected in each submodel of the sequence.
Statistically, while it is generally possible to test the significance of the individual
models in the sequence, it is not possible to undertake statistical tests of the
overall recursive system. The approach is also deficient in that it does not embrace
the fundamental tenet of travel demand, namely that it is derived from the desire
to participate in some final activity.
One of the reasons why disaggregate economic models have been developed
is to avoid some of these problems and to approach more closely to the basic
decision-making unit, the household (Ben-Akiva and Lerman, 1985).

12.5 Disaggregate Modeling

The substantial data required to satisfactorily calibrate the submodels, and the
difficulty of transferring models once estimated from one data area to another,
combined with dissatisfaction with the basically mechanistic and physical nature
of the sequential approach, has resulted in alternative mathematical approaches
being developed. This second generation of models employing economic-based,
disaggregate methods of travel demand forecasting emphasizes the economic–
psychological influences on travel behavior at the individual household level.
The idea is that households are utility maximizers who, mainly for math-
ematical convenience, are considered to make travel decisions in isolation from
other activities. The emphasis is on short-run decisions rather than long-run
mobility decisions. Small stratified samples of households (about 700 or so)
provide the data input into the models, which tend to be probabilistic, rather than
deterministic, in nature (that is, they forecast the probability of household travel
patterns rather than the average number and type of trips to be undertaken).
A personal disaggregate model of trip-making would be of the general form:

P(f,d,h,m,t) = P(f)P(d/f)P(h/f,d)P(rn/f,d,h)P(r/f,d,h,m) (12.7)

where P(f,d,h,m,r) is the probability that an individual will undertake a trip with
frequency (f) to destination (d) during time of day (h) using mode (m) and via
route (r) out of a choice set comprising all possible combinations of frequen-
cies, destinations, times of day, modes, and routes available to the individual. For
actual planning or assessment, the forecasts produced from such models must be

BUTTON 9781786435668 PRINT.indd 460 27/04/2022 11:07

 TRANSPORT PLANNING AND FORECASTING ­ 461

aggregated up to the level of the geographical zone – it is the inter-zonal level, for
example, which determines the level of public transport demand. While there are
claims that the approach can be used in comprehensive transport planning, its
main role to date has been in policy assessment (for example, looking at carpool-
ing proposals, pollution controls, public transport subsidies, etc.).
Broadly, disaggregate models are characterized by two main features. First,
they explicitly recognize that travel decisions emerge out of individuals’ optimiz-
ing behavior and, if it is pointed out that the final goods consumed as a result of
travel are normal, then at the very minimum the demand for travel ought to be
related positively to disposable incomes and negatively to the prices of transport
services. Second, most have their origins in the ‘attribute theory of demand’ asso-
ciated with Kelvin Lancaster (1966). This approach to human behavior assumes
that people desire to maximize a utility function that has, as its arguments,
commodity attributes rather than the quantities of the actual goods consumed.
In other words, if we represent the amounts of attributes by the vector z, the
amounts of commodities (in this case, travel alternatives) by the vector x, posit a
utility function, U(z), and a production of attribute function, G(x), which reflects
the attributes of different travel alternatives, and assume that potential travelers
are constrained by income, y, and the price of travel, p, then we can reduce the
problem to solving:

max U(z) (12.8)

subject to:

z = G(x), x ≥ 0, p.x ≤ y

The selection of the attributes can also be seen diagrammatically in

Figure 12.7, which looks at a very simple mode choice situation. Potential travel-
ers have a fixed budget, and a choice of three modes (car, train, and bus) which
have different attribute-supply characteristics in terms of speed and economy.
With a given budget the traveler can enjoy the combination of attributes seen
on the frontier – for example, if all trips were made by train then the bundle of
attributes would be ST and ET. Of course, combinations of modes may be made
for the budget. The final selection will depend on the consumers’ preferences for
speed and economy and, since the aim is to maximize utility, this will be at the
point where the highest possible constrained indifference curve (I2) is reached. By
using a combination of train and car, the traveler will enjoy E of economy and S
of speed.
As an example, if one is considering air transport between the United
Kingdom and the United States, then the alternative commodities would be the
different fare packages offered by the airlines, and the attributes of each would
be characteristics such as money costs, speed, period of advance booking, timing,
type of aircraft, in-plane service (food, drink, films, etc.), stop-over regulations,
required length of stay at destination, etc.

BUTTON 9781786435668 PRINT.indd 461 27/04/2022 11:07

 462 TRANSPORT ECONOMICS, 4TH EDITION

Economy

I2

Bus I1
ET
E

Train

Car

0 STS Speed

Figure 12.7 The attribute choices in mode selection

In terms of application, direct attempts were initially made to apply Lancaster’s

theory at the aggregate level by Richard Quandt (1976) and associates in the
context of abstract mode modeling. At the inter-urban level Quandt and Baumol
(1966) developed an abstract mode model for air, bus, and car journeys between
16 city pairs in California using cost and time (both absolute and relative) as the
determining attributes, but the results were inferior to those obtained from more
conventional trip-distribution models.
Howrey’s (1969) early study of air travel out of Cleveland produced an
abstract mode model with correct signs, significant coefficients, and a good
overall fit to survey data. However, while its explanatory power proved statisti-
cally superior to a conventional gravity model it turned out to be inferior in
terms of ex post forecasting quality. Talvitie (1973) concentrated on developing
the framework to handle intra-urban travel, while Baumol and Vinod (1970), by
combining the attributed theory of demand with an inventory theory of goods
handling, adapted the approach to deal with urban freight transport demand.
While some of the calibration problems that were associated with the early
aggregate abstract mode models have been resolved, it has been the introduction
of disaggregate approaches that has marked the greatest advances. The major
advance in this context was the realization that everyone has a different utility
function, partly because of quantifiable differences in their personal characteris-
tics, but also partly because of random factors. While the heterogeneous nature
of the population poses serious problems, the work of Domenich and McFadden
(1976) in modeling the random factors, and especially their theoretical work on
justifying the use of ‘multinomial logit models’, forms the basis of much modern
disaggregate analysis. The forms of disaggregate model that have been developed
along these lines, and the range of applications to which they have been applied,
are many.

BUTTON 9781786435668 PRINT.indd 462 27/04/2022 11:07

 TRANSPORT PLANNING AND FORECASTING ­ 463

Most of the recent work on disaggregate modeling has centered on math-

ematical and calibration issues and there is no intention of taking discussion of
such matters any further in this book. The mathematical complexity stems from
the fact that one is looking at discrete choices – that is, whether a person makes
a trip or not – and this involves complications not normally encountered when
considering continuous functions.

12.6 Interactive and Stated-preference Modeling

While the sequential and disaggregate approaches to transport demand analy-

sis concentrate on developing sophisticated mathematical simulations, recently
there has been a growth of interest in ‘behavioral realism’, and an emphasis
on ‘understanding the phenomenon’ (Dix, 1977). This is sometimes called the
activity-based approach (McNally and Rindt, 2008), because it has sought to
embrace a richer, holistic framework in which travel behavior is analysed as a
daily or multi-day pattern of behavior related to life-styles and participation in
various activities.
The idea is that the demand for travel is derived from the activity patterns of
individuals and households. The basic idea has much to do with the concept of
time-geography developed by Torsten Hägerstrand (1970). Underlying this, is the
idea that an individual’s choice of a specific activity pattern is viewed as a solu-
tion to an allocation problem involving limited resources of time and space. In
this sense actual behavior is not that useful, but rather there is a need to put more
emphasis on the constraints that limit people’s actions.
These approaches to modeling have also been tied in with the greater use of
stated-preference techniques that we have already encountered in Chapter 6 in
the context of contingency evaluation procedures. This latter approach, which, to
adopt Kroes and Sheldon’s (1988) definition, is ‘a family of techniques which use
individual respondents’ statements about their preferences in a set of transport
options to estimate utility functions’, is often claimed to be helpful when:

• there is insufficient variation in revealed preference data to examine all vari-

ables of interest;
• there is a high level of correlation between the explanatory variables in
a revealed preference model, making statistical estimation of parameters
difficult;
• a radically new technology or policy takes the analysis outside of the realms
where current revealed behavior is relevant; and
• variables are not easily expressed in standard units (for example, when the
interest is in the effects on demand of less-turbulent travel by air).

The aim of interactive modeling is to develop models that get closer to the
essential decision process underlying travel behavior. Rather than simply incor-
porate variables such as household status in mathematical models because the

BUTTON 9781786435668 PRINT.indd 463 27/04/2022 11:07

 464 TRANSPORT ECONOMICS, 4TH EDITION

statistical ‘explanation’ of the model appears to be improved, interactive mode-

ling seeks to explain why status affects travel behavior. Theoretically, travel is seen
as one of a whole range of complementary and competitive activities operating
in a sequence of events in time and space. It is seen to represent the method by
which people trade time to move location in order to enjoy successive activities.
Generally, time and space constraints are thought to limit the choices of activities
open to individuals. The technique is still far from fully developed, but it has been
applied only in a relatively limited number of small-scale forecasting studies (for
example, by Peter Jones, 1978 in the United Kingdom to school bus operations;
by Fowkes and Preston, 1991 to new local rail services; and by Phifer et al., 1980
in the United States to energy constraints).
The emphasis of interactive models is upon the household (or individual)
as the decision-making unit. According to Heggie (1978), ideally an interactive
model should exhibit six main properties:

• It should involve the entire household and allow for interaction between its
members.
• It should make existing constraints on household behavior quite explicit.
• It should start from the household’s existing pattern of behavior.
• It should work by confronting the household with realistic changes in its
travel environment and allowing it to respond realistically.
• It should allow for the influence of long-term adaptation.
• It should be able to tell the investigator something fundamental that he/she
did not know before.

In general, the approach is typified by a small sample and careful survey tech-
niques, often involving such things as ‘board games’, such as the ‘household activi-
ties travel simulator’ (HATS) developed by the Oxford University Transport Studies
Unit, or other visual aids, usually computer-based simulations these days, to permit
participants to fully appreciate the implications of changes in transport policy upon
their own behavior. In a sense it represents an attempt to conduct laboratory experi-
ments by eliciting responses in the context of known information and constraints.
The HATS approach, for example, was to confront a household with a
map of the local area together with a 24-hour ‘strip representation of colored
pieces’ showing how current activities of the household are spread over space and
throughout the day. Changes to the transport system were then postulated (for
example, reduced parking availability in the local urban center) and the effects
on the household’s activities throughout the day were simulated by adjustments
to the strip representation. In this way, changes in the transport system could
be seen to influence the entire 24-hour life pattern of the household, and appar-
ently unsuspected changes in ‘remote’ trip-making behavior can be traced back
to the primary change. It makes clear the constraints and linkages that may affect
activity and transport choices. The emphasis is on the micro unit with the aim of
being able to develop simple models that permit much clearer insights into the
overall effects of transport policy. By asking respondents to trace the effect of

BUTTON 9781786435668 PRINT.indd 464 27/04/2022 11:07

 TRANSPORT PLANNING AND FORECASTING ­ 465

changes in transport provision on the entire set of activities undertaken during

a day (or week), information on important travel intentions can be seen and the
relationships between travel and non-travel activities become explicit.
More recent studies have adopted more sophisticated experimentation pro-
cedures, often involving computers, which provide for greater flexibility and easier
interaction with those being ‘interviewed’. Examples of programs of this genre
include ALBATROS (A Learning-Based Transportation Oriented Simulation
System), the first computational process model of the complete activity schedul-
ing process that could be fully estimated from data, and TRANSIMS, an attempt
to replace the entire traditional transport paradigm, which has an activity-based
front end linked to a population synthesizer and integrated with a micro simula-
tion of modeled travel behavior.
While this aspect of the approach has been refined, important technical
issues remain regarding using the information gathered from stated-preference-
type experiments for forecasting. There is still, for example, much to be learned
about why some households give strategically biased responses; there are difficul-
ties in handling habit, inertia, and hysteresis in an experimental framework. At
a more technical level, John Bates (1988) points to our lack of knowledge about
the error structures associated with stated-preference data and the problems of
pooling data across individuals.
In contrast to the more traditional revealed-preference schools, advocates
of this approach, however, point to both the specific recognition that travel is a
derived demand and the fact that transport policies have qualitative, as well as
quantitative, effects on people’s lives. In the longer term, when operational models
are more fully developed, the framework may offer the much-sought-after basis
for integrating land-use and transport planning assessments. In the short term,
the approach has offered useful insights and a method for cross-checking the
validity of conventional statistical analysis of behavioral data.
As techniques of data collection and analysis have improved, so more
recently hybrid stated-revealed preference techniques have been used to look at
such things as multi-modal trip choices in Madrid (Cascajo et al., 2017), choice of
mode and route by shippers in the Pacific Northwest, (Train and Wilson, 2007),
and local travel behavior (Rudloff and Straub, 2021).

References

Bates, J. (1988) Econometric issues in stated preference analysis, Journal of Transport

Economics and Policy, 22, 59–70.
Baumol, W.J. and Vinod, H.D. (1970) An inventory theoretic model of freight transport
demand, Management Science, 16, 413–41.
Ben-Akiva, M. and Lerman, S. (1985) Discrete Choice Analysis: Theory and Application to
Travel Demand, MIT Press.
Berechman, J. and Gordon, P. (1986) Linked models of land use–transport interactions:
a review, in B. Hutchinson and M. Balty (eds), Advances in Urban Systems Modelling,
North Holland.

BUTTON 9781786435668 PRINT.indd 465 27/04/2022 11:07

 466 TRANSPORT ECONOMICS, 4TH EDITION

Cascajo, R., Garcia-Martinez, A., and Monzo, A. (2017) Stated preference survey for
estimating passenger transfer penalties: design and application to Madrid, European
Transport Research Review, 9, 1–11.
Dix, M.C. (1977) Report on investigations of household travel decision making behaviour,
in E.J. Visser (ed.), Transport Decisions in an Age of Uncertainty, Martinus Nijhoff.
Domenich, T. and McFadden, D. (1976) Urban Travel Demand: A Behavioural Analysis,
North Holland.
Flyvbjerg, B. (2007) Cost overruns and demand shortfalls in urban rail and other
­infrastructure. Transportation Planning and Technology, 30, 9–30.
Flyvbjerg, B., Skamris Holm, M.K., and Buhl, S.L. (2002) Underestimating costs in public
works projects: error or lie? Journal of the American Planning Association, 68, 279–95.
Fowkes, A.S. and Preston, J. (1991) Novel approaches to forecasting the demand for new
local rail services, Transportation Research Part A, 25, 209–18.
Friedman, M. (1953) The methodology of positive economics, in M. Friedman (ed.),
Essays in Positive Economics, University of Chicago Press.
Hägerstrand, T. (1970) What about people in regions? Papers of the Regional Science
Association, 24, 7–21.
Heggie, I. (1978) Putting behaviour into behavioural models of travel choice, Journal of the
Operational Research Society, 29, 541–50.
Howrey, E.P. (1969) On the choice of forecasting models for air travel, Journal of Regional
Science, 9, 215–24.
Jones, P.M. (1978) School hour revisions in West Oxfordshire: an exploratory study using
HATS, Technical Report, Oxford University Transport Studies Unit.
Kroes, K.P. and Sheldon, R.J. (1988) Stated preference methods: an introduction, Journal
of Transport Economics and Policy, 22, 11–26.
Lancaster, K.J. (1966) A new approach to consumer theory, Journal of Political Economy,
74, 132–57.
Lay, M.G. (2005) The history of transport planning, in K.J. Button and D.A. Hensher
(eds), Handbook of Transport Strategy: Policy Institutions, Pergamon.
Mackinder, I. and Evans, S. (1981) The predictive accuracies of British transport studies in
urban areas, TRRL Report SR 699.
McFadden, D. (2001) Economic choices, American Economic Review, 91, 351–78.
McNally, M.G. (2008) The four-step model, in D.A. Hensher and K.J. Button (eds),
Handbook of Transport Modeling, 2nd edn, Elsevier.
McNally, M.G. and Rindt, C.R. (2008) The activity-based approach, in D.A. Hensher and
K.J. Button (eds), Handbook of Transport Modeling, 2nd edn, Elsevier.
Phifer, S.P., Neven, A.J., and Hartgen, D.T. (1980) Family reactions to energy constraints,
Paper presented at the Transportation Research Board Annual Meeting, Washington
DC.
Pickrell, D.H. (1989) Urban rail transit projects: forecast versus actual ridership and costs,
Report, US Department of Transportation.
Quandt, R.E. (1976) The theory of travel demand, Transportation Research, 10, 411–13.
Quandt, R.E. and Baumol, W.J. (1966) The demand for abstract transport modes: theory
and measurement, Journal of Regional Science, 6, 13–26.
Rudloff, C. and Straub, M. (2021) Mobility surveys beyond stated preference: introduc-
ing MyTrips, an SP-off-RP survey tool, and results of two case studies, European
Transport. Research Review, 13, https://doi.org/10.1186/s12544-021-00510-5.
Stopher, P.R. (1992) Use of an activity-based diary to collect household travel data,
Transportation, 19, 159–76.
Stopher, P.R. (2008) Survey and sampling strategies, in D.A. Hensher and K.J. Button
(eds), Handbook of Transport Modeling, 2nd edn, Elsevier.
Talvitie, A.P. (1973) A direct demand model for downtown work trips, Transportation, 2,
121–52.

BUTTON 9781786435668 PRINT.indd 466 27/04/2022 11:07

 TRANSPORT PLANNING AND FORECASTING ­ 467

Train, K. and Wilson, W. (2007) Estimation on stated-preference experiments constructed

from revealed-preference choices, Transportation Research Part B, 42, 191–203.
UK Department of Transport (1977) Transport policy, Policy Document, DoT.
UK House of Commons Committee of Public Accounts (1988) Road planning, Report,
HMSO.
UK Ministry of Transport (1963) Traffic in towns, Report, HMSO.
US Federal Highways Administration (1973) Urban origin–destination surveys, Report,
UDDOT.
Webster, F.V., Bly, P.H., and Paulley, N.J. (eds) (1988) Urban Land-use and Transport
Interaction, Avebury.
Weiner, E. (2008) Urban Transportation Planning in the United States: History, Policy, and
Practice, Springer.

BUTTON 9781786435668 PRINT.indd 467 27/04/2022 11:07

 13 Transport and Economic Development

13.1 Transport’s Role in Development

Economists have long been concerned with assessing the links between changes
in the transport sector and the evolving pattern of economic development within
the area that it serves. While the importance of transport in economic growth
and development has never seriously been questioned, its exact role and influence
have been subjected to periodic reappraisals. The subject is made more difficult
because the impacts of transport are not always intended, and thus can have
unforeseen consequences. The underlying problem is, however, a more general
one, in that our general understanding of what causes economic development is
poor. Charles Kindleberger (1958) made the point over 60 years ago:

We have suggested that there is no agreement on how economic development pro-

ceeds and have implied that this is because the process is not simple. There are many
variables involved, and there is a wide range of substitutability among ingredients –
land, capital, and the quality and quantity of labor, and technology can substitute
for one another, above certain minima, although there are at the same time certain
complementary relationships among them. The will to economize and organization
are probably the only indispensable ingredients. For the rest, none are necessary, and
none sufficient.

The interest in the topic, though, is not purely an academic matter. There
are public concerns regarding regional disparities within countries and with the
economic performances and differences between levels of national economic
prosperity, and especially so between the ‘North’ and the ‘South’. This has
brought forth efforts to stimulate growth in lagging economies by investing in
infrastructure, including transport, and in enhancing the performance of existing
infrastructure. The creation of the World Bank, various regional development
banks, such as the Inter-American Development Bank, and a plethora of other
bi-lateral and multi-lateral organizations is a tangible institutional manifestation
of these concerns at the macro level. This is setting aside any consideration of
the use of transport investment programs as part of a Keynesian-style macro-
economic stimulus package; it is about dynamics rather than comparative statics.
The form and scale of the measures that could be taken, and indeed their
general desirability, are matters of significant practical interest. In consequence,
while Kindleberger is still correct, in that our knowledge is in many ways very
limited, efforts to clarify the situation continue. The work on development eco-
nomics and the role that transport can play in the economic development process
is extensive.

BUTTON 9781786435668 PRINT.indd 468 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 469

To provide some context for our discussions we begin by considering the

theories of economic development and how transport fits into them. Without
an understanding of how economic development takes place, it is not possible
to study transport’s contribution. We then move to look at more specific issues,
beginning with a long-standing debate in economics about the role that public
transport infrastructure can play in stimulating economic productivity. This has
largely been an empirical debate that was initiated in the 1980s regarding ques-
tions about the role infrastructure policies should play in macroeconomic policy.
This leads on to questions regarding the ripple effects on economic performance
of individual transport initiatives.
We then move on to consider the more macro problems of stimulating
­economic growth in the Third World, before proceeding to consider the prob-
lems of formulating a common transport policy to foster the economic growth
ambitions of the member states of the European Union. At a more micro level
there are also questions concerning the ways in which transport provision can
stimulate economic growth within certain parts of a country or for a given urban
area.
While the discussion of transport and economic development has been sec-
tionalized for expositional convenience, it is important to emphasize that in prac-
tice considerable trade-offs may be necessary between, say, devising a transport
policy to stimulate national growth and one designed to assist the development
of specified under-developed regions. The poorest countries, especially, often feel
they must attempt to increase their national income. Indeed, if one were to accept
the ‘growth pole’ approach to economic development, then national growth
is generated by concentrating effort in several strong regional centers. Hence,
inter-regional inequality of growth is an inevitable concomitant and condition
of growth itself. Consequently, although it may be possible to design a national
transport strategy or investment program that assists in the maximization of
national economic growth, it may need modifying to ensure that an acceptable
degree of equity is retained among the different regions of the country.

13.2 Economic Growth Theory and Transport

First, a little background. In the neoclassical economic framework, which we

shall consider in more detail below and which largely represents an extension
of eighteenth-century trade theory, transport is not really considered at all. The
existence of optimal transport services is assumed, and, if they are lacking in the
short term, then it is assumed that a perfectly functioning market would soon
supply the optimal amount. The more recent endogenous approach to economic
development assumes imperfect markets that need not result in socially optimal
outcomes. Again, with reference to transport, these imperfections may entail a
lack of adequate financial markets to construct appropriate infrastructure and
imperfect markets in making use of the roads, ports, etc. that are available, owing
to either market or government failures.

BUTTON 9781786435668 PRINT.indd 469 27/04/2022 11:07

 470 TRANSPORT ECONOMICS, 4TH EDITION

Traditionally, it was argued by Adam Smith and others that transport

exerted a strong positive influence on economic development, and that increased
production could be directly related to improved transport. In the United
Kingdom context, for example, over 150 years ago, Robert Baxter (1866) argued
that ‘[r]ailways have been a most powerful agent in the progress of commerce, in
improving the conditions of the working classes, and in developing the agricul-
tural and mineral resources of the country’. Perhaps the strongest advocate of the
positive role of transport, however, is Walter Rostow (1960) who, in accounting
for economic growth, maintains that ‘[t]he introduction of railroads has histori-
cally been the most powerful single indicator to take-offs. It was decisive in the
United States, France, Germany, Canada and Russia’.
A broader brush approach is that of Andersson and Stromquist (1988),
who claim that all the major transitions in the European economic systems were
accompanied (or initiated) by major changes in transport and communications
infrastructure. Four main transport and logistics revolutions can be distinguished:

• the period from the thirteenth century, in which water transport emerged as
a new logistic system connecting cities along the rivers and coastal areas (the
Hansa economy);
• the period from the sixteenth century (the Golden Age), characterized by a
dramatic improvement in sailing and sea transport and by the introduction
of new banking systems which stimulated trade to the East Indies and West
Indies (with Lisbon, Antwerp, and Amsterdam as major centers);
• the middle of the nineteenth century, marked by the Industrial Revolution,
in which the invention of the steam engine generated new transport modes,
thereby creating new market areas such as North America; and
• from the 1970s, which is marked by increased information and flexibility;
just-in-time systems and material requirements planning have evolved within
this framework.

Positive linkages between transport provision and economic development can

be divided between the direct transport input and indirect, including multiplier,
effects. Good transport offers low shipping costs that have permitted wider markets
to be served and the exploitation of large-scale production in an extensive range
of activities. In the past, for example, Hunter (1965) postulated a causal linkage
between low-cost transport and economic growth. The Industrial Revolution was
successful because of a prior revolution in transport technology. Similarly, Owen
(1964) argued that a widening of domestic markets through improved transport
services is a pre-requisite for national economic development. Further, most unde-
veloped countries are, for a variety of geographical, economic, and historic reasons,
dependent upon international trade and an expansion of this trade is an essential
pre-requisite for growth. In these circumstances the provision of efficient port
facilities will, according to this school of thought, positively assist development.
The indirect effects stem from the employment created in the construction
of transport infrastructure and the jobs associated with operating the transport

BUTTON 9781786435668 PRINT.indd 470 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 471

Exhibit   Employment implications of the United States’ federal highway system

The Interstate highway system was initiated in 1956 with the passage of the Federal Highway
Act. The National Highway System (NHS) Designation Act of 1995 added $5.4 billion to the
United States’ federal-aid highway program targeted to the NHS. This motivated analysis of
the economic impacts of the network.
The NHS consists of 260,000 kilometers of roads that serve major population centers,
international border crossings, ports, airports, public transportation facilities, and other
inter-modal transportation facilities and major transportation destinations; meet national
defense requirements; and serve inter-state and inter-regional travel. It constitutes 4
percent of public roads, but carries more than 40 percent of highway traffic and 70 percent
of truck traffic.
Assessing its implications in national labor market studies, the US Federal Highways
Administration (FHWA) looked at both the direct effects and the indirect. Regarding the
former, it estimated that $1 billion (in 1995 prices) of federal-aid highway program spending
in 1996 would nationally support about 7,900 full-time, on-site highway construction jobs.
The table provides the aggregate picture derived from the FHWA studies: aggregate
employment supported per $1 billion of NHS investment generated approximately 42,100
jobs in the United States. This is with the caveat that the estimates refer to ‘gross’ job
creation, meaning that the employment impacts of alternative investments have not been
considered in this broad level of analysis.

Direct, indirect, and induced employment impacts of federal aid

Employment type Jobs per $1 billion federal spending

Direct 7,900
Indirect 19,700
Induced 14,500
Total 42,100

See also: G. Michaels (2008) The effect of trade on the demand for skill: evidence from the
Interstate highway system, Review of Economics and Statistics, 90, 683–701.

services. Further, there may be multiplier effects stemming from the substantial
inputs of iron, timber, coal, etc. required to construct a modern transport system
and which, at least in the context of development in the nineteenth century, were
supplied by indigenous heavy industries. Transport also often provided some
initial experience of business for many industrialists of the period. The potential
multiplier effects for Third World countries today are likely to be substantially
less given the growth (itself a function of improved transport) of international
trade and tied development aid. Additionally, the technical expertise required to
engineer and plan modern transport systems is often unavailable in less developed
countries and must be bought from more advanced nations.

BUTTON 9781786435668 PRINT.indd 471 27/04/2022 11:07

 472 TRANSPORT ECONOMICS, 4TH EDITION

The broad theories that underlie these views of the role of transport have
evolved over time, and can be divided into two broad conceptual approaches,
each seeing transport as impacting on development in somewhat different ways.

Neoclassical Economics

Economists have struggled over many years to explain why some regions’ econo-
mies, or indeed national economies, perform better than others. Traditional neo-
classical, long-run macroeconomic growth theory of the type espoused by Robert
Solow (1957) and others argues that per capita income in a region or country
depends on local factor endowment, the savings rate, and the impact of exog-
enously determined ‘technical progress’. Transport is seen as one of the inputs in
this context, on a par with other industries.
More formally, an exogenous growth model of the Solow type can be formu-
lated by taking a simple Cobb–Douglas production function of the form:

Y = A(t)K1–bLb (13.1)

where:

Y is net national product;

K is the capital stock;
L is the labor stock; and
A is the level of technology.

The fact that A is a function of time (t) indicates the standard neoclassical
assumption that technology only improves over time for reasons external to the
model.
Where regional, as opposed to macro, neoclassical growth models differ
from this to some extent is that they treat labor and other factor supplies as both
dependent upon the internal demographics and an inherent resource base of
a region, and dependent upon factor movements between regions. While most
national economies experience relatively small flows of factors of production
between them, the much laxer border restrictions within a region tend to foster
such flows. This makes factor growth within a region more elastic.
Basically, this spatial mobility of factors of production can help to compen-
sate for an initial shortage of a factor in a region. Hence a specific region may
grow when there is capital abundance by ‘importing’ labor from other regions.
This, combined with the natural growth in labor supply within the region, can
lead to growth through more efficient use of the complementary capital stock,
although in the long run there will be convergence in per capita income between
the regions, albeit at a higher average wage rate. One can consider parallel move-
ments of capital, but this is less relevant to air transportation.
Figure 13.1 offers a simple illustration of the larger point. There are
two regions, A and B. Region A enjoys high income and low unemployment

BUTTON 9781786435668 PRINT.indd 472 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­473

Region A

U– : Y+

Capital Labor

U+ : Y–

Region B

Region A

U– : Y+

Capital Labor

U+ : Y–

Region B

Figure 13.1 Simplified theories of migration

(represented as U– : Y+) whilst region B is the mirror image of this. There are
decreasing returns to factors of production, including capital. The neoclassical
economic model assumes that, with zero costs of migration (including zero trans-
portation costs) and a homogeneous labor force, labor will move from B to A
seeking work and higher pay, whereas, on the assumption of uniform commercial
risk across regions, capital will move from A to B where it can be combined with
abundant, cheap labor to maximize returns.
Wages will fall in region A, unemployment will increase, and the return on
capital will rise as the labor supply grows and capital becomes scarcer. The addi-
tional capital and the decreasing size of the labor force in region B will push down
the marginal return on investment and concurrently push up wages. The process
continues until labor costs and unemployment levels are equalized. This equaliza-
tion is achieved in a world of zero transportation costs and full information about
opportunities.
The problem with this way of looking at regional development is that it
relies on relatively strong assumptions. Factor movement is not frictionless: there
are distance, information, and money costs. Also, factors of production are not
homogeneous, and some are more mobile than others. Young, skilled, highly

BUTTON 9781786435668 PRINT.indd 473 27/04/2022 11:07

 474 TRANSPORT ECONOMICS, 4TH EDITION

Exhibit   Railroads and canals in the United States’ economic development

Robert Fogel won the 1993 Nobel Prize in economics for his work on cliometrics: the use
of quantitative methods in history. His early work focused on the role of the railroads in
the economic development of the United States. He questioned the established view in the
1960s that the railroads were the main driver of economic growth in the 1800s. His main
argument was that, while railways had important implications for the economy, most of
the history had been focused on correlations and not on causality. He attacked the ‘axiom
of indispensability’ – the notion that railroads were crucial to the economic growth of the
United States in the 1890s – by focusing on two major corollaries: (i) railroads brought
great savings in transportation costs; and (ii) their construction generated demands vital
for the development of domestic industry. In doing this, he paid attention to the degree to
which railways reduced the costs of transporting four agricultural products – wheat, corn,
beef, and pork – that made up 80 percent of agricultural shipments. The alternative to rail
would at the time have been waterborne transport.
To complete this task, Fogel made several assumptions. The work was aggregate in nature,
while there were clear regional variations in the efficiency of the railroad sector and in their
strategic importance. The costs of rail transport were also largely embedded in the railroad
rates, while those of canals were more complex and some approximations had to be made.
The quality of service varied, with railroads having a natural advantage in terms of reliability
over canals in the winter when the latter froze, and sometimes in the summer in drought
conditions. Canals were simply not a tenable alternative in areas where production was
at higher altitudes, due to the necessity to install cost- and time-expensive lock systems in
order to reach it. Adjustments for such things as cargo losses in transit and capital costs
funded by government, and trans-shipment costs, were added to the canal calculations.
Fogel calculated the costs of both inter-regional and intra-regional distribution and
estimated the costs that would have been incurred without the railroads, the difference
being the ‘social saving’. The main water routes considered were the Great Lakes and Erie
Canal route, the Mississippi route, and the inter-coastal route which carried the bulk of
traffic. He estimated that savings in transporting these commodities amounted to less than
2 percent of GNP for that year. A hypothetical extension of the formulae to all commodities
hauled by rail yielded a saving well below 5 percent of GNP.
The second half of his study concerned measuring the demand represented by railroad
construction during the ‘take-off’ period of 1840 to 1860, the purchase of rails and
the impact for the iron industry being his primary focus. The results again countered
conventional wisdom. He found that the addition to pig iron production required for rails
between 1840 and 1860 only amounted to less than 5 percent of the output of domestic
furnaces. Similar findings were discovered for other industries supplying the railroads. His
overall estimate was that railroad purchases from the rest of the manufacturing sector
accounted for only 3.9 percent of production.
See also: R.W. Fogel (1964) Railroads and American Economic Growth: Essays in Econometric
History, Johns Hopkins Press.

BUTTON 9781786435668 PRINT.indd 474 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 475

educated workers, for example, tend to be more mobile than those lacking these
characteristics. Investors consider returns on their capital relative to risk incurred;
they are not risk-neutral. There is also the issue of technical progress that is
simply treated as a sort of residual and uncontrollable factor in the neoclassical
framework. This seems excessively fatalistic.
In addition, empirical analysis hardly gives solid support for the neoclassi-
cal idea. Testing the validity of the alternative theories, in the absence of easily
quantifiable counterfactuals, has frequently involved looking at secondary evi-
dence, and at evidence shedding light on whether there is convergence in the
economic growth paths of regions or, at the macro level, nations. The empirical
question that is explored becomes one of whether there is convergence in regional
economic growth rates in, generally, per capita income or productivity, as is an
outcome of the neoclassical model.
The extensive empirical analysis that has emerged has been assisted by the
improved data made available in recent years, as well as new models, enhanced
econometric techniques, and better understandings of how to measure conver-
gence. There has been the development of the concept of β-convergence measures
that has allowed a more rigorous analysis of economic convergence than the more
traditional σ-convergence measure that looks for changes in standard statistical
indicators of dispersion – normally the variance.
The estimation of possible β-convergence involves a mean-reversion calcula-
tion, which occurs if there is a negative relationship between the growth rate of
income per capita and the level of initial income. Much of this type of analysis
has been carried out at the national level, but it, and the more limited body of
analysis that has been conducted at the regional level, does offer some insights
into the validity of the idea of endogenous economic growth.
Much of the early work on spatial economic convergence relied upon aggre-
gate, national data sources and focused on σ-convergence measurements. The
findings indicated that labor productivity, and per capita income in the world,
was converging in the long run and thus provided support to the neoclassical
growth theory. The difficulty with this work was that the datasets only contained
countries that had already become industrialized, and even for those there were
periods of divergence. Improved data subsequently indicated a lack of any
general economic convergence.
The more recent work that has made use of β-convergence measures,
and embraces a number of subnational studies, tends to find little support for
overall convergence, and ipso facto the neoclassical model. Robert Barro and
Xavier ­Sala-i-Martin (1991), in several studies that, for example, have exam-
ined the economies of American states and European Union regions, find that
while there is evidence of convergence in per capita income, it is slow: about
2 percent per annum and well below the 12 percent or so that neoclassical
theory would suggest. These are also conditional convergence measures that
allow for homogeneity between, for example, the regional economies within
a European economy, but diversity between countries that would suggest
potential differences in steady-state growth rates for the regions within them.

BUTTON 9781786435668 PRINT.indd 475 27/04/2022 11:07

 476 TRANSPORT ECONOMICS, 4TH EDITION

Hence, there is little support in this body of work for the exogenous growth
idea.

‘New Economic Growth Theory’

An alternative to the neoclassical model attempts to circumvent some of these

intellectual problems. In endogenous growth models, growth over time entails
increasing returns to scale for a region. A proportionate increase in labor and
capital gives rise to more than proportionate gains in output. This, when embod-
ied into the ‘New Growth Theory’ or ‘New Economic Geography’, is consist-
ent with divergence in economic growth along the traditional lines of Gunnar
Myrdal’s ideas of ‘circular-and-cumulative causation’, albeit for somewhat differ-
ent reasons.
The situation can be illustrated by returning to Figure 13.1. Taking the
initial starting positions for the two regions, the endogenous growth approach
argues that not only will equalization of real wages and employment levels not
be attained, but that there may be cases where they diverge further. Labor mobil-
ity between regions A and B may be impeded by the various costs of migration
(embracing social and search costs as well as simple financial costs) and heteroge-
neity in the labor market (the jobs available in region A not being compatible with
the skills and knowledge of labor in region B).
Equally, capital does not always move from region A to region B because of
the higher returns due, not to an actual shortage of capital, but rather to the lesser
uncertainty that is to be found in regions that already have a high level of prosper-
ity and a pool of complementary skilled labor. While the earlier regional models
focused largely on divergent growth rates in the context of traditional-style indus-
tries, the more contemporary form of the theory pays particular attention to the
endogenous growth that occurs in regions which have an established highly skilled
workforce and the ability to further develop their knowledge-based industries.
The relevance of the emergence of the new thinking, and the empirical
analysis that has been found to support at least parts of it, is that it gives a role
for government in regional economic policy. If there is circular-and-cumulative
causation caused by scale effects of one kind or another, combined with imperfec-
tions in the mobility of factors of production, very broadly defined to embrace
knowledge, then policies can be devised to compensate. Part of this policy may
embrace appropriate transportation elements, and especially those relating to
infrastructure provision.

13.3 Transport Infrastructure Investment and Economic

Productivity

Large transport infrastructure investments, such as the United States Federal

Highway System, have often been credited with having major impacts on the eco-
nomic efficiency of a country. Fogel’s work questioned whether that was always

BUTTON 9781786435668 PRINT.indd 476 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 477

the case, but the issue resurfaced in the late 1980s in the context of the role of
public investment, when David Aschauer (1989) produced work suggesting that
the return on public infrastructure investment, of which transport works form the
largest element, was exceeded by a large margin that found for private investment
between 1949 and 1985 in the United States. Analysis by Biehl (1991), relating to
European regions, came to the same basic conclusion.
Aschauer’s analysis took a simple macro national time-series production
function for the United States of the general form:

Y = f(L,K,J) (13.2)

where:

Y indicates output;
L is labor;
K is the stock of private capital; and
J is the shock of public capital.

Aschauser used a Cobb–Douglas specification to estimate the output elastic-

ity of public sector investment. Later work sometimes modified the production
framework by adding additional variables, for example Prud’homme (1996) used
a function of the general form Y = f(L,K,J,F), where F reflected employees in
the public sector. Other studies have used a cost, rather than a production, func-
tion of the form C = f(IP,Q,PK), some have used varying levels of aggregation,
and some have taken cross-sections of data, often by region, rather than a single
national time series. Table 13.1 provides a listing of the results of the main studies.
These do not refer specifically to transport investments, but transport is by far the
major component in each.
The early studies indicated that when they were conducted, infrastructure
expansion would significantly enhance productivity; a 1 percent increase in
the public capital stock would raise total factor productivity by 0.39 percent.

Table 13.1 
Summary of estimated output elasticities of public infrastructure investments

Study Level of aggregation Output elasticity of public capital

Aschauer (1989) National 0.39

Holtz-Eakin (1988) National 0.39
Munnell (1990) National 0.34
Costa et al. (1987) States 0.20
Eisner (1991) States 0.17
Munnell (1990) States 0.15
Mera (1973) Regions 0.20
Duffy-Deno & Eberts (1989) Urban areas 0.08
Eberts & Fogarty (1987) Urban areas 0.03

Source: Button (1998). This paper contains the full references to the studies cited.

BUTTON 9781786435668 PRINT.indd 477 27/04/2022 11:07

 478 TRANSPORT ECONOMICS, 4TH EDITION

Exhibit   Infrastructure investment and economic productivity

While its physical form has changed over time, investment in public sector transportation
infrastructure has, from the days of Adam Smith, been considered a stimulus to economic
development. Defining economic development and agreeing on a common form of
quantification, however, is difficult, and economists have deployed a variety of definitions
over the years. Finding an acceptable measure of infrastructure is equally difficult.
The neoclassical economic work of David Aschauer, although its findings were
subsequently questioned, stimulated serious economic debate on the subject. His
empirical work was stimulated by the slow-down of economic productivity growth in
many advanced economies in the late 1970s and early 1980s. Simple data plotting in
the figure, which relates the amount of national productivity growth to the domestic
product devoted to public investment between 1973 and 1985, suggests a strong positive
relationship at the national level. And public transportation infrastructure constitutes
a significant part of any country’s non-military public capital. What Aschauer was
interested in was a more detailed examination of the United States, and especially whether
over time declining public investment had impacted on the United States’ economic
productivity.

3.5
Japan
3.0

2.5
Productivity growth

France West Germany

2.0 United
Kingdom Italy

1.5 Canada

1.0

0.5 United States

0
0 1 2 3 4 5 6
Public Investment/Gross Domestic Product

More detailed analysis adjusting for other things that affect economic performance makes
use of an aggregate production function. This involves relating a country’s real output of
goods and services over time to its technical progress, its employment, its stock of non-
residential capital, and a measure of public capital stock. Making fairly standard economic
assumptions (for example, a Cobb–Douglas production function and allowing for possible
economies of scale and congestion effects), and assuming the government provides services
directly to the private sector, allows the specific functional form to be specified. This can
then be manipulated to see how private sector productivity has changed with respect to
public investment.

BUTTON 9781786435668 PRINT.indd 478 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 479

Estimation of the equations makes use of annual American data for 1949–85, taken from a
variety of government sources. As seen in the table, the analysis involves looking at several
subperiods. While the average growth in total factor productivity was 1.5 percent from
1950 to 1985, when the capital stock was growing at just over 3.0 percent, there were
variations. In the initial period to 1970, factor productivity increased by 2.0 percent,
accompanied with capital stock growing by 4.1 percent, but the latter only rose by
0.8 percent in the second part of the period when the public capital stock increased by
1.6 percent, the increase in total factor productivity in the first part of the early 1980s
being particularly slow when the public capital stock only increased by 0.7 percent.

Subperiod Annual growth in total factor Change in on-military

productivity (%) public capital stock (%)

1950–85 1.5 3.0

1950–70 2.0 4.1
1971–85 0.8 1.6
1981–85 0.7 0.7

Aschauer does not claim that public expenditure on infrastructure is the only thing
influencing changes in the short-term economic productivity of a country. It is seen in the
study as one of several factors, and others may include expenditure on R&D and the costs
and availability of various factor inputs, and especially fuel and migrant labor. But it would
seem to be an important one.
See also: D.A. Aschauer (1989) Is public infrastructure expenditure productive? Journal of
Monetary Economics, 23, 177–200.

Later studies, and especially as the level of aggregation moved down from
national to state to urban levels, brought this into question. As Ronald Fisher
(1997) put it:

The research is all over the board and somewhat inconsistent in its results as to
whether investments in public services can increase levels of economic development.
For example, the relationship between transportation investments and development
is generally positive (but not overwhelming), and it is statistically significant only half
of the time.

Several technical points have also been raised about the way this sort of
analysis has been conducted (Gramlich, 1994):

• While the various studies have thrown up positive correlations between eco-
nomic performance and the quantity of infrastructure, the direction of cau-
sality is not immediately clear. Wealthier areas, or times of greater economic
growth, may simply provide more resources to be put into public projects.
Testing for causality is difficult, and unfortunately the econometric work that

BUTTON 9781786435668 PRINT.indd 479 27/04/2022 11:07

 480 TRANSPORT ECONOMICS, 4TH EDITION

has been done is not conclusive. Looking at tourism, a transport-intensive

activity, for example, studies of Spain by Balanguer and Cantavella-Jorda
(2002) and of Turkey (Gunduz and Hatemi-J., 2005) have suggested that
there is a causal link running from tourism to economic development, but
contrarily Narayan (2004) finds the linkage to be in the other direction for
Fiji.
• The term ‘infrastructure’ is flexible; there is no agreed definition and taking
accountancy data may disguise important measurement, qualitative, and
definition factors. In the United States, for example, Aschauer used a four-
category core subset of the nine US Bureau of Labor Statistics infrastruc-
ture categories, whereas other studies have employed the full set.
• The way that infrastructure is managed, used, and priced may be as impor-
tant as the level of investment involved. In terms of impact, therefore,
account needs to be taken of the short-term levels of utilization, mainte-
nance, and so on, in addition to the stock of, and investment in, capacity.
• While transport is the largest element of most nation’s infrastructure, it is not
the only one, and ideally one should separate it out from energy, education,
health, defense, and so on.
• As Morrison (1993) points out, ‘[a] clear consensus about the impacts of
infrastructure investment has as yet been elusive, at least partly because dif-
ferent methodologies generate varying results and implications’. From an
econometric perspective, for example, Sturm and De Haan (1995), looking
at American and Dutch data series on the economic effects of public
investment, found them to be neither stationary nor co-integrated, sug-
gesting the estimation techniques used in many time series studies are often
inappropriate.
• From a purely a priori perspective, some of the studies have thrown up rates
of return that are outside of normal expectations, and well outside of what
have been found in most individual micro projects.
• There are also issues regarding how economic development is to be meas-
ured. Gallen and Winston (2021) used 2017 US National Household Travel
Survey data to develop a 17-equation general equilibrium model to examine
the effects of government increases in publicly funded transportation infra-
structure on individuals’ accessibility to jobs and firms’ accessibility to
workers, on the availability, price, quality, and variety of consumer goods
and services, on the intensity of competition among firms and their produc-
tivity, and on economic growth attributable to agglomeration economies,
incorporating effects on households and firms. They focused on permanent
changes and economic welfare. By incorporating shopping and commut-
ing time wedges, they examined the costly time to build infrastructure, and
efficient government pricing and investment policy, which are important for
assessing the effects of infrastructure spending on welfare and GDP. The
evidence shows that such spending can result in greater firm productivity
and reductions in commuting and shopping–travel time. But they also show
that it is important to distinguish an infrastructure spending policy’s effect

BUTTON 9781786435668 PRINT.indd 480 27/04/2022 11:07

 481
TRANSPORT AND ECONOMIC DEVELOPMENT ­

on GDP. Welfare improvements in GDP may overstate the improvements in

an economy from increased infrastructure spending. Specifically in the case
study, a 5 percent increase in transportation infrastructure financed by tax-
payers generated a $79 billion increase in GDP, but a much lower welfare
gain of $25 billion. The divergence occurs because GDP may increase
without increasing welfare when households are indifferent to a marginal
increase in work. Given that transportation infrastructure acts as a comple-
ment to labor, it is also possible to increase labor even as welfare decreases
because of increased taxation.

13.4 The Multiplier Impacts of a Transport Investment

While at the abstract macro level there seems, from the empirical evidence, to be
only a very general link between transport provision and economic development
at best, it is at the more meso level that investment decisions are made. Major
transport investments are generally neither simple in engineering terms, nor in
their economic implications. Chapter 11 dealt with the microeconomic matters of
assessing the pros and cons of a transport investment; here we look at transport
provision in terms of its implications for stimulating economic activity at a higher
level of aggregation.
A new airport, which for specificity is what is considered, goes through
several stages from its planning to becoming fully operational and heavily used.
Each of these phases generates its own type of income multiplier effect on the
region. Figure 13.2 offers a simple diagram tracing out the temporal and spatial
impacts together with some indication (the size of the arrows) of their respective
magnitudes. It should be noted that the monikers attached to the various levels of
multiplier sometimes differ in the literature, as indicated below:

Time
Perpetuity effect

Tertiary effect

Secondary effect

Primary effect

Multiplicand
0 Geographical range

Figure 13.2 The various economic multipliers associated with an airport investment

BUTTON 9781786435668 PRINT.indd 481 27/04/2022 11:07

 482 TRANSPORT ECONOMICS, 4TH EDITION

• Primary (or direct). The primary multiplier stems from the income that is
associated with the multiplicand inherent in the construction of the facil-
ity, and the rounds of expenditure that emanate as part of that money
are recycled through the local economy. The size of the local multiplier is
often tempered in the case of an airport, if there is a need for significant
inflows of labor, raw materials, and equipment to plan and construct the
facility. Leakages of this kind are often substantial because new airports
or major extensions are rare events in aggregate let alone in a particular
region. As a result, there is seldom either adequate local expertise or equip-
ment available. Even when local resources are used, there may be crowding-
out effects if these are taken from other sectors of the regional economy
and, thereby, reduce the multiplier effects of these sectors. Added to all
this are frequent implementation delays due to planning, input shortage,
and other factors that can produce small or even negative labor and output
responses to increases in government investment in the short run (Leeper
et al., 2010).
• Secondary (or supplier). Once an airport is operational, it pumps money into
the local economy through the staff that it directly employs and the activities
of the airlines that make use of the facility. This income, in turn, has multi-
plier effects on the regional economy. Airports can be major employers but
there is a largely bi-modal distribution in the labor force. While airports do
employ many highly skilled and, thus, generally highly paid workers, many
jobs are unskilled or semi-skilled (for example, in terms of drivers, aircraft
cleaners, workers in concessions, and baggage handlers). Equally, while the
airlines using the airport may maintain a staff to handle ticketing and air-
craft maintenance, as well as have aircrew stopping overnight at local hotels,
the numbers are generally relatively small compared to the land-take and
investment in infrastructure and operational equipment.
• Tertiary (or induced). The tertiary multiplier is the one that normally attracts
most attention in the economic development literature. It is concerned with
the amount of economic activity drawn to the region by the existence of
an airport, and with the subsequent ripple effects that result as this pumps
income into the area. These regional economic effects can be substantial.
For example, Memphis Shelby Airport, the major United States hub for
FedEx, is surrounded by warehouse and distribution facilities – one of the
largest concentrations in the world. These facilities house products as varied
as ­just-in-time surgery and orthopedic devices, home decor products, and
DVDs, all of which are shipped from Memphis to destinations around the
world. But elsewhere, and at much smaller facilities, the presence of air ser-
vices is important for many industries, not only in their cargo needs but more
often in terms of allowing their employees and customers easy access to their
facilities and markets. High-technology industry makes extensive use of air
transport, as do tourists.
• Perpetuity. While the tertiary multiplier effect can often be pronounced
when an airport is suitably located and an extensive range of destinations

BUTTON 9781786435668 PRINT.indd 482 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 483

offered, its effect largely stems from a movement up the local production
function. The airport itself, or extensions to an existing facility, is justi-
fied because there is an existing demand for air transportation services.
The ­perpetuity effect is often, although not always, associated with the
­development of a large airport that shifts the regional production func-
tion upwards. Essentially it changes the structure of the regional economy.
For example, many islands in the Caribbean and the Mediterranean have
had their entire economies changed, usually from fishing and agriculture
to tourism, as the result of the construction of a large airport. Air travel,
after the construction of airport infrastructure, is also vital to the growth
of tourism in larger regions, such as the coasts of Spain and of Miami.
Additionally, as alluded to earlier, high-technology corridors have emerged
at locations that had either been farmland before or dependent on more
traditional industries.

Some general estimations by the Economic Policy Institute of the multiplier

effects associated with an additional $1 million of expenditure on various modes
of transport are seen in Table 13.2. These would suggest that, in general, expendi-
tures on transport activities such as transit, trucking, and tourism have the largest
multiplier effects.
Looking at less-aggregate cases, however, reveals that empirical evidence has
not always been very helpful in sorting out the diverse ideas concerning the links
between specific transport initiatives and economic development. This is seen,
for example, in the summaries of studies conducted in industrial countries in the
1980s set out in Table 13.3. One of the difficulties is that economic development
itself is not easy to define, and as can be seen from the table, several measures can
be used in addition to simply the impact on local GDP. Additionally, the nature of
the transport investments or other initiatives have implications for its influence on
the wider economy. In other words, generalization concerning multiplier effects is
highly context-specific.

Table 13.2 
United States employment multipliers per $1 million in final demand, all
private sector industries

Direct Supplier Induced Total indirect

Air transport 2.50 4.03 2.91 6.93

Rail transport 2.64 4.27 2.64 6.91
Water transport 1.06 5.80 2.43 8.22
Truck transport 5.01 5.42 4.16 9.58
Transit & ground passenger transport 8.04 8.84 4.93 13.77
Pipeline transport 1.46 3.18 1.45 4.63
Scenic & sightseeing transport
& support activities for transport 4.99 6.71 4.39 11.11
Couriers & messengers 6.90 5.83 5.01 10.83
Warehousing & storage 9.18 4.88 5.37 10.25

BUTTON 9781786435668 PRINT.indd 483 27/04/2022 11:07

 484 TRANSPORT ECONOMICS, 4TH EDITION

Table 13.3 Summary of findings looking at transport and development in industrialized countries

Study Geographical scale Infrastructure Conclusions

Botham (1980) 28 zones (UK) Changing nature of Small, centralizing on

highways employment
Briggs (1981) Non-metropolitan Provision of Presence of Interstate
counties (US) highways highway is no guarantee of
county development
Clearly & Thomas Regional level (UK) New estuarial Little relocation but changes
(1973) crossing in firms’ operations
Dodgson (1974) Zones in north (UK) New motorway Small effect on employment
Eagle et al. (1987) 87 counties (US) New highway No increased employment
Evers et al. (1987) Regional level High-speed rail Some effect on (Netherlands)
employment
Forrest et al. (1987) Metropolitan Light rapid transit Property blight – good areas
(US) for urban renewal
Judge (1983) Regional level (UK) New motorway Small economic impact
Langley (1981) Highway corridor Highway Devalued property in areas
(US)
Mackie et al. (1986) Regional level New estuarial Small overall effect
crossing
Mills (1981) Metropolitan areas Interstate highways No significant effect on
(US) location patterns
Moon (1986) Metropolitan areas Highway crossings Existence of ‘interchange
(US) villages’
Pickett (1984) Local districts Light rapid transit Properties close to the (UK)
line benefitted
Stephandes (1990) 87 counties (US) New highway Could affect employment –
expenditure depends on county
Stephandes et al. 87 counties (US) New highway Some positive association
(1986) expenditure with employment
Watterson (1986) Metropolitan area Light rapid transit Modest growth in land use
(UK)
Wilson et al. (1982) Regional level (US) Existing highways Transport affects locations
decisions but not
development

Source: Button et al. (1995). This paper contains the full references to the studies cited.

There are also some serious caveats that can often lead to over-estimations using
income or employment multipliers at the regional level. Multipliers were derived
as part of demand-side macroeconomics at a time when factor supply constraints
were not an issue. In practice, regions, and especially those where the economy
has been performing poorly, often have supply constraints that limit the size of
multiplier effects. Most empirical work tends to use macro-parameters that may
not be relevant for the region under consideration.
While multipliers are often useful for considering the impacts on the region
enjoying the new air transportation services, they only consider the gross effects.
The initial injection of resources is often from outside the region; the multipli-
cand has to come from somewhere, and there are opportunity costs associated

BUTTON 9781786435668 PRINT.indd 484 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 485

with resources drawn in during successive multiplier rounds, and this means that
there is an opportunity cost involved. In the case of an airport intended to open
up a tourist area, for example, this may stimulate more tourism in the aggregate,
but some of the visitors will be attracted from alternative destinations. As with
any activity that allows trade, airports have both a traffic generation and a traffic
diversion effect.
In addition, while there may be positive multiplier effects, the closure of an
airport or airline service may produce a downward spiral as income circulation is
reduced. This can lead to retention of a facility or service that may not be justi-
fied. Investment, therefore, should be viewed over a long horizon, and simply to
assess the short-term effects of, say, developing a small airport to stimulate local
residential development or tourism may not be enduring and leave the area with
significant stranded costs. Reliance on one industry that is in turn dependent on
another is also a risky development strategy.
This causal view of transport and economic development has come under
question in recent years. The cliometrics (the systematic application of eco-
nomic theory, econometric techniques, and other mathematical methods to
the study of history) work of Fogel (1964) in the United States, for example,
offers evidence that American growth in the nineteenth century would have
been quite possible without the advent of the railways – waterways supplying
a comprehensive transport system at comparable costs. The view that the rail-
ways were the motivating force behind American economic development has
given way to a weaker position, namely that good transport permits economic
expansion.
Economic development is, thus, now generally seen as a complex process
with transport permitting the exploitation of the natural resources and talents of
a country; it is, therefore, necessary but not sufficient for development. Transport
can release working capital from one area that can be used more productively as
fixed capital elsewhere, although a necessary prior condition is the existence of
suitable productive opportunities in potential markets. Public infrastructure in
this sense should be set in the context of the availability of private capital: many
parts of the world, for example, would not benefit from more transport infra-
structure because of the lack of private resources.
Looked at from a slightly different perspective, improved transport can help
overcome bottle-necks in production and thus further foster economic expansion.
This is, for example, the underlying position taken by Roger Vickerman (1987)
when examining the regional impacts of the Channel Tunnel between the United
Kingdom and France. A difficulty, of course, if this is true, is that the bottle-neck
may be some distance from the region and superficially appear unconnected with
it. Accepting this caveat, the basic view of this school of thought is that transport
is seen more as a facilitator than as a generator of development.

In many developing countries the inadequacy of transport facilities is one of the

major bottlenecks to socio-economic development and a national integration. Often
the lack of transport makes it difficult to introduce other social infrastructure such

BUTTON 9781786435668 PRINT.indd 485 27/04/2022 11:07

 486 TRANSPORT ECONOMICS, 4TH EDITION

as education and medical services. The dissemination of the modern techniques and
inputs of agricultural production and the linking of agriculture to other sectors of the
economy through the market is hampered by the absence or inadequacy of transport
facilities. As a result of these and other factors, the productivity of agriculture – the
dominant sector in developing economics – is deplorably low. (Ahmed et al., 1976)

While the approaches sketched out above ascribe a positive role to transport
in economic development, albeit in different ways, there is a feeling among some
economists that an excess of scarce resources sometimes tends to be devoted
to transport improvements. As with any scarce input it is possible to define an
optimal provision of transport to facilitate development so that resources being
drawn from other activities are not wasted. Excess provision can have an eco-
nomic cost such as reduced or missed opportunities; the aim is thus to optimize
‘crowding out’.
At a given point in economic development, a country requires a critical level
of transport provision so that its growth potential is maximized – hence there is
an optimum transport capacity for any development level. It has been argued,
however, that there are economic forces that tend to lead to an excess of transport
provision (especially high-cost infrastructure) at the expense of more efficient and
productive projects. More specifically, Wilson (1966) pointed to the lumpiness
of transport capital that, together with its longevity and associated externalities,
makes it particularly difficult to estimate future costs and benefits. Consequently,
decisions to devote resources to transport are not easily reversible or readily
corrected.
Albert Hirschman (1958) highlighted the political acceptability of trans-
port, arguing that the sector attracts resources simply because it is difficult for
mistakes of an economic nature to be proved even after major projects have
been completed. Also, development planners tend to be mainly concerned with
allocating public investment funds, and it is, therefore, natural that they should
claim transport, communications, energy, drainage, etc. as being of over-riding
and fundamental importance. Further, given the industrial composition of
wealthier developed countries with an established heavy industrial base, tied aid
for transport schemes has a firm attraction. Those adopting this rather skeptical
approach to the role of transport, therefore, accept that an adequate basic trans-
port system is an obvious sine qua non for modern economic development but
question whether the opportunity costs involved in further improving transport
are necessarily justified.

13.5 Transport Economics in Less Developed Countries

Transport investment forms a major component of the capital formation of less

developed countries, and expenditure on transport is usually the largest single
item in the national budget. Up to 40 percent of public expenditure is devoted to
transport infrastructure investment with substantial supplements coming from
outside international agencies such as the World Bank or with direct bi-lateral

BUTTON 9781786435668 PRINT.indd 486 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 487

assistance from individual countries. At one level it is important to know whether

this is, in aggregate terms, the most practical and efficient method of assisting the
poor countries of the world, while at another level it is necessary to be able to
assess the development impact of individual transport schemes.
Broadly, transport may be seen to have four functions in assisting economic
development:

• First, it is a factor input into the production process, permitting goods and
people to be transferred between and within production and consumption
centers. Because much of this movement is between rural and urban areas
it permits the extension of the money economy into the agricultural sector.
• Transport improvements can shift production possibility functions by alter-
ing factor costs and, especially, reducing the levels of inventory tied up in the
production process.
• Third, mobility is increased, permitting factors of production, especially
labor, to be transferred to places where they may be employed most
productively.
• Transport increases the welfare of individuals, by extending the range of
social facilities to them, and provides superior public goods such as greater
social cohesion and increased national defense.

Transport economists have made significant contributions in assessing in

detail the role that transport may play in assisting economic development in
Third World countries. At the microeconomic level they have developed tech-
niques of project appraisal that permit a more scientific assessment of the costs
and benefits of individual transport projects to be conducted.
Many of the techniques of investment appraisal employed in the developed
parts of the world (see Chapter 11) are applicable in Third World conditions, but
local situations often require changes of emphasis. This is not surprising consid-
ering these techniques were devised to look at transport systems based almost
entirely upon mechanical modes, while head-porterage and canoes still account
for the greatest proportion of goods movement in many less developed countries.
Basic data are also often not so readily available or reliable in the Third
World as in most developed countries, thus limiting the precision of any analysis.
Nevertheless, the development of investment appraisal techniques of the type
set out by Ian Little and James Mirrlees (1974) permit consistent analysis across
investment alternatives both within the transport sector and between the trans-
port sector and other areas of economic activity. Such techniques emphasize
the importance of estimating appropriate shadow prices for both inputs into
transport and the benefits derived from it. In particular, the shortage of foreign
exchange suffered by many less developed countries is highlighted, while it is
recognized that higher levels of under-employment and unemployment require
adjustments to the wage costs of labor. (The shadow price of labor, estimated
as any lost production by diverting it from elsewhere in the economy, is usually

BUTTON 9781786435668 PRINT.indd 487 27/04/2022 11:07

 488 TRANSPORT ECONOMICS, 4TH EDITION

negligible, plus an allowance for the disutility associated with the work in the
transport sector.)
At the macroeconomic level economists have pointed to the general influ-
ence that appropriate transport planning can have in assisting overall economic
development. While it may be argued that ideally one should expand transport
provision to balance developments elsewhere in the economy, this is not always
possible. The balanced growth approach maintains that if transport services are
inadequate, then bottle-necks in the economy will curtail the growth process, but
if the services are excessive this is both wasteful, in the sense that idle resources
could be earning a positive return elsewhere in the economy, and can become
demoralizing if the anticipated demand for transport does not materialize rela-
tively quickly. Hirschman takes a somewhat different view, arguing that the rela-
tionship between economic development and the provision of social overhead
capital, such as transport, is less flexible than members of the balanced growth
school believe.
In Figure 13.3, the horizontal axis shows the provision and cost of social
overhead capital (which is normally provided by the public sector and will
embrace transport as a major component), while the vertical axis measures the
total cost of direct productive activities (which are normally undertaken on purely
commercial criteria). The balanced growth approach assumes that DPA output
and SOC activities should grow together (that is, along the growth path repre-
sented by the ray from the origin), passing through the various curves from a to d
representing successively higher amounts of DPA/SOC output.
Hirschman, however, argues that less developed countries are in practice
not in a position to follow such a path – partly because of the lack of necessary
expertise to ensure the balance is maintained and partly because of inherent
indivisibilities in the social overhead capital schemes available. Consequently,
growth is inevitably unbalanced and may follow one of two possible courses:

Total cost of
DPA output

C1

B1 C

B
c

A b
B2
A1
a
0 SOC availability and cost

Figure 13.3 Balanced and unbalanced growth paths

BUTTON 9781786435668 PRINT.indd 488 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 489

one based upon excess capacity of SOC (that is, path A => A1 => B => B2 =>
C), the other upon a shortage of SOC (that is, path A => B1 => B => C1 =>
C). If a strategy of excess SOC capacity is preferred, it is hoped that this will
permit DPA to become less expensive and encourage investment in that sector.
Alternatively, with the second approach, DPA expansion occurs first and DPA
costs will rise substantially. Therefore, considerable economies will be realized
through the construction of more extensive SOC facilities. The actual effective-
ness of the alternatives depends upon the strength of the profit motive in the
DPA sector, and the responsiveness of the public authority in the SOC sector to
public demand.
The type of transport provision most suited to developing economies is often
of as much importance as the aggregate level of provision. Many developing
countries tend to spend scarce development funds on prestige projects to dem-
onstrate visually their capacity to emulate the performance of more developed
nations; in other words, X-efficiency is sacrificed for a modern, if superficial,
image. More critical is the way in which funds are spent on internal transport pro-
vision and whether there are advantages in concentrating limited capital resources
in either the road or rail modes.
The appropriateness of different modes often depends upon the geographic–
demographic nature of the country. Most less developed countries may be catego-
rized as one of the following:

• densely populated tropical lands;

• tropical lands with low population density;
• mountainous, temperate lands with a low overall density of population but a
concentration on a coastal plain or altiplane; or
• thinly populated desert areas with population concentrated along irrigated
channels.

The appropriateness of different transport modes changes according to the

type of country under consideration, thinly populated, tropical lands having dif-
ferent transport problems to highly urbanized countries with high population
densities.
While the railways were important in the development of nineteenth-
century economies and characterized colonial development in many countries,
the emphasis in recent years has switched to the provision of adequate road
infrastructure. This is particularly true in areas where a skeleton of roads already
exists and resources can be devoted to improving and extending an established, if
rudimentary, network. This approach may be especially fruitful if it links isolated
agricultural communities both with each other and with the more advanced areas
of the economy. Sixty years ago, Millard (1959) argued that, unlike economically
advanced countries, in Third World nations ‘the benefits from road construction
are almost entirely in the form of new development from traffic which the new
road will generate’. The effect is not purely on immediate output but can stimulate
a propensity for further development.

BUTTON 9781786435668 PRINT.indd 489 27/04/2022 11:07

 490 TRANSPORT ECONOMICS, 4TH EDITION

Wilson strongly supported road development in Third World countries for

this reason, arguing that ‘[i]nvestment options might usefully be analyzed in terms
not only of their direct economic pay-off but also in terms of their influence on
attitude’ and that ‘[t]he educational and other spill-over effects of road transpor-
tation appear to be greater than those of other modes of transport. This is espe-
cially significant at low levels of development’. Having said this, however, there is
a danger if integrated planning is not pursued that while improved road facilities
may stimulate the agricultural economy, the new links between rural and indus-
trial–urban areas could lead to increased polarization in the spatial economy with
an enhanced geographical, as well as sectional, dualism resulting.
In terms of transport, improved port and shipping facilities permit less
developed countries to export their products to wider markets and gain foreign
exchange, although, as we see later, there are some dangers here. Since the demand
for shipping services is derived from that for the final product, the benefits to less
economically advanced countries from reducing maritime shipping costs can be
illustrated by looking at the quantities of exports from a Third World country
and the imports into the market of a higher-income economy.
Figure 13.4 shows a back-to-back diagram where Si and Di are the supply
and demand schedule for the commodity in the developed country and Sj and
Dj are the supply and demand curves in the less economically developed nation.
Demand for imports and supply of exports is obtained by subtracting horizon-
tally the domestic supply from demand, the demand for imports (exports) at each

Si Di

De
a b
P*m Se
f Z
c
d e
C D E F

P*e
A B

Sj Dj

PT1 W DT

X Y
PT2
Import to i 0 Exports from j

Figure 13.4 Welfare gains from improved maritime services

BUTTON 9781786435668 PRINT.indd 490 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 491

price being the difference between the quantity supplied and demanded assuming
domestic and import commodities are perfect substitutes. De and Se in Figure
13.4 are derived in this fashion; the vertical difference between these curves then
represents the demand for shipping shown as DT. (If shipping charges were zero,
for example, then the free trade equilibrium would be Z.) Suppose actual shipping
rates were PT1, then at that rate the price of imports from country i confronting
country j is seen to be Pm* (the cost, insurance, and freight (CIF) price) while the
cost of exports to country j would be seen in country i to be Pe* (the free on board
(FOB) price). Country i would then import an amount ab equal to country j’s (the
less developed country’s) exports of AB.
There is now an improvement in shipping services; this may take the form
of better port facilities or more efficient ships or it may be administrative (for
example, relaxation of high conference shipping rates). The effect is that ship-
ping costs fall to PT2, resulting in exports from j rising to CF to match the higher
imports of cf into the developed country i. Trade has expanded for the less devel-
oped country. The net benefit of this trade to the two countries is represented by
the shaded areas in the figure. Area adc is the extra consumption enjoyed by the
developed, importing country as a result of the fall in the CIF price while bef is
a positive production effect resulting from a contraction of i’s relatively high-cost
industry. The areas ADC and BEF are the symmetrical benefits to the less devel-
oped country. (Interestingly the sum of these benefits can be measured directly as
the area WXY under the demand curve for shipping services.)
While it can be demonstrated that improved shipping facilities in our example
can aid development, it should be noted that the analysis is crucially dependent
upon the elasticities of demand for goods in developed and under-developed
countries, and the relative costs of supply. This often poses serious problems for
less developed countries, as pointed out by the United Nations Conference on
Trade and Development (UNCTAD) in 1969, with the less developed nations
often paying much of the costs of transport, that is:

For many of the world’s agricultural products, on which developing countries rely
for much of their export earnings, supply elasticities are low in the short run … .
Although overall demand elasticities for most of these commodities are also low, the
elasticity of demand facing the individual supplier or the whole group of suppliers in
a single country is likely to be relatively high, unless that country is the only source of
supply, and there is no ready substitute for the commodity … . The supplier in these
cases therefore normally bears the bulk of the transport costs.

In reality, the world is also a little more complicated, with trade involving not
simply the production and transport sectors but also various systems of interna-
tional finance and tariff barriers in effect.
One of the practical problems experienced by many less developed countries
has been the fact that shipping lines have combined in conferences to regulate
prices and thus have often led to shipping costs, which are often borne by the
less developed countries, being higher than in a free market situation. There are
also arguments that the existence of non-pricing competition within conferences

BUTTON 9781786435668 PRINT.indd 491 27/04/2022 11:07

 492 TRANSPORT ECONOMICS, 4TH EDITION

produces a much higher quality of service (and ipso facto cost) than would
prevail without collusion and that this is again detrimental to Third World coun-
tries. Empirical evidence produced by Devanney et al. (1975), looking at trade
between the United States’ east coast and Chile, Columbia, Ecuador, and Peru,
suggests that the conference system on these routes pushed up shipping rates by
about $20 per ton in 1971, most of which would have been borne by the poorer
countries.
It may not seem surprising that in these circumstances UNCTAD negotiated
a Code of Conduct for Liner Conferences that allocated maritime traffic on a
40:40:20 basis, with 40 percent of the trade allocated to the merchant marine of
each of the trading nations and 20 percent to cross-traders. This was intended to
give under-developed countries the chance to reap some of the financial rewards
from shipping and exert a more immediate influence on their own development.
In 1978, for example, Third World countries were in the disadvantaged position
of only having 8.6 percent of the world dead-weight tonnage of shipping but
generating over 30 percent of the bulk cargoes and over 90 percent of the tanker
cargoes.
In the short term, lack of capacity prevented some nations from enacting the
full implementation of the Code. Zerby (1979) estimated, for instance, that of the
26 less developed countries he studied, only nine had merchant fleets large enough
to handle 40 percent of their exports in 1975 and only 15 had sufficient capacity
to handle 40 percent of imports. Attempts to expand the fleets of less developed
countries to fulfil 40 percent of the market would, therefore, have resulted in
excess capacity in the fleets of the developed countries, but more importantly,
given the imbalance in the volume of imports and exports, attempts to meet
40 percent of shipping demand both into and out of Third World nations would
lead to a 50 percent excess capacity within their own fleets. Zerby, therefore, felt
that a rigid adherence to the 40:40:20 principle was likely to be an extremely costly
method of reducing the developing countries’ dependence on conference services.
In practice it had limited effect, and efforts to reduce the costs of shipping to
developing nations moved in other directions (Premti, 2016).

13.6 The Transport Policy of the European Union

Several national groupings have emerged in both the economically developed and
the economically less developed world. These entail countries that have come
together into loose economic unions with the aim of fostering their common
economic interests. The European Union (this term is used throughout for
simplicity; the title has changed several times) is one example, and the United
States–Mexico–Canada Agreement and the African Continental Free Trade Area
are others.
Each grouping has its own priorities and has set about achieving these in
its own way. The development of transport policies within the framework of the
European Union provides some indication of the importance that is increasingly

BUTTON 9781786435668 PRINT.indd 492 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 493

being attached to transport by these groupings and to some of the problems of

creating pan-national transport policies. (Indeed, the challenges are often similar
to those when trying to develop a transport strategy in a national federal struc-
ture.) The European case also provides a useful illustration of the problems of
devising an international transport policy designed primarily to foster economic
growth.
The objective of the European Union is ‘to promote throughout the
Community a harmonious development of economic activities, a continuous and
balanced expansion, an increase in stability, an accelerated raising of the standard
of living, and closer relations of member states’. A ‘Common Transport Policy’
(CTP) was important in facilitating the attainment of this aim. The difficulties of
agreeing on how to achieve this from a political-economy perspective is illustrated
in Figure 13.5. Before integration, there were several access points that allowed
interactions between two countries (either side of the dashed line) with internal
movements combining to influence location, traffic, and production. Most move-
ment was internal, as seen in the bottom segment of the figure. After integra-
tion brought about by the CTP, there was more international trade that affected
levels of production, the routes used for transport, and the relative importance
of domestic and international markets. The outcome was difficult to foresee, and
although aggregate economic welfare would rise, the ultimate distribution of the
welfare across locations and industrial sectors was uncertain.
Individual member states also each had their own set of transport objectives.
These required modifying to conform with commonly agreed aims and goals with
a context of each country having its own sets of institutions and policy tools
which would require standardization. Many countries in Europe had traditionally
used transport subsidies to protect specific industries or regions, but this may run
counter to the Union’s objectives of increasing overall economic efficiency or be
thought to result in undesirable redistributions of welfare.
Problems with the creation of a transport policy began early. The initial
Rome Treaty of 1957 contained an entire chapter on transport, although it limited
itself initially to movement of freight by road, rail, and inland waterways. It was
not until 1961, however, that a memorandum appeared setting out clear objec-
tives, and not until the following year that an Action Programme was published.
The emphasis of these initiatives was on removing obstacles to trade posed by
the institutional structures governing transport, and to foster competition once
a level playing field of harmonized fiscal, social, and technical conditions had
been established. That it took over 40 years to make significant progress towards
a substantive policy is in part due to the nature of European geography and the
underlying transport market, although continued insistence by nation states on
pursuing their individual agendas did not help.
Examination of a map of the European Union provides insights on some of
the problems of devising a CTP. Ideally, as we have seen, transport functions most
effectively on a hub-and-spoke basis with large concentrations of population and
economic activity located at corners and in the center, and with the various trans-
port networks linking them. The central locations act as markets for transport

BUTTON 9781786435668 PRINT.indd 493 27/04/2022 11:07

 494 TRANSPORT ECONOMICS, 4TH EDITION

Before Economic Union

International
Country A Country B
Domestic Domestic
Few ‘gateways’
Document checks
Regulated prices/market access Bi-lateral license Regulated prices/market access
National licensing Split tariffs National licensing
National establishment Different ‘gauges’ National establishment
National labor laws Different labor laws National labor laws
National technical standards Different laws of establishment National technical standards
Split tariffs Split tariffs

After Economic Union

International
Country A Country B
Domestic Domestic
Open borders
European licenses
Competitive markets Competitive markets
Open access to infrastructure
National/EU licensing National/EU licensing
Through tariffs
EU establishment EU establishment
Different ‘gauges’
EU/national labor laws EU/national labor laws
Coordinated labor laws
National/EU technical standards National/EU technical standards
EU laws of establishment
EU licenses EU licenses
Coordinated EU infrastructure
EU/national infrastructure policy EU/national infrastructure policy

Figure 13.5 Shifts in the European Union’s international and domestic transport from
1960

services in their own rights, but also as interchange and consolidation points for
traffic between the corner nodes. In many ways the United States fits this model
rather well, but the European Union never has. When there were six members, the
bulk of economic activity was at the core, with limited growth at the periphery.
The various enlargements over the years have added to the problems of serving
peripheral and often sparsely populated areas. The geographical separation of
some states and the logical routing of traffic through non-member countries,
together with the island nature of others, posed further problems.
The common policy was also not initiated with a clean slate: member states
had established transport networks and institutional structures that could not
rapidly be changed even if a common set of principles could have been estab-
lished. At the outset, countries such as France and West Germany carried a sig-
nificant amount of their freight traffic by rail, whereas others, such as Italy and
the Benelux nations, relied more on road transport. The resultant differences were
also not simply physical (including variations in railway gauges, vehicle weight
limits, and different electricity currents), they also reflected fundamental differ-
ences in the ways transport was viewed.
At a macro, political-economy level there are two broad views on the way
transport should be treated. Following the Continental philosophy, the objective
is to meet wide social goals that require interventions in the market through regu-
lations, public ownership, and direction. This approach particularly dominated
much of twentieth-century transport policy thinking in Continental Europe. Its
place was taken by a wider, but not complete, acceptance of the Anglo-Saxon

BUTTON 9781786435668 PRINT.indd 494 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 495

approach to transport policy. This treats the sector as little different from other
economic activities. Transport provision and use should be efficient in its own
right, with efficiency normally best attained by making the maximum use of
market forces. Of course, the extremes of either of the approaches never existed;
it has been a matter of degree. Even in countries such as the United Kingdom,
which was a bastion of the Anglo-Saxon ideology, there existed extensive regimes
of regulation and control, and large parts of the transport system were in state or
local government ownership.
From a more analytical perspective, the situation may be seen in terms of
the ways efficiency is viewed. The approach until the 1970s was to treat transport
efficiency largely in terms of maximizing scale efficiency while limiting any dead-
weight losses associated with monopoly power. Most transport infrastructure was
enjoying economies of scale that could only be exploited by coordinated and, ipso
facto, regulated, often subsidized, development and in many cases state-owned.
Many aspects of operations were also seen as potentially open to monopoly
exploitation and hence in need of oversight. This situation changed. From a prag-
matic perspective, the high levels of subsidies enjoyed by many elements of the
transport sector became politically unsustainable. Economists began to question
whether the regulations deployed were achieving their stated aims. Government
failures, it was argued, were often larger than the market failures they were trying
to correct.
New elements also came into play in the 1970s. Attitudes towards environ-
mental intrusion, for example, changed as part of a wider effort to improve the
overall environment and fulfil larger, global commitments on such matters as
reducing emissions of global warming gases. Local environmental effects were
largely left to individual countries, but as the implications of regional and global
environmental intrusions have become more widely appreciated, so the Union’s
transport policy has become proactive in these areas.
The early thinking regarding a CTP centered on harmonization so that a
level playing field could ultimately be created on which competition would be
equitable. The European Coal and Steel Community had initiated this approach
in the early 1950s and it continued as Union interest moved away from primary
products. The Community had removed some artificial tariff barriers relating to
rail movements of primary products and the Union’s policy initially attempted to
expand this idea in the 1960s to cover the general carriage of goods and especially
those moved on roads. Road transport was viewed rather differently to railways.
It was perceived that the demand and supply features of road haulage markets
could lead to excessive competition and supply uncertainties.
Early efforts included seeking to initiate common operating practices (for
example, relating to driving hours and vehicle weights) and accounting proce-
dures, and standardizing methods of charging. A forked tariff regime for truck-
ing, with rates only allowed between officially determined maxima and minima,
was aimed at meeting the dual problems of possible monopoly exploitation in
some circumstances, and of possible inadequate capacity due to excess competi-
tion in others. Such rates were stipulated on international movements within the

BUTTON 9781786435668 PRINT.indd 495 27/04/2022 11:07

 496 TRANSPORT ECONOMICS, 4TH EDITION

Union. There were practical problems in setting the cost-based rates, and ques-
tions were raised concerning a policy that was aimed at simultaneously tackling
monopoly and excess competition. Limitations on the number of international
truck movements across borders were marginally reduced by the introduction of
a small number of Community quota licenses, authorizing the free movement of
holders over the entire European road network.
The 1973 enlargement to nine member states stimulated a renewed interest
in transport policy and offered the opportunity to review a whole range of policy
areas. The new members – the United Kingdom, Ireland, and Denmark – were
more market-oriented in their transport policy objectives. At about the same
time, the European Commission raised legal questions concerning the inertia of
the Council of Ministers in creating a genuine CTP. It also followed a period of
rapid growth in trade within the Union, bringing infrastructure capacity issues to
the fore and pressures for more flexible regulation of road freight transport.
The outcome was not dramatic although new sectors entered the debates,
most notably maritime transport, and wider objectives concerning environmen-
tal protection and energy policy played a role. Overall, the actions in this period
were a gentle move to liberalization by making the quota system permanent
and expanding the number of licenses increased international intra-Union road
freight capacity. The option of using reference tariffs rather than forked tariffs
was a reflection of the inherent problems with the latter. A major element of the
measures involved improving decision-making regarding the provision of trans-
port infrastructure and about consideration of the way that appropriate charges
should be levied for its use. The importance of transport links outside of the
EU, but part of a natural European network, also began to play a part in policy
­formulation, with the Union beginning to develop mechanisms for financing
investment in such infrastructure.
The enlargement of the Union, as Greece and then Spain and Portugal
joined, had little impact on the CTP. It still essentially remained piecemeal. The
only significant change prior to major developments in the early 1990s was the
gradual widening of the modes covered. There were, for example, moves to bring
maritime and air transport policy in line with Union competition policy.
The accession to membership in the 1970s and early 1980s of countries such
as the United Kingdom and Greece with established shipping traditions brought
maritime issues to the table and then the Single European Act of 1986 provided
a catalyst for initiating a maritime policy. A series of measures were introduced
aimed at bringing shipping within the Union’s competition policy framework.
This came at a time when major changes were beginning to permeate the way in
which maritime services were provided. Technical shifts, such as the widespread
adoption of containerization, had begun to influence the established cartel
arrangements that had characterized scheduled maritime services. (These initial
arrangements were ‘conferences’ that coordinated fares and sailings but later were
more integrated ‘consortia’.)
The size of the European Union’s shipping sector declined significantly in
the 1980s in the face of competition from Far East and communist bloc fleets.

BUTTON 9781786435668 PRINT.indd 496 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 497

The ‘First Package’ of measures in 1985 sought to improve the competitive struc-
ture of European shipping by giving the Commission power to react to preda-
tory behavior by third-party ship owners, and reinterpreted competition policy
to allow block exemptions for shipping conferences, albeit with safeguards. In
1986 a ‘Second Package’ set out to establish a common registry, although this did
not prove successful. Also, as part of the general effort to liberalize the market,
agreement on cabotage (the provision of a domestic service within a country by
a carrier from another nation) was reached but with exceptions in some markets,
for example the Greek Islands.
Ports policy was largely ad hoc. Initial concerns in the early 1990s centered
around modernizing European ports to ensure that they could handle the large
ships that were being introduced. Progress was relatively slow until 2000 when sea
and inland ports were incorporated into the Trans-European Networks (TENs)
initiative with the objective of integrating and prioritizing investment in transport
infrastructure.
The European bi-lateral system of air service agreements covering sched-
uled air transport between member states was, like those in other parts of the
world, tightly regulated. Typical features of a bi-lateral agreement meant: only
one airline from each country could fly on a particular route with the capacity
offered by each bi-lateral partner also often restricted; revenues were pooled; fares
were approved by the regulatory bodies of the bi-lateral partners; and the desig-
nated airlines were substantially state-owned and enjoyed state aid. Domestic air
markets were also highly controlled Air transport in general, however, since the
liberalization of United States domestic markets in the 1970s, was moving away
from a tradition of strict regulation. Until the early 1980s, however, it had been
thought that European aviation policy was outside of the jurisdiction of the
European Commission and a matter for national governments. This changed, fol-
lowing several legal decisions by the European Court of Justice beginning in 1979.
The basic philosophy became that deregulation would take place in stages
(an approach to regulatory reform we shall discuss in more detail in Chapter 14),
with workable competition being the objective and a series of ‘packages’ being
subsequently enacted. The 1987 package aimed at relaxing existing intra-Union
bi-lateral regimes by, for example, allowing deviations from the traditional air
services agreement that set a 50:50 split of traffic between the two member coun-
tries. The decision also required member states to accept multiple designations on
a country-pair basis by another member. It also became easier for countries to
allow more than one of their airlines onto routes. This was extended in the 1989
‘Second Package’ when all capacity limits between bi-lateral partners were to
disappear, and, further, only if both civil aviation authorities refused to sanction
a fare application could an airline be precluded from offering it to its passengers.
Governments could no longer discriminate against airlines of other member
states provided they met safety criteria, effectively removing national airline iden-
tities within the Union.
The creation of the Single European Market in 1992 and the subsequent
moves towards greater political integration brought important changes to the

BUTTON 9781786435668 PRINT.indd 497 27/04/2022 11:07

 498 TRANSPORT ECONOMICS, 4TH EDITION

CTP and related transport policies. Broadly, the 1987 Single European Act
removed institutional barriers to free trade in transport services. At about
the same time, efforts at further political integration and economic develop-
ment led to major new initiatives to provide an integrated European transport
­infrastructure – for example, the TENs. While there were moves to liberalize
industries such as air transport from the late 1980s, the broad basis of European
transport policy was established in ‘The future development of the Common
Transport Policy’ (Commission of the European Communities, 1992), which
advocated as a guiding principle the need to balance an effective transport system
for the Union with a commitment to the protection of the environment.
The changes reflected developments in economic theories that provided new
ways of thinking about transport markets. There was also a switch away from
concern about problems of optimal scale and monopoly power that had been
the intellectual justification for state ownership and regulation of such industries
as railways and air transport, to seeking ways of creating conditions favorable
to X-efficiency and dynamic efficiency. Technically, this was largely, but not
exclusively, a concern with reducing costs replacing that of containing consumer
exploitation. There was mounting concern about the costs of regulated transport
that had macroeconomic implications for the overall economic development of
the Union.
Although terms such as multi-modalism abound in the official literature of
the European Union, and, indeed, some initiatives have transcended the conven-
tional bounds of modal-based actions, a useful and pragmatic way of treating
these recent developments is by mode.

Road Transport

Road transport is the dominant mode of both freight and passenger transport in
Europe; the share of freight going by rail, for example, has fallen from 32 percent
in the Union in 1970 to about 18 percent in 2019. Over the period, the freight
tonnage in Europe has increased 2.5 times and the share of this going by road
has risen from 48 percent to 76 percent. The initial efforts to develop a common
policy regarding road transport, however, proved problematic. Technical matters
on things like driving hours were more easily solved than those of creating a
common economic framework of supply, and economic controls lingered on as
countries with less efficient road haulage industries sheltered them from the more
competitive fleets. There were also more legitimate efficiency concerns throughout
the Union over the wider social costs of road transport, regarding both environ-
mental matters and questions of infrastructure utilization.
The single-market initiative, also later influenced by the potential of new
trade with the post-communist states of Eastern and Central Europe, many of
which have joined the Union, has resulted in significant reforms to economic
regulation in recent years. Earlier measures had helped expand the supply of
international trucking permits in Europe and, as part of the 1992 single-market
initiative, a phased liberalization was initiated that gradually removed restrictions

BUTTON 9781786435668 PRINT.indd 498 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 499

on trucking movements across national boundaries and phased-in cabotage that

had hitherto not been permitted. There was also a considerable reduction in
cross-border documentation.

Railways

Rail transport, while largely filling a niche market in many countries, is an

important freight mode in much of Continental Europe and provides important
passenger services along several major corridors. At the local level, it serves as
a key mode for commuter traffic in larger cities. The European Coal and Steel
Community undertook considerable economic reform of Europe’s freight rail-
ways in the 1950s, including the removal of discriminatory freight rates. From the
late 1960s and 1970s the emphasis had shifted to the rationalization of the sub-
sidized networks through more effective and transparent cost accountancy. The
recent focus has been more on widening access to networks and on technological
developments, especially regarding the development of a high-speed rail network
as part of the TENs initiative. The Union has also instigated measures aimed at
allowing the trains of one member state to use the track of another, with charges
based upon economic costs. The implementation of the strategy has been slow
and has had limited impact.
The Union has found it difficult to devise practical and economically
sound common pricing principles to apply to transport infrastructure despite
the ­proposals of the Oort Report (Oort and Maaskant, 1976). Regarding rail-
ways, the gist of the overall proposals is for short-run marginal costs including
environmental and congestion costs (as well as wear on the infrastructure) to be
recovered. Long-run elements of cost are only to be recovered in narrowly defined
circumstances, and in relation only to passenger services. This clearly has implica-
tions, especially on the freight side, if genuine full cost-based competition is to be
permitted with other transport modes.
Rail transport has also received considerable support as an integral part of
making greater use of multi-modal transport systems. Such systems would largely
rely upon rail (including piggy-back systems and kangaroo trains) or waterborne
modes for trunk haulage, with road transport used as the feeder mode. This is
seen as environmentally desirable and as contributing to containing rising levels
of road traffic congestion in Europe.
The success of some of the French TGV (Train à Grande Vitesse) services,
and especially that between Paris and Lyon, where full cost recovery has been
attained, has led to a significant interest in this mode. In 1990 the Commission
set up a high-level working group to help push forward a common approach to
high-speed railway development and a master plan for 2010 was produced. The
Union’s efforts to harmonize the development of high-speed rail has not been
entirely successful and there are significant technical differences, for example
between the French and German systems.
In legal terms, between 2001 and 2016, four legislative packages were adopted
aimed at gradually opening up rail transport service markets for competition,

BUTTON 9781786435668 PRINT.indd 499 27/04/2022 11:07

 500 TRANSPORT ECONOMICS, 4TH EDITION

making national railway systems inter-operable, and defining appropriate frame-

work conditions for the development of a single European railway area. These
include charging and capacity allocation rules, common provisions on licensing
of railway undertakings and train driver certification, safety requirements, the
creation of the European Agency for Railways, and rail regulatory bodies in each
member state, as well as rail passenger rights.
The 4th Railway Package adopted in 2016 is designed to complete the Single
European Railway Area. Its goal is the revitalization of the sector, making it more
competitive vis-à-vis other modes of transport. It comprises two ‘pillars’:

• The ‘technical pillar’ is designed to boost the competitiveness of the railway

sector by significantly reducing costs and administrative burdens for railway
undertakings wishing to operate across Europe. It will save firms from having
to file multiple applications in the case of operations beyond one single
member state. It aims to ensure that European Rail Traffic Management
System equipment is inter-operable, and to reduce the multiplicity of
remaining national rules.
• The ‘market pillar’ completes the process of gradual market opening started
with the 1st Railway Package. It establishes the general right for railway
undertakings established in one member state to operate all types of passen-
ger services everywhere in the EU, lays down rules to improve impartiality
in the governance of railway infrastructure and prevent discrimination, and
introduces the principle of mandatory tendering for public service contracts
in rail. The market pillar is designed to deliver more choice and better quality
of rail services for European citizens.

Inland Waterway Transport

Inland waterway transport has been important for the European Union since
the beginning. This is mainly because it is a primary concern of the Netherlands
and Germany, with France and Belgium also having interests in the mode.
Progress in formulating a policy has tended to be slow, in part because of his-
torical agreements covering navigation on the Rhine (for example, the Mannheim
Convention), but mainly because the major economic concern has been that of
over-capacity, which in 1998 was still estimated at between 20 and 40 percent at
the prevailing freight rates. Retraction of supply is almost inevitably difficult to
manage, both because few countries are willing to pursue a contraction policy in
isolation and because of the resistance of barge owners and labor.
As in other areas of transport, the Union has sought technical standardi-
zation, and principles for social harmonization were set out in 1975 and 1979.
Economic concerns took over in the 1990s, and in 1990 a system of subsidies
designed to stimulate scrappage of vessels was adopted. The introduction of
new vessels into the inland fleet is only allowed on a replacement basis. The labor
subsidies given in the Netherlands, Belgium, and France (the ‘rota system’) were
phased out by 2000.

BUTTON 9781786435668 PRINT.indd 500 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 501

These measures were coupled with an initiative in 1995 to coordinate

investment in inland waterway infrastructure (the Trans-European Waterway
Network), designed to encourage, for environmental reasons, the greater use of
waterborne transport. Subsequently, in ‘European transport policy for 2010: time
to decide’ (Commission of the European Communities, 2001), the Commission
put emphasis on its Marco Polo Programme on integrating inland waterways
transport with rail and maritime transport for the movement of bulk consign-
ments within an inter-modal chain.

Maritime Transport

Much of the emphasis of the European Union’s maritime policy in the late 1990s
was on the shipping market rather than on protecting the Union’s fleet; that is,
it is user- rather than supplier-driven. Globally, the sector became increasingly
concentrated as, first, consortia grew in importance, mergers took place, and then
the resultant large companies formed strategic alliances. An extension of the 1985
rules to cover consortia and other forms of market sharing was initiated in 1992
and subsequently extended as maritime alliances became more complex.
In 1994 the Commission acted to ban the Transatlantic Agreement reached
the preceding year by the major shipping companies to gain tighter control over
loss-making North Atlantic routes. It did so because the agreement manipulated
capacity and rate, and contained articles covering pre- and on-carriage over land.
It also fined 14 shipping companies that were members of the Far East Freight
Conference for price-fixing because the prices embodied multi-modal carriage
and while shipping per se enjoyed a block exemption on price agreements, multi-
modal services did not.
Ports also attracted attention in the 1990s mainly because advances in tech-
nology had led to significant concentrations in activities as shipping companies
have moved towards hub-and-spoke operations. The main European ports were
working at about 80 percent capacity with many at or near their design capacity.
Whether this is a function of a genuine capacity deficiency or reflects inappropri-
ate port pricing charges that do not contain congestion cost elements is debatable.
In 2001 the Commission launched an initiative to improve the quality of services
offered by ports that involves tightening access standards for pilotage, cargo han-
dling, etc., and to make more transparent the rules of procedure at ports, with the
aim of bringing ports more fully into an integrated transport structure.

Air Transport Policy

The final reform of air transport, the ‘Third Package’, came in 1992 and was
phased in from the following year with the aim of having a regulatory structure
by 1997 akin to that for US domestic aviation. Since 1997, full cabotage has
been permitted, and fares are generally unregulated. Additionally, foreign own-
ership among Union carriers is permitted, and these carriers have, for Union
internal purposes, become European airlines. One result has been an increase in

BUTTON 9781786435668 PRINT.indd 501 27/04/2022 11:07

 502 TRANSPORT ECONOMICS, 4TH EDITION

cross-share holdings and a rapidly expanding number of alliances between air-

lines within the Union. This change did not initially apply to extra-Union agree-
ments where national bi-lateral arrangements still dominate the market.
A ruling by the European Court of Justice in 2002 gave the Commission
the authority to negotiate with the United States on liberalizing the transatlantic
market. Negotiations subsequently took place but while there was some initial
progress through compromise in several technical areas, little substantive agree-
ment emerged for some time regarding the key and fundamental economic issues.
The interests of the domestic coalitions of the military and the air labor unions
that influence the American stance strongly favor an ‘Open Skies’ arrangement
with free airline access to markets, whilst the Union has favored a much more free
market approach that allows for cabotage and flexible movement of capital, via
an ‘Open Aviation Area’.
The impasse was broken in 2007 with a compromise, short-term agreement
that effectively opened the skies over the North Atlantic effective March 30, 2008.
This permitted United States airlines to operate intra-European Union flights if
this involved an all-cargo flight or a passenger flight as the second leg of a flight
started in the States. European airlines are neither permitted to operate intra-
United States flights nor to purchase a controlling stake in an American carrier.
In November the United Kingdom concluded an individual open-sky agreement
with the United States that will supersede the Union agreement post-Brexit.

Overall Impacts

The phased enlargement of the European Union under the Treaties of Nice in
2001 has had implications for transport by affecting the demands placed on the
networks of existing member states and those that have acceded. The accession
states are reforming their economic structures – important for influencing what is
transported and where – and their transport systems. Nevertheless, the difficulties
to be overcome are not trivial:

• Geographical. Enlargement has had implications for the economics of long-

haul transport operations as well as necessitating investment in infrastruc-
ture. What the enlargement has not done is to create a ‘natural’ transport
market. The spatial distribution of economic activities still does not have
the structure of the United States, where the overall physical market is essen-
tially rectangular with centers of population and economic activities at the
corners and in the center, which allow exploitation of efficiency benefits from
long-haul carriage and hub-and-spoke structures. In the European Union
economic activity is dichotomously distributed and enlargement has added
to the central/peripheral nature of the Union.
• Legal. There are issues concerning the role of central legal responsibili-
ties and the degree of local national autonomy will inevitably arise. This is
shaping the wider legal structure in which macroeconomic policies regard-
ing transport are being formulated, and is influencing the external policies

BUTTON 9781786435668 PRINT.indd 502 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 503

of the Union – important for transport in a world of global economies and

global trade.
• Economic. In 2002 the European Union’s Commission President raised the
issue of the need for a common fiscal policy within Europe, or at least that
part of it in the eurozone. Economists have long understood that a common
currency requires a common fiscal policy as a concomitant. This involves the
Union having responsibility for fiscal transfers above those it now controls.
While the exact amount is debated, and depends on the extent to which the
center considers distribution as well as macroeconomic stabilization as part
of its function, it may well be considerable. This would take away much
national autonomy over major transport infrastructure works and influence
short-term public expenditure patterns.

These are not trivial changes, and it is impossible to talk about any one in
isolation from the others, or without considering the background and the current
state of existing Union transport policy. The countries that gained membership
in 2004 (Poland, the Czech Republic, Hungary, Slovakia, Lithuania, Latvia,
Slovenia, Estonia, Cyprus, and Malta) also offer a variety of different challenges
from a transport perspective. Later adding Bulgaria, Romania, and Croatia has
compounded the diversity.
The nature of the economies of the post-Soviet transition states, and
their relationships to the European Union, has already changed considerably.
Nevertheless, there are numerous ways in which their transport systems differ
from much of the older Union. They are mostly distant from the core of the
Union, making railways a potentially more viable mode for long-distance freight
transport. Indeed, the physical area of an enlarged Union offers the prospects of
haul lengths comparable with those in the United States, where deregulated rail-
ways have at least been maintaining their market share. However, the rail freight
networks within transition economies are largely based on dated technologies and
are not oriented to meeting transport demands for movements to and from the
Union. They have traditionally been excessively labor-intensive and serve as job
creators rather than transport suppliers.
Car ownership is considerably lower in the accession countries than in the
older ones, but this is changing. This is putting strains on urban infrastructure
and poses mounting environmental problems. Smaller states, and some regions
within the larger ones, are also themselves subjected to significant transit traffic
flows, raising issues of infrastructure capacity and environmental degradation but
also matters of charging and pricing – a subject the Union has been singularly
poor at addressing. The transition economies also have poorly maintained trans-
port networks largely directed to moving bulk, raw materials to Russia. The road
freight sector has begun to develop in response to the needs of modern just-in-
time production and some countries are making use of the limited infrastructure
links to the West. Transition economies with maritime access and inland water-
ways make considerable use of them. Enlargement comes at a time of technical
change in the sector, with the increasing deployment of a post-Panamax fleet

BUTTON 9781786435668 PRINT.indd 503 27/04/2022 11:07

 504 TRANSPORT ECONOMICS, 4TH EDITION

exploiting scale economies and adding pressure for more hub-and-spoke opera-
tions and fleet rationalization.
A more recent event has been the departure (Brexit) of the United Kingdom
from the European Union in 2020. This re-introduced many of the barriers to free
transportation that existed before the United Kingdom joined the Union. While
a range of temporary agreements have been put in place, the new situation which
came into place in 2021 increases the paperwork required by United Kingdom
operators to serve European markets and in many cases removes their automatic
rights to enter these markets. British-based transport companies, which were
technically European citizens, effectively became simply United Kingdom citizens
with limited rights to operate within Union countries.

13.7 Transport Effects on Regional and Urban Development

We now move from macro issues to look at some of the meso-level effects
transport may have on economic development. The inter-regional spread of
economic activity within a country is of major concern to national governments.
Geographical variations in unemployment, income, migration, and industrial
structure are of importance because they both result in spatial inequalities in
welfare and, in many cases, reduce the overall performance of the national
economy. For these reasons, many countries actively pursue policies that attempt
to stimulate economic activity in depressed areas and to contain damaging explo-
sive growth in prosperous regions.
The policies, which have varied both in intensity and in form over time,
and differ in their nature across countries, have generally concentrated on giving
direct financial assistance to industry and on improving the mobility of labor. In
addition, there have been attempts at improving the economic infrastructure of
what are seen as particularly depressed areas, or in some cases proactive policies
where depression is anticipated, with specific emphasis being placed on providing
better transport facilities. The policy of biasing transport investment in favor of
depressed regions has been subjected to considerable debate over the years.
In the United Kingdom, skepticism about the effectiveness of such a policy
as a regional economic development aid was initially expressed by A.J. Brown in
a Minority Report of the Hunt Committee Inquiry into the Intermediate Areas
as long ago as 1969 and was supported by the findings of the Leitch Committee.
A common thread in these studies is that, in a country such as Britain where
infrastructure is already relatively comprehensive, transport is seldom an impor-
tant factor in explaining disparities in regional economic performance. It is now
accepted by many developed countries that transport policy motivated by regional
policy objectives must be pursued with circumspection and that, in many cases,
improved transport facilities may prove counterproductive for development areas.
A simple hypothetical example illustrates the difficulty. We have two regions, A
and B, producing a single homogeneous commodity. The centers of the regions (see
Figure 13.6) are M miles apart and the commodity can be transported over the area

BUTTON 9781786435668 PRINT.indd 504 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 505

at a constant money cost per mile of f – t per ton. The markets served by the regions
differ, however, because it costs $CA to produce a ton of the commodity in region
A and $CB per ton in region B. Consequently, and assuming no production centers
exist between the regions, a distribution boundary can be drawn (shown by the
dashed line in the figure), which is mA miles from the center of A and mB miles from
the center of B (where mA + mB = M). The boundary is determined by the relative
production costs of the regions and the costs of transport (that is, CA + tmA = CB
+ tmB). Basic manipulation of the algebra gives the form:
 C − CA 
mA =  M + B
 (13.3)
 t 
If, therefore, production is relatively cheap in region A then mA will increase
if infrastructure reduces the cost of transport. Thus, if A is a depressed area
then transport improvements could assist in expanding its potential market and,
therefore, generate more income and employment, but region A must be a low-
cost producer for this to be automatically true. If region B is the depressed one,
then quite clearly investment in improved transport will only worsen the regional
problem by contracting the market area served by the region. Indeed, at the
extreme (where (CA – CB) > Mt), region B may be forced from the market entirely
by the expansion of the low-cost region’s market area.
Of course, the model is a considerable simplification. Regions do not nor-
mally, for instance, specialize exclusively in the production of a single commodity,
but produce a range of goods. Thus, a transport improvement, while damaging
certain industries, may increase the competitiveness of others. The final effect of
the improved transport facility will then depend upon relative production costs
between regions and the importance of transport vis-à-vis production costs in the
overall cost functions for the various commodities.
Further, costs of production may vary with output and thus (following
the ‘infant industry argument’) it may be beneficial to reduce transport costs
if the government’s regional policy also involves using grants and subsidies for
encouraging the establishment of decreasing cost industry in a depressed area.
Supplementary measures of this kind may be necessary if the depressed area
is sparsely populated and, to be successful, its industry needs to penetrate the
markets of other, more populous, regions to benefit from scale economies. It
should be noted, however, that in these circumstances transport improvements
must be accompanied by other regional aids if the natural gravitation of decreas-
ing cost industries to centers of population is to be counteracted. Additionally,
transport costs tend not to increase linearly with distance because of discon-
tinuities and fixed cost elements in the overall cost function (see Chapter 4).
Consequently, the influence of any transport infrastructure improvement is much
more difficult to predict than the simple analysis implies.
A good example of misplaced transport investment is Montreal–Mirabel
International Airport, near Montreal, which was opened in 1975 as the second-
largest airport in the world in terms of surface area that was ever envisaged.
It was intended to replace the existing Dorval Airport (now Montreal–Pierre

BUTTON 9781786435668 PRINT.indd 505 27/04/2022 11:07

 506 TRANSPORT ECONOMICS, 4TH EDITION

A B

mA mB

Figure 13.6 Market areas served by centers A and B

Elliott Trudeau International Airport) as the eastern air gateway to Canada, and,
from 1975 to 1997, all international flights to/from Montreal were required to
use Mirabel. However, its distant location and lack of transport links, as well as
Montreal’s economic decline relative to Toronto, made it unpopular with travel-
ers, so Dorval was not closed as originally planned. Eventually, Mirabel was rel-
egated to the role of a cargo airport.
In summary, there is no general case for thinking that investment in trans-
port infrastructure will automatically improve the economic performance of
depressed regions. In a country such as the United Kingdom, where the transport
cost differences between the least and the most accessible regions has traditionally
only been about 2 percent across industries, the effect of transport investment on
regional policy is, in general, unlikely to be substantial. This is particularly true if
industrial location is influenced by objectives other than cost minimization (for
example, on satisficing principles) or where there is a high degree of X-inefficiency.

Exhibit   Some macro- and microeconomics of elevator travel

The density of population and production found in modern city centers would not be
possible without the elevator (or ‘lift’ in English). People are not prepared, and in many
cases not able, to climb more than about four or five flights of stairs. Without vertical cable
cars, which elevators essentially are, the considerable economies of density that come with
efficient urbanization would not be possible.
Something like the modern elevator appeared in 1853 at New York’s Exhibition of the
Industry of All Nations. Elisha Otis demonstrated the safety catch he had devised to stop
an elevator platform from falling. Gaining public confidence and commercializing meant the
first steam-powered passenger elevator was installed at 488 Broadway in New York City on
March 23, 1857. The first electric elevator was built in Germany by Werner von Siemens
in 1880, while Alexander Miles of Duluth, Minnesota patented in 1887 an elevator with
automatic doors that closed off the elevator when the car was not being entered or exited.
By 1900, completely automated elevators were available.
A real-estate crunch in Manhattan during the 1870s was a catalyst in getting the elevator
accepted. At the time there were discussions concerning moving the financial district to
uptown New York. Henry Hyde, founder of Equitable Life Assurance Society, argued that
by installing a pair of elevators in his headquarters, he could make it the tallest building in

BUTTON 9781786435668 PRINT.indd 506 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 507

the city: seven stories and 130 feet. Between 1910 and 2010, the ratio of the floor space in
a building to the area of land taken up doubled in Manhattan. Today, there are roughly 18
million elevators now installed globally.
While beneficial at the macro level, from an individual’s perspective elevators can be
frustrating. People get annoyed when someone boards an elevator with them only to ride
up one floor. There are stairs, could they not just walk up a single flight? But this is not the
reaction if someone boards the elevator on the first floor with a third-floor destination, but
instead of getting off at the second floor and walking the last flight of stairs, they ride all the
way to the third floor.
An intuitive reaction is that in riding to floor N rather than getting off at N – 1 is laziness.
Such a rider is creating an externality. Getting on the elevator only to ride up one floor
delays everybody else. However, the decision to ride to the second floor rather than the
third is not the same, because whichever floor the person chooses, the elevator is going to
have to stop once. If the person gets on and the floor 2 button is already pushed but 3 is
not, then the trade-off is the same because if the person were to get off at the second floor
and walk, they would spare everyone else the additional stop at the third floor. So, people
get annoyed at a single-floor rider if and only if they get annoyed at this marginal-floor rider.
But there is one more difference. After the person makes the sunk decision to get on the
elevator, but before making the marginal decision, the problem changes. In particular, as the
person is riding, new information is gained. The person can observe how many other people
get on the elevator and are going to be affected by the person’s decision.
In social welfare terms this puts the marginal-floor rider in a different position to the single-
floor rider, because the single-floor rider’s decision whether to board at all is made without
knowing how many others will be on the elevator. The marginal-floor rider’s decision can,
however, be conditioned by the latter. This means there may even be cause to forgive the
single-floor rider and be annoyed at the marginal-floor rider. The former may have reasonably
expected that few people, if any, were going to be inconvenienced. But if the elevator is nearly
full then the total of people’s delays due the single-floor rider’s decision to board is a sunk
cost, but it is an avoidable cost for the marginal-floor rider. If that rider does not get off at the
second floor and walk an extra flight, other riders may have cause to be annoyed.
See also: https://cheaptalk.org/2011/05/31/the-welfare-economics-of-elevator-travel/.

While in Table 13.2 several important studies were cited, it is nevertheless true
that empirical evidence on the specific regional effect of transport policies is
scant, and that which is available is often weakened by the difficulties of iso-
lating transport effects from the effects of other regional policy measures. A
counterfactual exercise conducted by Ronald Botham (1980) suggested that
road investment in Britain between 1957 and 1972 had little effect, although
there was some tendency for it to have a centralizing effect on the distribution of
employment in the country. At a more micro level, Linneker and Spence (1991)
concluded from their study of the impact of the M25 motorway around London
that while it has improved accessibility (and regional market potential) much of

BUTTON 9781786435668 PRINT.indd 507 27/04/2022 11:07

 508 TRANSPORT ECONOMICS, 4TH EDITION

this has subsequently been eroded by generated traffic. This study, however, does
not allow for the substantial range of other aid measures then operative, nor for
intra-regional movements between subregions adjacent to the new facility and
those more distant. A study of the high-speed TGV rail system in France by
Alain Bonnafous (1987) suggests that, while facilitating development, the overall
impact has been small. Studies of airports, for example by Button (2002), have
suggested that they may have significant regional impacts that extend beyond the
immediate area, but this varies on a case-by-case basis.
Turning to a slightly lower level of spatial aggregation, changes in transport
technology have, over time, exerted a strong influence upon the shapes and forms
of the urban areas in which we live. The development of steam locomotion in
the second half of the nineteenth century substantially improved inter-urban
transport and permitted urban growth. Local, distributional services evolved
much more slowly, leading, in most cities, to a concentric pattern of development
around the main rail (or occasionally port) terminal of the type we saw associated
with bid-rent curves in Chapter 3. The wealthy tended, because they could afford
transport which was available, to live in the outer rings of housing while industry,
being dependent upon good inter-urban transport, and the working-class poor
concentrated near the urban core, LGT, the central business district (CBD) – see
the upper part of Figure 13.7.
The introduction of motorized local public transport (initially the tramcar
and later the omnibus) followed by the motor car encouraged the growth of an
axial pattern of urban land use with the former succession of concentric rings of
housing being extended (star-like) in ribbon developments along the main road
arteries (see the lower part of Figure 13.7). Finally, the widespread adoption of
the automobile, combined with improved road systems, limited traffic restraint,
and more efficient road freight transport, has led to the growth of multi-nucleus
cities where there are numerous subcenters and suburbs – Los Angeles is often
cited as the extreme example, as is Phoenix. While this simplified account of
urban development misses many important subtleties, it serves to highlight the
historical role which transport has had in shaping urban growth.
More recently, the difficulties of most large urban areas in the United States
and Europe, unlike rapidly developing countries such as China or India, have not
been ones of containing or molding growth but rather of reversing decay. The
concern once focused on the role of transport in stimulating urban development
has been, since the early 1970s, transformed into concern for inner-city revitaliza-
tion and redevelopment.
As can be seen from Table 13.4 there were substantial outflows of popula-
tion from the centers of virtually all major United States cities. For example,
between 2000 and 2010, the country’s population grew by almost 10 percent, but
35 percent of United States counties, both urban and rural and in all regions,
experienced depopulation. Where cities are concerned, 18 percent of those with a
population of 100,000 or over in 2010 lost population (Franklin, 2021). This has
been accompanied by an even faster exodus of industry. The result of this has
been a decay in inner-city public economies (the tax base of such areas has fallen

BUTTON 9781786435668 PRINT.indd 508 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 509

CBD

Low
income
Middle
income
High
income

Major road
Railroad

Figure 13.7 Concentric (upper) and axial (lower) city geographies

while the composition of the population has become increasingly biased towards
the old, disabled, and poor) and a rise in unemployment (not only have firms left
faster than population, but also it has primarily been manufacturing industry
which has been leaving behind large numbers of unemployed, unskilled workers).
The causes of the decline of inner-city areas are complex; regional and urban
planning policies are partly responsible, but there have also been changes both in
the life-styles aspired to by the population (urban life becoming less attractive as
incomes have risen) and in the production functions confronting industry (the land/
output ratio has been rising). Improved personal transport, especially higher car
ownership levels, has also encouraged more commuting from more distant suburbs.
Official policy to counter the decline of the inner-city areas and to stimu-
late the redevelopment of urban cores has incorporated a substantial transport
component. For example, the UK white paper, ‘Policy for the inner cities’, made
specific reference to the fact that ‘Commerce and industry in inner areas needs
to be served by transport conveniently and efficiently’ and points to the need for

BUTTON 9781786435668 PRINT.indd 509 27/04/2022 11:07

 510 TRANSPORT ECONOMICS, 4TH EDITION

Table 13.4 Geographical population losses in United States cities (2000–10)

Geographical unit Units with loss Population Tracts with

count (%) loss loss (%)

All tracts 30,269 (41.7) 8,166,465 15,825 (21.8)

Cities of over 100,000 in 2010 49 (17.6) 1,066,549 6,203 (30.1)
Cities of 100,000 plus city CBSAs 10 (6.8) 511,158 10,753 (21.9)
All CBSAs 207 (22.0) 924,707 14,210 (21.4)
Metropolitan areas 42 (11.5) 664,521 12,758 (21.4)
Micropolitan areas 165 (28.7) 260,186 1,452 (19.6)

Note: CBSAs are core-based statistical areas and tracts are census track areas with populations of
1,200–8,000.

Source: US Census Bureau.

local authorities ‘to give weight to the implications for local firms when design-
ing traffic management schemes to improve access for central traffic, to ensure
efficient loading and to provide adequate and convenient parking’. Additionally,
it is argued that movement, notably in terms of journey-to-work trips, needs to be
made easier, especially for certain groups of travelers.
At the theoretical level, the bid-rent curve analysis set out in Chapter 3
would seem to imply that cheaper and better public transport would lead to
a spread of cities (that is, the residential bid-rent curves would shift up and to
the right), while traffic restraint policy would lead to greater concentration of
economic activity at the urban core. As Goldstein and Moses (1975) have shown,
however, this type of analysis rests upon the assumption of a single CBD with
no allowance for possible competing suburban centers. This is unrealistic in the
context of most modem conurbations. Figure 13.8 depicts a more typical urban
situation with a major urban center – the CBD – serving as the focal employment
point dominating the suburban center to which it is linked by road.
The urban core itself is served by good local public transport with people
up to B miles away being able to travel in to work by bus. But there are other
modes available for workers and they may opt for one of three main employment/
residential location choices. The situation we consider is one of equilibrium, with
every household achieving the same utility level:

CBD

Suburban center
S1 S S2
B
b
B'

U1 U U2

Figure 13.8 The impact of traffic restraint and public transport subsidies on the urban core

BUTTON 9781786435668 PRINT.indd 510 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 511

• Live within the immediate commuting area (radius B) and travel to work at
the CBD by bus.
• Live outside the immediate commuting area and travel to work at the CBD
by car.
• Commute to the subcenter by car.

In this situation – and excluding residents of the city who elect to take a
fourth option, namely working at home – the boundary U will separate those
workers employed at the core and those with jobs at the suburban center. This
state is one in which no household can improve its utility by changing its place
of work, residential location, or mode of transport. The impact of two common
alternative strategies on the economy of the central core area are:

1 The generalized cost of car travel in the central city area is increased by either
higher parking charges or the imposition of road pricing. This will tend
to produce a rapid decline of the urban core in this model. Higher motor-
ing costs will cause the immediate commuting belt to widen out (say to a
radius of B' in Figure 13.8) with a reduction in real income for those living
(B' – B) miles from the CBD. This will encourage more people either to work
at home or to cease to be active in the labor force (if there are other cities in
the economy, there may also be some out-migration), leading to a decrease
in the labor supply function at the CBD. Also, car travelers confronted by
higher costs will seek work at the suburban center, shifting the employment
boundary to, say, U1. The increased competition for jobs at the subcenter
will depress real income there, leading, once again, to a general reduction in
the overall labor supply in the city. Generally, therefore, traffic restraint in
the CBD makes labor conditions less favorable, which will, in the long term,
make non-core sites more attractive for industry. The empirical fact that
skilled labor tends to be more mobile (both between jobs and locations) is
likely to magnify this effect on industrial location.
2 A subsidized express bus service running in bus-only lanes is introduced
from a depot at location b to run non-stop into the urban core. This is
unlikely to affect those using the existing commuter public transport services
around the CBD, although, if fewer car trips and less congestion result, the
speed of their journeys may rise – but if the service is, in general cost terms,
cheaper than car travel, this will widen the employment market for the urban
core. Former car travelers from as far out as S2 will find that by driving to b
and transferring to the express service they reduce the overall cost of trave-
ling to the CBD. Consequently, the boundary marking employee catchment
areas will shift to U2. People living to the left of U2 will find their real income
has risen as a result of lower transport costs (indeed, many people to the left
of b may drive out to the depot to catch the bus in to the core) and the labor
supply function at the core will have shifted out, making the CBD a more
attractive place for industry. People formerly inactive or working at home
may also now find it attractive to seek employment at the CBD. The labor

BUTTON 9781786435668 PRINT.indd 511 27/04/2022 11:07

 512 TRANSPORT ECONOMICS, 4TH EDITION

supply for the subcenter has fallen, and wages will have to rise to generate the
real-income increase anticipated by employees there. In the long term, the
subcenter will become less attractive for employers.

This theoretical analysis suggests that while one common effect of both the
traffic restraint policy and the public transport improvement policy is to increase
local bus service utilization, the long-run effects on the distribution of popula-
tion and economic activity are likely to be quite different. Goldstein and Moses
(1975) argue, therefore, that even if transport policy cannot cure the malaise of
inner-city areas, improved public transport provision can at least slow down the
decline.

References

Ahmed, Y., O’Sullivan, P., and Sujono, W.D. (1976) Road investment programming
for developing countries: an Indonesian example, Report, Transportation Center,
Northwestern University.
Andersson, A. and Stromquist, U. (1988) The emerging C-society, in D.R. Batten and
R. Thords (eds), Transportation for the Future, Springer.
Aschauer, D.A. (1989) Is public infrastructure expenditure productive? Journal of
Monetary Economics, 23, 177–200.
Balanguer, J. and Cantavella-Jorda, M. (2002) Tourism as a long-run economic growth
factor: the Spanish case, Applied Economics, 34, 877–84.
Barro, R.J. and Sala-i-Martin, X. (1991) Convergence across states and regions, Brookings
Papers on Economic Activity, 1, 107–82.
Baxter, R.D. (1866) Railway extension and its results, Journal of the Statistical Society,
29(4), 549–95.
Biehl, D. (1991) The role of infrastructure in regional development, in R.W. Vickerman
(ed.), Infrastructure and Regional Development, Pion.
Bonnafous, A. (1987) The regional impact of the TGV, Transportation, 14, 127–37.
Botham, R.W. (1980) The regional development effects of road investment, Transportation
Planning and Technology, 6, 97–108.
Button, K.J. (1998) Infrastructure investment, endogenous growth and economic develop-
ment, Annals of Regional Science, 32(1), 145–62.
Button, K.J. (2002) High-technology employment and hub airports: infrastructure’s
contribution to regional development, in Y. Higano, P. Nijkamp, J. Poot, and V.W.
Kobus (eds), The Region in the New Economy: An International Perspective on Spatial
Dynamics in the 21st Century, Ashgate.
Button, K.J., Leitham, S., McQuaid, R., and Nelson, J.D. (1995) Transport and industrial
and commercial location, Annals of Regional Science, 29, 189–206.
Commission of the European Communities (1992) The future development of the
common transport policy: a global approach to the construction of a community
framework for sustainable mobility, Brussels, COM(92) 494 Final.
Commission of the European Communities (2001) European transport policy for 2010:
time to decide, CEC.
Devanney, J.W., Livanos, V.M., and Stewart, R.J. (1975) Conference rate-making and the
west coast of South America, Journal of Transport Economics and Policy, 9(2), 154–77.
Fisher, R.C. (1997) The effects of state and local government public services on economic
development, New England Economic Review, March/April, 53–82.

BUTTON 9781786435668 PRINT.indd 512 27/04/2022 11:07

 TRANSPORT AND ECONOMIC DEVELOPMENT ­ 513

Fogel, R. (1964) Railroads and American Economic Growth: Essays in Econometric History,
John Hopkins Press.
Franklin, R.S. (2021) The demographic burden of population loss in US cities, 2000–2010,
Journal of Geographical Systems, 23, 209–30.
Gallen, T.S. and Winston, C. (2021) Transportation capital and its effects on the U.S.
economy: a general equilibrium approach, Journal of Macroeconomics, 69, No 103334.
Goldstein, G.S. and Moses, L.N. (1975) Transport controls and the spatial structure of
urban areas, Papers and Proceedings of the American Economic Association, 85, 289–94.
Gramlich, E.M. (1994) Infrastructure investment: a review essay, Journal of Economic
Literature, 32(3), 1176–96.
Gunduz, L. and Hatemi-J., A. (2005) Is the tourism-led growth hypothesis valid for
Turkey, Applied Economics Letters, 12, 499–504.
Hirschman, A.O. (1958) The Strategy of Economic Development, Yale University Press.
Hunter, H. (1965) Transport in Soviet and Chinese development, Economic Development
and Cultural Change, 14, 71–2.
Kindleberger, C.P. (1958) Economic Development, McGraw-Hill.
Leeper, E.M., Walker, T.B., and Yang, S.-C.S. (2010) Government investment and fiscal
stimulus, Journal of Monetary Economics, 57, 1000–12.
Linneker, B.J. and Spence, N.A. (1991) An accessibility analysis of the impact of the M25
London orbital motorway on Britain, Regional Studies, 26, 31–47.
Little, I.M.D. and Mirrlees, J.A. (1974) Project Appraisal and Planning for Developing
Countries, Heinemann.
Millard, R.S. (1959) Road development in the overseas territories, Journal of the Royal
Society of Arts, 107, 270–91.
Morrison, J.C. (1993) Macroeconomic relationships between public spending on infra-
structure and private sector productivity in the United States, in J.M. Mintz and R.S.
Preston (eds), Infrastructure and Competitiveness, John Deutsch Institute for the Study
of Economic Policy.
Narayan, P.K. (2004) Economic impact of tourism on Fiji’s economy: empirical evidence
from the computable equilibrium model, Tourism Economics, 10, 419–33.
Oort, C.J. and Maaskant, R.H. (1976) Study of the possible solutions for allocating the
deficit which may occur in a system of charging for the use of infrastructure aiming at
budgetary equilibrium, European Economic Community.
Owen, W. (1964) Strategy for Mobility, Brookings Institution.
Premti, A. (2016) Liner shipping: is there a way for more competition? Report, United
Nations Conference on Trade and Development.
Prud’homme, R. (1996) Assessing the role of infrastructure in France by means of region-
ally estimated production functions, in D.F. Batten and C. Karlsson (eds), Infrastructure
and the Complexity of Economic Development, Springer-Verlag.
Rostow, W.W. (1960) The Stages of Economic Growth, Cambridge University Press.
Solow, R.M. (1957) Technical change and the aggregate production function, Review of
Economics and Statistics, 39, 312–20.
Sturm, J.-E. and De Haan, J. (1995) Is public expenditure really productive? New evidence
from the USA and the Netherlands, Economic Modelling, 12(1), 60–72.
Vickerman, R.W. (1987) The Channel Tunnel: consequences for regional growth and devel-
opment, Regional Studies, 21, 187–97.
Wilson, G.W. (1966) Towards a theory of transport and development, in G.W. Wilson,
B.R. Bergmann, L.V. Hirsch, and M.S. Klein (eds), The Impact of Highway Investment
on Development, Brookings Institution.
Zerby, J.A. (1979) On the practicality of the UNCTAD 40–40–20 code for liner confer-
ences, Maritime Policy and Management, 6, 241–51.

BUTTON 9781786435668 PRINT.indd 513 27/04/2022 11:07

 14 Political Economy and Transport
Regulation

14.1 Underlying Issues

A considerable amount of economists’ efforts on the public sector side are

attached to the regulation of transport, and probably an equal amount on the
private side to responding to the resultant regulations. A chapter providing
an overview of the economic regulation of transport activities must inevitably
involve some overlap with other chapters in this volume. This chapter is less than
complete in that sense. Economic regulation is also a very large and complex topic
and there is no pretense that the coverage here is complete or thorough. It is also
one that transcends the transport sector in numerous respects as many nations,
for example, modified the ways they looked at economic regulation more gener-
ally from the 1970s (Button and Swann, 1992).
Space has already been devoted to policies for containing pollution and for
the control of traffic congestion, for instance. We have also implicitly considered
the implications of some forms of intervention regarding pricing and market
entry. Transport, however, has been a sector that has been subjected to various
forms of economic regulation throughout history, and this justifies a more spe-
cific examination.
Wider questions concerning such matters as why have regulations been
imposed, how have they operated, and how they can be appraised have not been
examined in any depth so far. Equally, there are different ways in which transport
is supplied with differing degrees of public sector participation in ownership – the
ultimate form of regulation – that impact on efficiency and equity implications.
Just why this occurs and how it differs between countries raises a further set of
economic questions.
Besides the general issues of the various types of regulation and ownership
of transport, there have also been significant changes in the way in which many
countries approach these regulatory policy issues. We saw, for instance, a consid-
erable relaxation of long-standing entry and price controls in markets as diverse
as the United States’ domestic aviation and the United Kingdom’s bus transport
industry in the 1970s and 1980s. This was a trend that has extended into the new
market economies of the post-communist states and to many developing nations.
We have also seen extensive privatization of transport infrastructure. This
has included airports and air traffic control in the United Kingdom and many
other European countries. More build, operate, and transfer public–private part-
nerships have been formed with the aim of speeding up transport investments.

BUTTON 9781786435668 PRINT.indd 514 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 515

Many of these initiatives are as focused on the tapping of a wider pool of

technical expertise as on exploiting more diverse sources of funding. Private
sector involvement at a second tier has also expanded with the franchising-out
of operations (for example, of London’s bus services and of South American
airports) and the greater use of private sector funds in the financing of infra-
structure investments (for example, the Channel Tunnel and the French TGV
network). The subject, therefore, is an important one and is likely to remain so
for some time to come.
What we do not consider here are the generic laws and controls that are
imposed in most counties, and thus inevitably affect transport. For example, many
countries have generic minimum wage laws, anti-trust and monopoly regulations,
social justice constraints, and consumer protection legislation. These apply to all
sectors, including transport, although in some cases there are specific exemptions.
The implications of such regulations on transport are often far from trivial, but
discussion of them is moving us away from the main theme of this book.
While it is often difficult to precisely define the rationale and nature of regu-
lations, a very simple flow chart is seen in Figure 14.1 that gives a broad indica-
tion of the types of microeconomic measures involved. In addition, transport is
inevitably affected by macroeconomic policies, such as monetary and fiscal policy,
and internationally by trade policies. These can influence the aggregate size of the
economy and thus the total demand and supply for transport services. But these
macro aspects represent wider, cross-sector matters that are outside of the scope
of this book.
Also important for the efficiency of transport are the microeconomic regula-
tions that affect the economic and social environments in which transport services
are demanded and supplied. Since we have spent some time on the challenges of
dealing with areas of social regulation, and especially the environment, the focus
here is mainly on the nature of economic regulation in its more traditional sense
of seeking to increase the efficiency of transport suppliers. What we do not do,
however – although it can be very important – is to offer any overview of generic
anti-trust or merger regulations that affect all types of industry or national laws

Regulation of the economic system

Macroeconomic regulation Microeconomic regulation

Generic anti-trust Social regulation Economic regulation

regulation

Environmental Health and Consumer Affirmative Private Public

protection safety protection rights ownership ownership

Markets Internal by Command and Fiscal

self-regulation control instruments

Figure 14.1 The basic breakdown of types of regulations affecting transport

BUTTON 9781786435668 PRINT.indd 515 27/04/2022 11:07

 516 TRANSPORT ECONOMICS, 4TH EDITION

governing things such as minimum wages. We essentially push these to one side
and attend to areas of economic regulation that are specifically focused on trans-
port matters.
The objective of this chapter is not to describe all the changes in regulation
and ownership of transport that have taken place – that would be impossible,
although some of the more important developments are reviewed – but rather
to explore the economic arguments underlying the various positions taken by
policy-makers. Some contextual historical information is helpful in doing this.
Initially, however, a little time is spent looking at the various economic theo-
ries of regulation and setting trends in transport policy in the context of these
theories.
In doing this, we not only look at domestic policies and their affects, but
also at international policy, a matter of increasing importance as international
trade grows and as globalization takes place. Changes in international transport
regulations, which may range from the removal of transport barriers within the
European Union to the spread of Open Skies agreements in air transport, have
implications for international markets.
In Figure 14.2, for example, if international air services are deregulated then
they have direct effects on traffic into and out of the country of interest. But this
action also has secondary effects on airlines and other modes within the country
that take traffic to and from the international airport. This, in turn, because of
economies of scope and density, will lower the costs of some purely domestic
services and hence the use made of them. The additional markets that are now
opened up will lead to further changes in both domestic and international traffic as
industries such as tourism and high-technology companies that use air transport
change the scale and nature of the products they supply. In effect, it will have a
macro-income effect of increasing national income, inducing a further positive
ripple effect on the demand for air transport. Most transport regulation, however,

International out
Domestic feed and
Trade-generated
domestic traffic

distribution

International in

Income-generated
domestic traffic

Figure 14.2 Links between international and domestic transport markets

BUTTON 9781786435668 PRINT.indd 516 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 517

focuses on specific elements of the system, for example regional aviation, domestic
trunk haul services, or international aviation. Indeed, they are often regulated by
different agencies.

14.2 Regulation Theory

Economists have long recognized that markets may, in practice, suffer from
serious imperfections, indeed some discussion of this has been set out in previ-
ous chapters. Most obviously, these imperfections, or market failures, could
adversely affect the users of transport services, perhaps because fares would be
sub­optimally high or the service offered dangerous. But equally, they may harm
third parties through, for example, generating excessive environmental pollution,
or the predatory pricing behavior of incumbent operators may reduce the poten-
tial viability of other firms wishing to supply transport services and thus deter
them from entering the market.

Why Regulate?

A wide range of arguments, some of doubtful economic logic, have been drawn
into debates over transport regulation. Broadly, the types of market failure that
have attracted most attention embrace:

• The containment of monopoly power. Transport often involves large, indivis-

ible, and specialized infrastructure that lends itself to monopoly provision to
maximize scale economies. This has led to long-standing concerns about the
potential economic exploitation of monopoly power. This was an issue in the
case of the railways, which dominated inland transport for nearly a century
from the late 1830s, but while some monopoly power persists in certain areas
of transport activity, technical advances across a range of transport modes
have reduced the potential for pure monopoly exploitation, at least in most
developed countries. More common perhaps is the fear that suppliers of
transport services may combine in cartels to limit output and prevent new
entrants coming into the market.
• The control of excessive competition. Unregulated competition may limit
the quality of service offered to customers and result in instability in the
industry – in technical terms there may be no sustainable equilibrium: an
empty core may exist. The actual problem is not competition per se but
rather the possibility that externalities may result or that some sections of
the community may not be provided with adequate services. Additionally, in
some instances, notably trucking and inter-urban passenger transport, the
potential for conditions of monopolistic competition developing also pose
problems of possible excess capacity being supplied.
• The regulation of externalities. Imperfections in the market mechanism may
result in transport activities imposing costs that are not directly included in

BUTTON 9781786435668 PRINT.indd 517 27/04/2022 11:07

 518 TRANSPORT ECONOMICS, 4TH EDITION

the private sector’s decision-making – pollution and congestion being the

main causes for concern. This subject has been explored more fully in earlier
chapters and we shall not dwell on it here. It is, however, becoming an aspect
of regulation that is attracting increasing attention at the global as well as at
the national and local levels.
• The provision of public goods. Because certain items of infrastructure, such
as roads, are thought to exhibit public goods characteristics (that is, non-
excludability, whereby use cannot be controlled, and non-rivalness, whereby
congestion cannot arise), their provision, it is argued, would be at best
inadequate without government intervention. In effect, suppliers could not
charge for their use. The fact that one cannot exclude use makes it difficult
to price the facilities. The degree to which such infrastructure should be con-
sidered as conforming to a public good, however, often depends upon the
initial policy in place – for example, it is relatively easy to exclude cars from
a road if wished.
• The provision of high-cost infrastructure. The sheer cost, long pay-back
period, and complexity, combined with possible high levels of risk, make it
unlikely that all major pieces of infrastructure would be built or expensive
transport engineering research undertaken without some form of govern-
ment involvement. Adam Smith, as we saw earlier, even conceded that on
occasions the government may need to directly provide such facilities, but
there are also ways of regulating the private sector to ensure it provides
infrastructure.
• The assistance of groups in ‘need’ of adequate transport. As was seen in
Chapter 4, the idea of ‘need’ embraces the notion that, for a variety of
reasons, including faults in the existing pattern of income distribution, effec-
tive demand is not an adequate guide to transport resource allocation and
wider, social criteria should, therefore, be sought. The market thus needs
regulations to meet this objective.
• The existence of high transaction costs. While free markets may theoretically
be capable of optimizing output, this may involve high transaction costs: the
costs of ‘doing business’. Drivers confronting each other on a road could
bargain as to who has the right of way, but a simple rule, say giving priority
to the left, is likely to prove more efficient. The problem may also be linked
to the difficulties of negotiating for rights of way to construct transport
infrastructure.
• The integration of transport into wider economic policies. Land use and
transport are clearly inter-connected and some degree of coordination may
be felt desirable if imperfections exist in either the transport or the land-use
markets. Additionally, intervention in the transport sector may form part of
a wider government macroeconomic strategy (for example, price controls
or investment programs) or industrial policy. The traditional notion of
Keynesian fiscal macroeconomic policies of the type adopted in the United
States in 2008–2009 to counteract economic recession, for example, often
involve significant public investments in transport facilities.

BUTTON 9781786435668 PRINT.indd 518 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 519

• The need to reflect the genuine resource costs of transport. In the case of
certain finite, non-renewable resources (for example, mineral fuels), the
market mechanism may fail to reflect the full social time preference of
society. The government may, therefore, intervene to ensure that the deci-
sion-maker is aware of the true shadow price.
• The improvement of transport coordination. Because there are numerous
suppliers of transport services, inefficient provision may result if their deci-
sions are made independently. There is also the prospect of duplication of
transport facilities, and consequential wastage of resources, without some
degree of central guidance. This is a topic that we return to in more detail
later.
• To instigate market stability. Although transport is a derived demand, it
can influence the final markets that make use of it. In particular, as seen
in Chapter 3, it interacts with land-use patterns. Continual and difficult-
to-predict changes in transport services available can negatively affect the
confidence of those making locational decisions. Market interventions, such
as guaranteeing rail or bus services for a given period of time inject a greater
degree of certainty into locational selections (Spiller, 2013). Railroads
in the United States, for example, must gain approval from the Surface
Transportation Board before railroad lines can be abandoned.

In practice, most official policies claim to cover a range of different problems,

although conflicts may and do emerge. For example, policies designed primarily
to contain externalities may have adverse effects on income distribution or could
run counter to a national economic policy that is pursuing a course of maximiz-
ing GNP.
Similarly, measures to ensure that adequate high-cost research is conducted
may mean conferring monopoly powers on private suppliers (for example,
through the patent system or in terms of government contracts to purchase the
fruits of a new technology). Consequently, there is an inevitable blurring across
these justifications for government involvement when various policy measures
are discussed or introduced. Even more uncertain than the exact justification or
objectives underlying some policies is the exact effects the different policy tools
employed by policy-makers are likely to exert.

Instruments of Policy

The instruments of transport policy are sometimes broken down into those
which, adopting American jargon, are aimed at economic regulation and those
which are aimed at social regulation – quantitative and qualitative regulation, in
the British vocabulary. The former is concerned with controlling the amount sup-
plied in transport markets, who supplies it, and the price which consumers pay.
Social regulation specifies the nature of the transport services provided, relating,
for instance, to vehicle design, maximum emissions levels, driving hours, training
of personnel, etc. In practice there is an inevitable overlap between the two sets

BUTTON 9781786435668 PRINT.indd 519 27/04/2022 11:07

 520 TRANSPORT ECONOMICS, 4TH EDITION

of instruments. Limiting market entry, for example, can contain many adverse
environmental effects of transport, while strict quality controls can act to contain
competition.
It is perhaps more useful, therefore, simply to provide a listing of the various
policy instruments under the following general headings:

• Taxes and subsidies. The government may use its fiscal powers either to
increase or decrease the costs of various forms of transport or services over
different routes – or, indeed, the cost of transport in general. While they
may be designed in some cases to consciously affect the costs of transport
(for example, subsidies to promote transit use or fuel taxes to finance road
construction), in other instances they are pure sumptuary taxes aimed at
raising revenue from sources where demand is less price-sensitive. This often
involves taxes on transport services used by higher-income groups.
• Price controls. Concerns that a monopolist transport service supplier may
exploit its ability to price above marginal costs has led to widespread price
regulations over the years. These have taken a variety of forms ranging
from rate-of-return regulation (which sets prices allowing a profit level to be
attained) to price-capping (which allows average prices to rise regularly but
at a rate less than general cost levels), each with its particular advantages and
challenges.
• Direct provisions. Local and central government are direct suppliers, via
municipal and nationalized undertakings, of a wide range of transport ser-
vices. They are also responsible for supplying a substantial amount of trans-
port infrastructure, notably roads, and supplementary services, such as the
police. In the context of bodies such as the European Union, the supply may
be by pan-national agencies. The link between supply, the infrastructure, and
its use may also involve direct public provision of both, as was often the case
with airport and state-owned airlines, but this is decreasingly so.
• Laws and regulations. Government (and to a lesser extent, local authorities)
may legally regulate the transport sector and there has grown up an extensive
body of law that, in effect, controls and directs the activities of both trans-
port suppliers and users. Linked with this, there are generic laws that govern
financial, labor, and other markets which provide essential inputs into trans-
port industries and thus affect the way in which it functions.
• Competition policy and consumer protection legislation. It is useful to dis-
tinguish general industrial legislation, governing such things as restrictive
practices and mergers, and consumer protection legislation, covering such
things as advertising, which embraces all forms of activity in the economy
and not just transport. Obviously, as we mentioned earlier, they also apply to
transport. In addition, many countries have specific transport-oriented poli-
cies aimed at protecting users.
• Licensing. The government may regulate either the quality or quantity
of transport provision by its ability to grant various forms of licenses to
operators, vehicles, or services. The system of driving licenses for cars and

BUTTON 9781786435668 PRINT.indd 520 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 521

trucks also influences the demand for private transport. A specific form
of licensing may be seen in franchising whereby a private sector undertak-
ing may operate a publicly owned transport facility for a designated period.
At the much broader level, there have been long-standing policies in inter-
national transport that have, de facto, imposed a licensing structure to limit
the supply of transport services, or to restrict the carriers involved in the
trade.
• The purchase of transport services. Various non-transport activities of gov-
ernment require the use of transport services – for example, to move military
personnel and to supply medical services. Hence, by means of its position as
a large consumer, government may exert a degree of countervailing power
over transport suppliers. This is often more relevant at the local level of
administration, where large fleets of vehicles can be loaned or leased. More
indirectly, by purchasing one type of transport service, a government may
influence the larger market for these services. For example, the purchase of
military transport equipment may allow economies of scale in production
that reduces the cost of producing related civilian products. Aircraft engines
and frames are often seen in this context.
• Moral suasion. In many instances this is of a weak form, usually being educa-
tional or the offering of advice on matters such as safety (for example, adver-
tising the advantages of the wearing of seat belts), but it may be stronger
when the alternative to accepting advice is the exercise, by government, of
others of its powers (for example, the refusal of a license or the withdrawal
of a subsidy).
• Research and development. Government may influence the long-term devel-
opment of transport through its own research activities. These are, in part,
conducted by its own explicit agents (for example, the Transportation
Research Board in the United States) and, in part, through the funding
of outside research via contracts. There are also fundamental research ini-
tiatives overseen by such bodies as the National Science Foundation in the
United States that can impact significantly on transport.
• Provision of information. The government, through various agencies, offers
certain technical advice to transport users and provides general information
to improve the decision-making within transport. Many of these services
are specific to transport (for example, weather services for shipping and air
transport), while others assist the transport sector less directly (for example,
information on trading arrangements overseas). In terms of safety and secu-
rity, most governments offer warnings of any severe dangers that may exist
in other countries.
• Policies relating to inputs. Transport is a major user of energy, especially oil,
and utilizes a wide range of other raw materials and intermediate products.
Government policy relating to the energy and other sectors can therefore
have an important indirect bearing on transport. It can, for instance, stipu-
late the types of fuel that may be sold – lead-free gasoline being required in
many countries.

BUTTON 9781786435668 PRINT.indd 521 27/04/2022 11:07

 522 TRANSPORT ECONOMICS, 4TH EDITION

It seems logical, therefore, that if such market imperfections can be identi-

fied, it is in the ‘public interest’ to intervene and reduce their distortive effects.
This is not a very contentious view. Difficulties arise, however, not from the
notion of public interest but rather from the degree to which intervention can,
in practice, produce public benefit. In the 1970s, as we see below, a growing body
of opinion emerged that regulation had become excessive and no longer served
the public interest. In particular, the ideas of the Chicago School of economists
questioned the motivation underlying the actions of regulators – most notably
that they tend to act as rational economic entities and pursue policies aimed at
furthering their position rather than necessarily that of the public interest (Stigler,
1971). Others considered the power of the regulated industries and argued that,
because of their control of information flows (for example, regarding cost data
which is needed to regulate fares) and their power to lobby, there was a tendency
for them to capture the regulatory process.
Even if regulation is initiated in the public interest, it is not free from
problems. Ideally, policy-makers like to match one policy instrument with one
objective, but this is seldom possible in transport. The problems of interdepend-
ence of objectives have already been alluded to, but, in addition, the instruments
themselves frequently have diverse effects. Taxation policies to reduce the use of a
specific mode of transport may prove regressive, while licensing to contain exter-
nalities may result in quasi-monopoly powers being given to certain suppliers.
Forecasting, monitoring, and appraising the effects of alternative policy instru-
ments also usually proves difficult. Transport policy is pursued through a package
of instruments and policy changes, usually resulting in several of these instruments
being varied in their intensity at once. The effects of such changes are also only
likely to be fully felt after a lag as agents in the transport market gradually respond
and adjust to the new situation. In the short term, a trucker, for instance, may do
little in response to higher fuel taxation (save pay the additional money), but in the
longer term he or she is likely to modify the method of ­operation employed and,
in the very long term, may even change the type of vehicle fleet used.
Even if one could isolate occasions when a change is made to a single policy
instrument, it is unlikely that the full effect could be recorded before further
changes take place. Finally, there is the problem of determining the counter-
factual – the course of events which would have ensued if policy had not been
changed. Government is frequently reactive in its approach – suggesting that
changing circumstances are already observable by the time policy is enacted – but
on other occasions policy changes represent initiatives and anticipate change.
One can never simply assume that events would have continued the course set
prior to a major policy change.

14.3 Monopoly Power

Although economic regulations have been motivated by a variety of factors,

concern over the exercise of monopoly power has dominated much of the debate

BUTTON 9781786435668 PRINT.indd 522 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 523

and legislation. The large-scale infrastructure that is needed to support most

forms of transport, and the various forms of network economies that can exist in
providing transport services, suggest that concentrated supply can minimize cost.
Such a concentration, however, allows for the opportunity for suppliers to exploit
users by pushing up prices or it can lead to them not having an adequate incentive
to manage the system efficiently.

Rate-of-return Regulation

These problems have long been recognized and regulations have been common-
place. State provision has often been used to ensure that users are not exploited by
keeping user prices in line with costs. This has widely been used in the provision
of physical infrastructure, but also, as we see below, in the provision of services in
some countries. The issue of ‘government failure’, and the capture of systems by
politicians and management, has led to other options being pursued in some cases.
Countries such as the United States have sought to control the prices at which
the private sector may offer transport services. The traditional methods used, for
example by the Civil Aeronautics Board (CAB) and the Interstate Commerce
Commission until the 1980s, relied upon rate-of-return regulation. This sets trans-
port prices so that suppliers can earn a ‘fair rate of return’ on their investments.
This allows the transport undertaking to choose its use of inputs, levels of output,
and prices when there is no excess above the officially defined fair rate of return. In
other words, with only one non-capital input labor, L, the rate-of-return is:

(PQ – wL)/K (14.1)

where:

P is price;
Q is output;
w is wages; and
K is capital, and this must not exceed the fair rate-of-return:

f ≥ (PQ – wL)/K (14.2)

The main difficulties with this approach involve the incentive structure it
offers to over-capitalize – the ‘Averch–Johnson effect’ (Averch and Johnson,
1962) – and, as discussed in the previous section, the ease with which the system
may be captured both by the industry under regulation and the regulators seeking
to serve their own ends.
The Averch–Johnson effect is the tendency of companies to engage in exces-
sive amounts of capital accumulation to expand the scale of their profits. If
companies’ profit/capital ratio is regulated as in equation (14.2), then at a certain
percentage there is a strong incentive for them to over-invest to increase profits
overall.

BUTTON 9781786435668 PRINT.indd 523 27/04/2022 11:07

 524 TRANSPORT ECONOMICS, 4TH EDITION

Specifically, the main Averch–Johnson result is that the capital/labor ratio

selected by a profit-maximizing regulated firm will be greater than that consist-
ent with a cost-minimizing one for any output it chooses to produce. If what is
considered by a regulator to be the fair rate of return is greater than the cost
of capital, a firm will have an incentive to invest as much as it can, consist-
ent with its production possibilities; the difference between the allowed rate
and its actual cost of capital is pure profit. This goes against any optimal effi-
ciency point for capital that the company may have calculated, as higher profit
is almost always desired over and above efficiency. This is sometimes called
‘gold-plating’.
As an illustration, we assume a railroad uses input labor (L) and capital (K).
The production function for the undertaking is therefore R(K,L), with w the cost
of labor and r the cost of capital.
The railroad’s objective is to maximize profits (Π):

Π = R(K,L) – wL – rK (14.3)

Microeconomic theory tells us that this requires that the ratio of the mar-
ginal products of the inputs be equal to the ratio of costs:

(δR/δK)/(δR/δL) = r/w (14.4)

The rate-of-return regulator fixes the cost of capital at s – the rate of profit
that the utility earns on its rate base. If the regulator sets s = r (the market
rate of capital), then there is no problem. But regulators tend to set s larger than
r, (s > r), to ensure that railroad is able to acquire investment capital, thus:

(δR/δK)/(δR/δL) = (s/w) > (r/w) (14.5)

The railroad will thus tend to use capital and other inputs in inefficient quan-
tities. This can also be seen diagrammatically in Figure 14.3. The isoquant shows
a given level of railroad services using various combinations of inputs K and L.
With no rate-of-return regulation, the optimal capital–labor substitution rate is
– r/w at point E, and using K amount of capital. Introducing rate-return regula-
tion, with ‘α’ being the amount this is below that prevailing in the market, causes
the rate of factor substitution to shift to –(r – α)/w. The outcome is an increase of
capital use to points A – J, and the amount of K1 – K.
Concern with the possible adverse effects of rate-of-return regulations’
economic efficiency has been supported by empirical analysis, and with the
way it affects American transport provision and prices in the 1960s and 1970s.
Table 14.1 provides a brief summary of a series of studies conducted on the
rate-of-return regulation of the CAB. In particular, the fact that only inter-state
services were regulated, whereas many state systems were not, allowed compari-
sons to be made.

BUTTON 9781786435668 PRINT.indd 524 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 525

Isoquant

E
Slope = – (r – α)/w
A–J

Slope = – r/w

0 K K1 K

Figure 14.3 Averch–Johnson effect

Table 14.1 Results of United States studies on the implications of airline regulation prior to 1978

Study Data Comparisons Conclusions

Caves (1962) CAB routes Structure, conduct, Problems with the industry’s
& performance performance that required changes
in the regulatory structure but did
not oppose regulation
Levine (1965) Intra-state & Fares Regulation caused higher fares &
CAB routes resulted in lower load factors
Jordan (1970) California & Fares Regulation caused excess capacity,
CAB routes benefitted aircraft manufacturers,
labor unions, & airlines
Keeler (1972) California & Fares Regulation caused excess capacity
CAB routes that dissipated any profits from
fares set at cartel levels by the CAB
Douglas & Miller CAB routes Fares & flight Regulation resulted in high fares &
(1974) frequency suboptimal high qualities of service
being offered
DeVany (1975) California & Fares & flight Regulation protected the consumer
CAB routes frequency with fares set close to the output-
maximizing level
Keeler (1978) California & Fares CAB regulation led to excess
CAB routes charges amounting to $2.7 billion
per annum

Note: CAB = Civil Aeronautics Board.

Source: Button (1989). This paper contains the full references to the studies cited.

BUTTON 9781786435668 PRINT.indd 525 27/04/2022 11:07

 526 TRANSPORT ECONOMICS, 4TH EDITION

Price-capping

An alternative to rate-of-return regulation developed initially for telecommunica-

tions regulation, but subsequently adopted in numerous transport contexts, is
‘price-capping’. The aim of this policy of lighted-handed regulation is to squeeze
out inefficiencies, and especially X-inefficiency, by forcing transport industries
to provide their services at increasingly low real prices. It has, for example, been
applied to privatized airports in the United Kingdom’s National Air Traffic
Services (NATS) air traffic control system and its airports.

Exhibit   Regulated and unregulated airlines

Prior to deregulation of the United States’ domestic inter-state passenger airline market in
1978, fares and airline market entry were federally regulated for cross-state border traffic,
but within states the state authority had jurisdiction. While federal regulation was very
restrictive, large states such as California were more liberal. Bill Jordan made use of this
different structure in 1965 to look at inter-state airline fares in the Northwest Corridor and
concluded that they could be 32 to 47 percent lower than they could be if the trunk airlines
were relatively unregulated on the Californian model.

Year Trunk airline load factor (%) California intra-state load factor (%)

1951 69.6 69.0

1952 67.1 65.9
1953 64.7 67.1
1954 63.4 69.2
1955 64.1 72.2
1956 64.1 75.7
1957 61.5 80.6
1958 60.0 72.4
1959 61.4 71.1
1960 59.5 71.1
1961 56.2 72.1
1962 53.3 75.3
1963 53.8 72.8
1964 55.4 74.9
1965 55.2 63.3
1966 58.5 –
1967 57.2 –
1968 53.0 –
1969 50.3 –
1970 49.3 –

When examining the implications of lower inter-state load factors on the formula-based
regulated fares, Keeler constructed a hypothetical optimal cost function for these routes
based upon load factors akin to those found in California. This involved developing a

BUTTON 9781786435668 PRINT.indd 526 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 527

long-run cost function embracing such factors as trip length and congestion to provide a
counterfactual for comparison with the fares charged. The analysis made use of a long-run
cost model, taking account of such things as available ton-miles. A number of adjustments
had to be made to allow reasonable comparisons. For example, as can be seen from the
table, there were considerable differences in load factors. Keeler assumes that if inter-state
routes operated as efficiently as intra-state routes, the former would have a 60 percent load
factor.
Keeler found that in 1968 regulated inter-state routes had fares involving 20 to 95 percent
over unregulated fares, with the mark-ups tending to rise with distance. Using a slightly
different method for 1972, mark-ups were found to range between 48 percent for short and
medium routes and 84 percent for long-hauls. These findings support mark-ups of between
47 and 89 percent found by Jordan.
Much of the differential found by Keeler results from the presence of the pioneering low-
cost carrier, Pacific Southwest Airlines, in the Californian market. The airline was not only
offering lower fares compared to those found on the major inter-state routes, but was also
more profitable than a typical trunk carrier. The difference in costs was largely attributed to
excess capacity on trunk routes that manifested itself in the very much lower load factors
seen in the table.
See also: W.A. Jordan (1970) Airline Regulation in America, Johns Hopkins Press; and
T.E. Keeler (1972) Airline regulation and market performance, Bell Journal of Economics
and Management Science, 3, 399–424.

In more detail, price-capping regulation adjusts the transport operator’s prices

according to a price-cap index that reflects the overall rate of inflation in the
economy, the ability of the operator to gain efficiencies relative to the average
firm in the economy, and the inflation in the operator’s input prices relative to the
average firm in the economy. The idea is that a technologically dynamic, profit-
maximizing supplier will have the incentive and the ability to respond to a price-
cap by reducing costs.
Price-capping regulation is sometimes called ‘CPI–X’ regulation (known
in the United Kingdom as ‘RPI–X’), after the basic formula employed to set
price caps. This takes the rate of inflation measured by the consumer (CPI) or
retail price index (RPI) and subtracts an expected annual efficiency saving (X).
The system provides an incentive for greater efficiency, as any saving above the
predicted rate X can be passed on to share-holders until the price caps are next
reviewed. The rate X is based not only on a firm’s past performance, but also on
the performance of other firms in the industry, and is intended to be a proxy for
a competitive market, in industries which are natural monopolies. The problem is
that, where there is manifest inefficiency, it is relatively easy to generate a reasona-
ble value for X, but as a transport undertaking moves towards a quasi-competitive
situation, X approaches a normal profit margin. In other words, it converges on
rate-of-return regulation.

BUTTON 9781786435668 PRINT.indd 527 27/04/2022 11:07

 528 TRANSPORT ECONOMICS, 4TH EDITION

Contestable Markets

There have been other perspectives on regulating what are seen as natural monop-
olies in transport. William Baumol et al.’s (1982) idea of contestable markets,
which was touched upon in Chapter 7, for example, argues that with free market
entry and exit, effectively the absence of any ‘sunk costs’, the forces of potential
competition temper the monopoly powers of any incumbent transport suppliers.
This logically moves to notions of unbundling those elements of transport supply
that may be subject to contestable forces and focusing regulation on those that
are characterized by elements of sunk costs. This was, for instance, the rationale
underlying the deregulation of European airlines and, more generally, modes
such as buses and trucking.
Although some studies have supported a weak notion of contestability, the
evidence supporting this tends to show that the forces of contestability are weaker
than those of actual competition. Scheduled airlines do not seem as affected by
potential competition as was initially thought. Moore (1986), for example, found
that in the United States fares were lower in markets with more actual carriers,
while Steven Morrison (2001) found that actual competition with Southwest
Airlines produced aggregate fare savings of $12.9 billion in 1998, whereas poten-
tial and adjacent competition produced a saving of only $9.5 billion. As a result,
even some of its strongest initial supporters have questioned the limits of its
usefulness: ‘We now believe that transportation by truck, barges, and even buses
may be more contestable than passenger air transportation’ (Baumol and Willig,
1986).

Regulating Sunk Costs

Where there are sunk costs, there have been arguments that the monopoly power
of transport undertakings may be controlled through initiating ‘competition for
the market’, rather than trying to stimulate competition in the market (Demsetz,
1968). This essentially involves auctioning off the rights to provide the transport
service. This may, for example, be used to supply infrastructure such as roads,
whereby a concession to build, operate, and toll the facility may be auctioned
off with the state taking back the road after some specified period. Basically, the
competition is in the rights to have the market for road space rather than in the
market for road use.
Several different auctioning methods, with numerous variants, each with
their advantages and limitations, are available and have been used in various con-
texts in the transport sector:

• English or ascending-bid auction. This is the most common form of auction

for the sale of goods, with an auctioneer allowing successively higher bids
until only a single bidder is left. At each point in the auction, the various
potential bidders know the prevailing highest bid. It is simple, but can lead to
the winner paying well below the maximum they would have paid.

BUTTON 9781786435668 PRINT.indd 528 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 529

• Dutch or descending-bid auction. In this model, the auctioneer gradually

lowers the price until a bidder accepts it. This tends to result in the winner
bidding close to the maximum they would be willing to pay because they
have no idea of any potential highest offers that may come from competitors.
• First-price sealed-bid auction. This approach involves bidders submitting
sealed bids, and the ‘winner’ is the one with the highest bid when they are
opened. In this case the bidders have no idea what the competing bids are
when they make their offer. This is the most common approach employed in
the transport context.
• Vickrey (1961) or second-price sealed-bid auction. This variant on the sealed-
bid model awards the ‘prize’ to the highest bidder but the winner only pays
the amount of the second-highest bid. This encourages each bidder to make
an offer of at least the maximum, but generally more than they would be
willing to pay because they realize any penalty of over-bidding will be off-set
by only paying the runner-up’s bid.

The difficulty with any auction is that it involves game-playing between the
auctioneer and those bidding, but with transport there can be specific challenges.
In many cases, because transport is essentially a network activity, there are issues
of the boundaries of the auction. For example, should it be for a single highway
or a set of highways, or one airport or a set of airports? There is also the physi-
cal longevity of much transport infrastructure and the problems that may arise
if the demand for its use unexpectedly declines significantly during the period of
the contract, making it impossible for the company winning the auction to earn
a viable return. This may involve the need for renegotiations. In some cases, for
example airport slots, the value of one service may be affected by the availability
of another. For example, gaining a slot at airport A to offer a service A => B is
only useful if a slot at B is also obtained. Secondary trading is common as a way
of reconciling this after the initial allocation has been decided.
Despite the challenges of devising these various franchising systems, compe-
tition for the market has been widely adopted in many transport contexts, ranging
from the auctioning of subsidized bus and rail routes in the United Kingdom (the
bid with the lowest subsidy requirement winning), to the awarding of tolled road
contracts in the United States, to auctions for airport concessions in many South
American countries.

Monopoly Suppliers and the Envelope Theorem

The envelope theorem highlights the differential effects of a price change on a

monopoly transport supplier’s profit (ΔΠ in Figure 14.4) and of social welfare.
What it shows is that a relatively large increase in output brought about, for
example, by fare or rate controls, can have a large first-order effect on social
welfare but limited second-order effects on the supplier’s profits.
Assuming for simplicity that the supply and demand functions are linear, the
profits curve seen in the figure rises with the quantity sold, is maximized (QM), and

BUTTON 9781786435668 PRINT.indd 529 27/04/2022 11:07

 530 TRANSPORT ECONOMICS, 4TH EDITION

∆SW Social
Welfare

∆Π

Profits

0 QM QR QC
Quantity

Figure 14.4 Social welfare and monopoly profit effects of a price reduction

then declines. In the traditional demand/cost presentation, the zenith is where the
supplier’s marginal cost is equated with its marginal revenue. Regarding the social
welfare function, representing the combination of profits and consumer surplus,
this takes a similar path but is maximized when profits fall to zero (QC). This is
where average costs equal marginal revenue in the standard monopoly model.
If a regulation or a price control is introduced, this will cause the monopolist
to increase the quantity of transport services that are sold which in turn will affect
both profits and social welfare. Because the profits curve is relatively flat about its
zenith, this will mean a small decline in profit (ΔΠ), but will be accompanied by
a larger increase in social welfare (ΔSW). This result obviously only holds locally
for arbitrary small changes from the profit-maximizing point.
Additionally, the idea relates to situations where a monopolist transport sup-
plier is charging a single price. It is less valid when the supplier engages in price
discrimination, as is common in the scheduled airline market and in many mer-
cantile markets. In these latter cases the profit curve, because it involves charges
that extract considerable amounts of consumer surplus, will approach that of the
social welfare curve.
While the findings are not universal, substantial gains in social welfare can
come from reductions in fares or increases in transport services without a large
impact on supplying transport companies’ profits.

14.4 Prioritizing Transport Policies

We turn now to examine the various landmarks and phases of transport policy.
Recalling Figure 1.1, our attention is largely on the economics of property rights
elements of institutions and governance issues involving transaction costs. Social
theory is not really a concern, because our attention is largely on the market econ-
omies of North America and Europe, and micro issues are market outcomes of
the regulatory structures in place and the ways in which they are operationalized.

BUTTON 9781786435668 PRINT.indd 530 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 531

This section is intended to highlight the way in which transport policy has
evolved to match both the changing technical and organizational structure of
the sector, and has responded to the differing attitudes of society to transport
over the past century and a half. It is in no way intended as a comprehen-
sive piece of economic history. Much of the focus is on the United Kingdom
­situation because it offers both a more diverse portfolio of regulatory policies
and because, unlike countries with federal systems such as the United States,
policy is largely controlled by a single authority. But changes elsewhere in the
world are also touched upon. The ‘story’ is slanted in favor of more recent
developments.

The Anti-monopoly Phase, Pre-1930

Government has always taken some interest in transport. Traditionally, it had

not been the efficiency of the transport market that was the principal concern
but rather the insurance that adequate transport was available to fulfil the
commitments of defense, administration, and internal stability. The Industrial
Revolution placed greater emphasis on the need for an economically efficient
transport system which, combined with the rapid technical changes that took
place in transport during the later years of the eighteenth century and the early
decades of the nineteenth century, resulted in more official intervention. (Indeed,
even the Duke of Wellington called for a National Transport Plan for the United
Kingdom at one stage.)
While much of the early involvement was to permit canals and railways to
be constructed, there was also a political awareness of the potential monopoly
powers which such companies could exercise. The 1844 Railway Act in the United
Kingdom, for example, gave government the option of purchasing newly formed
companies after 21 years (a power never exercised) and maximum rates were nor-
mally included in the enabling acts. In all, over 200 regulatory acts were passed
before 1930 to appease public concern about the private control exercised over the
monopoly of the railway companies.
The Americans followed closely and copied British railroad technology. The
Baltimore and Ohio Railroad was the first common carrier and started a passen-
ger train service in May 1830, initially using horses to pull trains. While fostering
private ownership and finance, America had passed its first legislation on rate
regulation in 1856. Subsequently, Congress passed the Interstate Commerce Act
in 1887, creating the Interstate Commerce Commission. The agency’s original
purpose was to regulate railroads to ensure fair rates, to eliminate rate discrimina-
tion, and to regulate other aspects of common carriage.
By the 1920s the British railway companies had to publish their rates, were
subject to common carrier obligations, were not allowed to show undue prefer-
ence, had to present accounts in a prescribed manner, and were subjected to
controls over wage bargaining. An element of social service obligation was also
apparent in some of this legislation with, initially, railway companies required by
law to provide specified cheap services and, subsequently, workmen’s tickets.

BUTTON 9781786435668 PRINT.indd 531 27/04/2022 11:07

 532 TRANSPORT ECONOMICS, 4TH EDITION

This type of situation was not unique to the United Kingdom. At the extreme,
many Continental European countries, partly for strategic reasons but also in part
because of their general approach toward natural monopoly, had developed their
rail systems as public enterprises from the outset. Equally, in Canada the inherent
advantage of the railways led to concern about the abuse of their powers, includ-
ing discriminatory practices, and to the passing of the Railway Act of 1903.

The Anti-competition Phase, 1930–45

Advances in trucking and passenger transport in the period after the Second
World War substantially eroded the near-monopoly that had been enjoyed by
the railways for the previous 80 or so years. The nature of road transport, and
especially the relative ease of entry and the low capital requirement necessary,
changed the focus towards the regulation of dangerous operating practices and
excessive competition. Political pressures from the railway companies, still ham-
pered by numerous restrictions on their commercial freedom, to contain cheap
road transport, added emphasis to the debate.
In Britain, besides the financial strains on the regulated railway industry,
fears were expressed that excessive competition produces dangerous operating
practices and results in inadequate and unreliable services for those situated away
from the main transport arteries. The Salter Conference, which reported on truck-
ing in 1932, found that:

Any individual at present has an unlimited right to enter the haulage industry without
any regard to the pressure on the roads or the existing excess of transport facilities ….
This unrestricted liberty is fatal to the organization of the industry in a form suitable
to a carrier service purported to serve the public.

(It should perhaps be said that the composition of the Salter Conference – it was
made up of railwaymen and representatives of established truckers – may have
colored its conclusions.)
The legal manifestations of this policy were the passing of the Road Traffic
Act, 1930, and of the Road and Rail Traffic Act, 1933, which introduced, respec-
tively, quantity licensing into road passenger transport and trucking. In addition
to trying to temper the competitive environment of road transport, the operation
of the licensing system for road passenger transport encouraged the provision and
cross-subsidization of unprofitable social bus services by virtue of the method of
license allocation. Other measures designed to produce greater equality in the
operating conditions encountered on the roads and railways included relaxation
of certain constraints on railway pricing and the introduction of minimum wage
rates and specified employment conditions into the industry.
Elsewhere a plethora of similar regulations were appearing. In the United
States, federal legislation was introduced, controlling coastal shipping (1933),
inter-urban bus operations (1935), trucking (1935), airlines (1938), inland water-
ways (1940), and freight forwarders (1942), and many states also tightened their
regulatory regimes. Canada initiated regulations over its aviation industry under

BUTTON 9781786435668 PRINT.indd 532 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 533

the 1938 Transport Act, and encouraged coordination, rather than competition,
between its main railway operators under the 1933 Canadian National–Canadian
Pacific Act.

Central Control and Nationalization in Europe, 1945–51

The years immediately following the end of the Second World War saw in the
United Kingdom a period of reconstruction and industrial policies emphasiz-
ing the nationalization of industry. The 1947 Transport Act, for the first time
outside of war years, brought a substantial part of the British transport system
under direct government control. (Although peacetime nationalization was not
entirely new, the state-owned British Overseas Airways Corporation was formed
in 1939.)
The objectives of the 1947 Act were to secure the provision of an efficient,
adequate, economical, and properly integrated system of public inland transport
and port facilities. The earlier success of the London Passenger Transport Board
gave credence to this view and the British Transport Commission (BTC) was
established to emulate. It was given the responsibility of coordinating the newly
nationalized railways, long-distance road haulage, sections of public road trans-
port, London Transport, and publicly owned ports and waterways.
It was recognized that for railways to compete efficiently, a much closer
price–cost relationship was required. It was impossible, however, for a new set
of freight charges to be set before 1955, and meanwhile the BTC was impotent
to direct traffic to the mode for which it thought it was best suited. The difficulty
was compounded by consignors having a free choice of transport mode, and that
own-account trucking vehicles were free from government control. Further, the
1947 Act had little opportunity to work because of, first, the lack of time needed
to develop the necessary administration and to reorganize the newly nationalized
undertakings, and, second, the inadequate funds available to carry out the neces-
sary investment required in the rail sector.
In the United States, the post-Second World War boom saw many rail-
roads driven out of business due to competition from airlines and the emergent
Interstate highways. The rise of the automobile led to the end of passenger
train services on most railroads. Trucking businesses had become major com-
petitors by the 1930s and acquired an increased market share of freight business.
Railroads continued to carry bulk freight such as coal, steel, and other commodi-
ties. Despite the considerably increased competition the railroads encountered,
they continued to have their rates and operations regulated, offering them little
flexibility in responding to changing market forces. There were no significant
moves to nationalize.

A Competitive Framework, 1951–64

The period of successive Conservative administrations between 1951 and 1964 in

the United Kingdom witnessed a movement away from a regulatory approach to

BUTTON 9781786435668 PRINT.indd 533 27/04/2022 11:07

 534 TRANSPORT ECONOMICS, 4TH EDITION

transport coordination and towards one based upon competition that made use
of the natural interplay of economic forces. A policy of decentralized control was
pursued and, under the 1953 Transport Act, the railways were freed from many
of their long-standing statutory obligations (for example, regarding common
carriage and publication of rates and charges). Large sections of nationalized
trucking were returned to private ownership. The basic argument for this com-
petitive approach was that even if integration in the fullest sense were practicable,
it would result in a large, unwieldy machine, ill-adapted to respond rapidly to the
varying demands of industry.
The culmination of these policies was the 1962 Transport Act which freed
the railways from most of their remaining legal constraints, regionalized their
boards, separated their overall administration from that of trucking, and recog-
nized that commercial viability required some rationalization of the rail network.
Nationalized road haulage was given terms of reference requiring it to perform as
though it were a ‘private enterprise concern’.
The general policy of ‘coordination by competition’ pursued by the
­government met with considerable problems after 1962. The railways and public
transport in general ran into increasing financial difficulties, while public concern
began to grow over both the need to provide additional road space to accommo-
date growing motor numbers and over the environmental effects of this growth,
especially on urban areas. The established method of providing social services
by cross-finance from profitable services ceased to be financially viable. Further,
there were questions arising concerning the most appropriate ways of exploiting
new transport technologies, especially containerization.

Controlled Competition, 1964–74

The Labour administration elected in the United Kingdom in 1964 undertook a

series of detailed policy studies that culminated in the 1968 Transport Act. This,
in conjunction with the 1968 Town and Country Planning Act, sought to produce
a policy based upon ‘controlled competition’. It hoped to combine the advan-
tages afforded by the automatic processes of the market mechanism with those
of direct control. The policy was comprehensive, setting up passenger transport
authorities (PTAs) to control and coordinate urban public transport, devising
new systems of quality licensing for trucking, providing specific finance to support
socially necessary transport, drawing up a national road-building program, and
reconstructing the accounts and activities of the railways. It used all means of
policy: market measures such as taxation and subsidy, as well as ‘administrative’
measures such as licensing and the establishment of new institutions.
The Act was also liberal in the sense that the framework of legislation and
licensing that was introduced was intended to provide a basis for ‘fair’ competi-
tion, leaving the coordination of modes, in most sectors, to market forces. The
PTAs were introduced in the major conurbations to control urban public trans-
port, but this was in a market where externalities are widespread and competition
felt to be unlikely to solve problems of congestion and environmental decay. The

BUTTON 9781786435668 PRINT.indd 534 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 535

inter-urban market was left to competitive forces with intervention limited to

preventing excess instability or dangerous practices.
The Conservative government returned to office in 1970 and pledged itself to
the repeal of many of the clauses in the 1968 Act, although, in practice, few of the
measures they found exceptionable – such as quantity licensing for long-distance
trucking – had in fact been implemented. Indeed, the 1972 Local Government
Act set up authorities that logically extended the PTA concept and further inte-
grated the long-term planning of overall urban development. The Civil Aviation
Authority, which was set up in 1971, rather than reversing the philosophy of the
1968 Act, was designed to control operating practices and improve the stability
of the sector. As the Minister for Trade said at the time: ‘There are close links
between the economics and the financial health of the airlines and the safety of
their operations and between operational safety, air worthiness, air traffic control
and navigational services’.

The Search for Efficiency, 1974–80

The policy emphasis of the Labour government from 1974 to 1979 was spelt out
in the government’s document, ‘Transport policy’:

First, to contribute to economic growth and higher national prosperity … . Second,

to meet social needs by securing a reasonable level of personal mobility … . Third, to
minimize the harmful effects, in loss of life and damage to the environment, that are
the direct physical results of the transport we use.

In many ways the policies, which reduced government expenditure con-

siderably, moved further towards the free market than the 1968 Transport Act,
with greater emphasis now placed on allocative efficiency. Although government
expenditure on ‘roads and transport’ fell from £3,820 million to £3,023 million
per annum (at 1978 prices) between 1974 and 1978, this does not mean official
intervention was ignored. The 1974 Railway Act, for example, had already intro-
duced the idea of a ‘Public Service Obligation’, and subsidies were made available
to provide services ‘comparable’ to those existing at the time of the legislation.
Methods of allocating public monies were also modified with the introduc-
tion of transport supplementary grants and the need, from 1974, to produce
transport policies and programs by local authorities to demonstrate the evolution
of coherent local transport policy. This latter idea was later extended in the 1978
Transport Act to embrace public transport services (the Public Transport Plans)
at the shire level. But, on the other hand, grand strategies for road building were
replaced by piecemeal assessments emphasizing a further shift from the active
policy of the Labour administration of the 1960s to rather more reactive policy-
making (Gwilliam, 1979).
The emphasis on allocative efficiency and market mechanisms was further
strengthened in the 1980 Transport Act introduced by the Conservative govern-
ment that had been returned the previous year. Quantity licensing in the express
coach sector was abolished and free competition (with quality controls) was

BUTTON 9781786435668 PRINT.indd 535 27/04/2022 11:07

 536 TRANSPORT ECONOMICS, 4TH EDITION

permitted. Further relaxation of sharing and lift-giving laws also allowed more
opportunity for private transport to offer limited forms of public service in rural
areas, while policies of denationalization resulted in sales of limited amounts of
publicly owned transport assets to private industry. (It is interesting to compare
this trend towards both reactive policy-making and a greater reliance on market
forces with the developments in the European Community’s Common Transport
Policy discussed in Chapter 13.)

The Age of Regulatory Reform, 1980 to Late 1990s

The period since the late 1970s has witnessed major reforms in transport regu-
lation. While there are marked national differences in the nature and pace of
change, this process has, in the case of public transport modes, been character-
ized by moves toward more liberal regimes and a withdrawal of government from
the ownership of transport operating companies. In contrast, there has been an
increasing emphasis on the containment and the management of automobile use
both on environmental grounds and because of congestion problems.
The liberalization measures that have been adopted not only represented
legal reforms, but also embraced de facto changes in interpretation and enforce-
ment of regulations. In addition, they have extended across national boundaries
with, for instance, the removal of institutional barriers to free transport opera-
tions being an explicit element of the European Community’s 1992 initiative.
Whereas previously the majority view was that, because of scale economies and
the potential for serious market failure, it was in the public interest for govern-
ment to take an active role in regulating the industry, the prevailing wisdom is
now that intervention failures are often potentially more damaging than market
imperfections.
The generality of the regulatory trends across industrialized countries raises
questions concerning the underlying causes of change. Certainly, in more recent
years one can point to the demonstration effects exerted mainly by the deregula-
tion of United States domestic aviation, but there also seem to be more funda-
mental issues that need addressing.
The first American economists to oppose regulation did so for a straight-
forward reason. Most subscribed (and still do) to the basic theory that welfare
is maximized when the price of each good or service equals its long-run social
marginal cost, but evidence mounted in the 1950s, 1960s, and 1970s that regula-
tory agencies caused prices to diverge from long-run social marginal costs, rather
than converge with them. Economic studies of United States aviation provided
a cornerstone in this debate. It was the combination of direct evidence that the
CAB kept long-haul, high-density fares high and evidence of low fares in the
unregulated intra-state markets in California that provided the first evidence that
deregulation of inter-city air passenger transportation was justified. Indeed, in
the case of airlines, the California markets provided evidence against airline regu-
lation, evidence that could be understood not only by economists, but also by the
general traveling public (see again Table 14.1).

BUTTON 9781786435668 PRINT.indd 536 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 537

By the mid-1960s, a strong case existed for transport deregulation in the

United States but there was much research needed to get a full picture; for
example, early studies disagreed as to whether the costs of excess rail capacity
resulting from closure regulation were high or low. Several subsequent studies
considered these issues, as well as the more basic issue regarding how high the
social loss from transport regulation was in the United States.
There were also theoretical developments in economics that encouraged
deregulation. The Chicago models of regulation made it clear that the public
interest is not necessarily served by regulation. However, these models also implied
that existing regulations were the result of a rational process, and by the interpre-
tation of some they could be thought to work against deregulation, because they
implied that regulation was itself a result of social optimization. Nevertheless,
subsequent analyses have extended the analysis to indicate how it could indeed
explain regulatory reform. The reason is that changes in technologies, markets,
and the balance of political power among different groups can easily stimulate
regulatory reform, even in a Chicago-type model. This has been examined in the
context of the bus (Button, 1989) and airline industries (Keeler, 1984).
Another theory which supported deregulation was that of contestability,
which asserted that, with sufficiently easy entry and exit in a market (there are
no sunk costs but there are lags in matching price cuts), even a natural monopoly
could have a zero profit, competitive outcome. The model, however, served more
as an after-the-fact rationale for deregulation than as a stimulus, because it was
only developed after deregulation occurred in most United States industries.
Furthermore, as we have seen, subsequent empirical evidence for the United
States airline industry has not fully supported the contestability hypothesis.
The resultant changes that have taken place across the world since the mid-
1960s have not been uniform either across countries or across modes. Part of this
stems from the different starting points that existed in the early 1970s. In very
general terms, the United States led the way in legal change with a rush of liberal-
izing measures during the 1970s and early 1980s, which removed much govern-
ment control from its domestic transport system – see Table 14.2. Other countries

Table 14.2 Major legal changes in United States transport regulation (1976–89)

1976 Railroad Revitalization and Reregulation Reform Act: removed many

regulations over rate setting
1977 Air Cargo Deregulation Act: initiated free competition for air cargo services
1978 Airline Deregulation Act: initiated a phased removal of fare setting and market
entry controls
1980 Staggers Rail Act: removed many regulations over line abandonments and gave
further freedom over rate setting
1980 Motor Carriers Service Act: increased entry and rate setting freedom and
reduced the role of rate fixing bureaux
1981 Northeast Rail Reform Act: enabled Conrail to abandon little-used lines
1982 Bus Regulatory Reform Act: eased conditions of market entry and exit and
phased in relaxation of rate controls

BUTTON 9781786435668 PRINT.indd 537 27/04/2022 11:07

 538 TRANSPORT ECONOMICS, 4TH EDITION

have followed the trend, in part because of demonstration effects that indicated
significant benefits from change but in some cases because of the direct impacts
of changes in the United States – Canada, for example, was so affected.
In effect, the systems of Boards and Committees, which regulated the indus-
try, have gradually had their powers curtailed and transport industries are increas-
ingly being treated like other commercial undertakings. Where controls remain,
there has been a shift in their emphasis. In inter-state trucking, for example, the
1980 legislation shifted the burden of proof for market entry from the applicant
to the protester and this effectively eliminated the major entry barrier. It also
created a zone of reasonableness within which rates could vary. Equally, with
respect to freight railroads, the reforms gradually removed rate controls, allowed
rationalization, and facilitated reorganization, especially mergers.
This quest for efficiency has been emulated in the United Kingdom as, first,
the 1980 Transport Act, and then the 1985 Transport Act liberalized bus markets
and introduced new financial mechanisms for providing social services. De facto
reform was brought about in domestic aviation as the Civil Aviation Authority
adjusted its position on licensing in 1982. Developments within the European
Union meant that there would be further de jure changes as cabotage was phased
in within the Community by 1997. Indeed, as outlined in Chapter 12, the creation
of a Single European Market meant that cabotage rights, albeit initially often
in rather limited forms, extend across all transport modes within the European
Union.
Similar changes occurred elsewhere with the domestic Canadian truck-
ing market, partly as a result of increased competition from the more efficient
American carriers being liberalized in 1987 and its aviation market, after some
de facto changes in 1984, being legally deregulated in 1988. Reforms of a similar
nature ended Australia’s ‘two airlines’ domestic aviation policy in 1990, but liber-
alization of surface transport has progressed more slowly.
The extensive public ownership of transport that existed in countries such
as Canada, Australia, Japan, and most of Europe has meant that reforms there
have often also involved elements of privatization. The privatization process,
while reducing the direct control of government, has stimulated the creation of
new regulations, for example, to limit market power and to meet social objectives.
Many of these relate to quality of service, especially safety, and to the nature of
ownership (such as governing allowable foreign ownership of shares), but eco-
nomic regulations have also been imposed.
The sale of transport undertakings through stock market flotation (for
instance, British Airways and the British Airports Authority) attracted considera-
ble attention and raised large sums of money for the United Kingdom Exchequer
(Table 14.3). Privatization in transport has also taken a diversity of other forms.
The former National Bus Company in the United Kingdom, for instance, was
broken up and privatized mainly through management buy-outs. In France and
Japan private money is increasingly being used to finance railway operations
through commercial loans and investments. Open tendering for formerly publicly
supplied bus services is now widespread in the United Kingdom, and franchising

BUTTON 9781786435668 PRINT.indd 538 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 539

Table 14.3 Transport privatization proceeds in the United Kingdom

Undertaking Year Amount (current prices)

National Freight Corporation 1982 £7 million

British Rail Hotels 1983 £30 million
Associated British Ports 1983–84 £34 million
Sealink 1984 £ 66 million
British Airways 1987 £892 million
British Airports Authority 1987 £1,200 million
National Bus Company 1988 n.a.

Note: n.a. = not available.

systems, albeit on a somewhat different basis, exist in a few Continental European

states. In a similar way, Sweden attempted to increase efficiency on parts of its
railway network by separating operations from track and allowing private oper-
ating companies to tender for services. Actions were also initiated in the United
Kingdom although extended, for a period, to embrace privatization of the track
with somewhat less success (Bowman, 2015). Overall, the experiences have been
mixed.
The regulatory actions in higher-income countries during the 1980s and
1990s had wider knock-on effects. Gradually the effects of reform strategies,
especially through the influence and actions of agencies such as the World
Bank and regional multi-lateral agencies such as the Asian and Inter-American
Development Banks, are also being felt in developing countries. Table 14.4, for
example, lists some of the regulatory reforms of local public transport systems
that occurred in lower-income countries, including transition economies. There
was a general relaxation of economic controls, but these took a diversity of forms,
often dependent on the wider responsibilities of the relevant authority and wider
national economic and political reforms.

The Notion of ‘Sustainable Development’, Late 1990s to Date

We saw in previous chapters the impact that transport can inflict on both the built
and the natural environments. Regulations of various kinds have long been used
to minimize these problems, but the nature of the problems being addressed and
the forms of regulation deployed have, as we have seen in Chapter 8, been chang-
ing. The use of regulation, and to a lesser extent, fiscal instruments, have reduced,
if not necessarily optimized, many of the local and regional adverse external
effects. There seems every indication, however, that the political economy of
optimizing such things as global warming gas emission will remain challenging.
In broad terms this fits with a wider policy shift that focuses on sustainable
development, within which transport policy plays a significant role.
The period since 1980 has witnessed a complete transformation in the way
in which policy-makers think about regulating transport markets. The role of
government in regulation has moved from directing the operational sides of the

BUTTON 9781786435668 PRINT.indd 539 27/04/2022 11:07

 540 TRANSPORT ECONOMICS, 4TH EDITION

Table 14.4 
Summary of changes in the role of public transport regulatory authorities in
developing counties

Change in role of regulator/transport authority and Examples

market structure examples

Regulation of urban passenger transport introduced Ghana; Manila (Philippines)

or restored
Regulation of passenger transport transferred to the Jordan
city
Regulatory capacity developed within local Accra & Kumasi (Ghana); Amman
government (Jordan); Tbilisi (Georgia); Manila
(Philippines)
Changes made to rules for market entry &/or Accra (Ghana); Lagos (Nigeria);
procurement Tbilisi (Georgia)
Regulators develop new models for procurement of Recife (Brazil)
services
Regulators develop new contractual basis for bus Kaunas (Lithuania); South Africa;
services, or amend existing ones Santiago (Chile)
Regulators make direct intervention to restructure Sri Lanka
private operators
Transport authority makes vehicles available to Accra (Ghana); Tbilisi (Georgia);
operator Dhaka (Senegal)
Transport authority provides funding or financial Kaunas (Lithuania); South Africa;
incentives for bus purchases &/or technology rural areas in Brazil
upgrades
Transport authority finances transport Accra (Ghana); Amman (Jordan);
infrastructure to enhance public transport system China; Dar es Salaam (Tanzania);
quality & level of service & to provide improved Delhi (India); Mauritius; many
operating & maintenance facilities to operators other cities

Source: Finn and Mulley (2011).

sector to looking for innovative ways to finance its infrastructure and control its
wider impacts on society. Part of this stems from wider developments concerning
the perceived role of government per se, but the changes in transport have also
come about because of specific concerns. The outcome has seldom been as pre-
dicted and certainly the more liberal conditions which now exist in most transport
markets around the world are not without their problems. In general, however,
the changes have afforded the opportunity for users to express their preferences
through the market and for regulators to pinpoint specific distortions where inter-
vention can be justified.
In summary, if there is a single conclusion to be drawn from the experience
of regulatory reform in transport, it has been that reform has, in general, been
an economic success. Though there have been areas in which the policies need
improvement (anti-trust, infrastructure, franchise arrangements, etc.), these are
small matters of ‘fine-tuning’ compared with the broad picture of overall success
in improving the efficiency of transport provision. What have been lacking are
comparable reforms in the regulation of private transport use.

BUTTON 9781786435668 PRINT.indd 540 27/04/2022 11:07

 541
POLITICAL ECONOMY AND TRANSPORT REGULATION ­

14.5 Paths of Regulatory Reform

Regulations are continually changing as we saw in our discussion of institutions

in Chapter 1. This reflects a variety of factors, but here the focus is more on the
ways in which the regulatory reforms are enacted. We saw in Chapter 13 that the
European Union, for example, staged its reforms of aviation policy over a series
of ‘Packages’. This was not the way the United States approached the challenge
in the case of air transport. This raises issues about the speed at which change in
transport regulations is optimal from an economic perspective. Linked to this is
the matter of how the transition is best managed, and the role that different incen-
tives and instruments may play in this.

‘Big Bang Approach’ Versus ‘Phased Reform’

There are two basic ways in which regulatory reforms are initiated; they either
come in at once as a ‘big bang’ or they are phased-in in stages. There are eco-
nomic advantages and disadvantages of each. The big bang approach, typified by
the individual modal regulations in the United States in the late 1970s and early
1980s, involves a comprehensive legal change in regulations that comes into effect
very rapidly. There are potential costs in this (Figure 14.5).
First, the reforms may leave significant ‘stranded costs’ in the industry. For
example, the infrastructure and hardware of the industry will be designed for the
prior regulatory regime and some of it will be redundant under the structures
of supply that emerge after reforms, or at least not be ideal in the new situation.
Second, the new measures may not themselves be optimal. Changes in regulatory
structures impose a wide variety of unforeseen costs on any industry and some
of these may be larger than anticipated. Equally, some of the possible benefits of
reform may be lost if the reforms are not optimal.

+
Big bang approach

Net
benefit
Phased approach
0
A B Time

Figure 14.5 The ‘big bang approach’ versus the ‘phased reform’

BUTTON 9781786435668 PRINT.indd 541 27/04/2022 11:07

 542 TRANSPORT ECONOMICS, 4TH EDITION

The advantages, on the other hand, are that the big bang change removes con-
siderable uncertainty from the market about the nature of future institutional
structures and should, therefore, remove some of the challenges in making invest-
ments. It also means that there is less opportunity for those in the industry, or
for others with related interests, to manipulate the various stages in the reform
process, as may be the case if the changes are phased-in over time. Additionally,
there is the expectation of a significant flow of benefits once the reforms have
worked their way through the system.
In contrast, a series of smaller changes phased-in allows for stranded costs
to be minimized as transport hardware is physically worn out before it needs to be
replaced and it allows modification to the new regulatory structure in the light of
the experiences gained in the previous phases. But it is prone to capture by vested
interests seeking to protect their positions and it takes longer for the benefits of
the new regime to work their way through.
In practice, the decisions regarding which approach to adopt are often more
to do with political economy than the simple calculation of costs and benefits.
Reform generally requires a common perspective of a coalition of interests and
this shapes the pattern of change. A relatively strong and established federal
structure, such as in the United States, makes it much easier to adopt a big bang
approach than in the European Union where basically a strong informal coalition
of members is needed to carry through change.
Considerable attention is paid in economics to comparative-static compari-
sons, normally in terms of efficiency, between the implications of one regulatory
regime and another, and to a lesser extent to the merits of the big bang and
phased approaches, but there are also more detailed issues of the transition
process (Meyer and Tye, 1981). Much depends on the peculiarities of the situa-
tion. The matter is seldom purely economic in nature and, for example, there are
often long-standing legal contracts that make sudden change difficult, or can only
be circumvented by expensive buy-outs. There are also equity considerations. The
loss of sunk costs is not just a matter of efficiency but adversely impacts on those
who had incurred them under the old regulatory regime, whereas the gains of
change are enjoyed by new investors.

United States Versus European Union Transatlantic Aviation Policy

The development of international air transport markets on the North Atlantic

offers an example of the ways phasing of regulatory changes may occur and their
economic implications. The restrictive bi-lateral air service agreements that typi-
fied the institutional structure of international airline markets from the Chicago
Convention of 1944 have gradually been eroded by multi-lateral actions of enti-
ties such as the European Union and the advent of the Open Skies initiative by
the United States (see Table 14.5 for details). This initially involved a loosening
up of individual bi-lateral agreements between countries such as Germany and
the Netherlands and the United States, but this was done in an ad hoc fashion,
and agreements with countries such as the United Kingdom remained restrictive.

BUTTON 9781786435668 PRINT.indd 542 27/04/2022 11:07

 Table 14.5 Main features of United States bi-lateral air service agreements

Pre-1978 bi-lateral air service agreements 1978–91 open market bi-laterals Post-1991 Open Skies
bi-laterals

BUTTON 9781786435668 PRINT.indd 543

US airlines Foreign airlines

Market Only to specified points From any point in the US to Access limited to a Unlimited
access specified points in foreign countries number of US points
Limited 5th freedom rights granted to US     Extensive 5th freedom rights granted Unlimited 5th freedom rights
carriers
Charter rights not included    Unlimited charter rights
     7th freedom rights not granted
      Cabotage not allowed
Designation Single – some multiple Multiple
Airlines must be ‘substantially and effectively controlled’ by nationals of designated state
Capacity Capacity agreed or shared 50:50 No frequency or capacity controls
No capacity/frequency controls under liberal
bi-laterals, but subject to review
Break of gauge permitted in some Break-of-gauge rights
agreements granted

Tariffs Approval by both governments (double- Double-disapproval (filed tariffs Free pricing
approval) or as agreed by IATA operative unless both governments
disapproval) or country of origin
rules
Code- Not part of bi-laterals Code-sharing permitted
sharing

27/04/2022 11:07
 544 TRANSPORT ECONOMICS, 4TH EDITION

Negotiations to reach a more comprehensive structure emerged when the European

Commission was given authority to negotiate on behalf of all Union airlines. The
effects of such changes have not been easy to isolate, but Figure 14.6 offers a
general representation of the issues that are involved in looking at what has hap-
pened. It highlights the potential fare and output implications of the various types
of transatlantic regulatory regimes that have been considered.
The initial position of the demand curve for transatlantic services under
the pre-1980s regulatory regime is assumed linear and shown as D1 in the figure,
and the average cost curve per passenger, which for simplicity is assumed to rise
more than linearly with quantity, as C1. Market forces, however, because of the
institutional interventions in place, did not determine fares and capacity across
the Atlantic. Capacity under this system was limited (seen as the ‘capacity con-
straint’) and fares were regulated. If we assume that the terms reached under the
bi-lateral agreement regarding fares allowed for at least cost recovery by the part-
ners’ airlines, this implies a fare level up to F1. The removal of both this capacity
constraint and of negotiated pricing, as happens under the Open Skies arrange-
ments, results in competition for air services, and a move toward cost-recovery
pricing strategies by the carriers. This would reduce fares to F*1.
Open Skies policies that free-up market entry by carriers and the fares they
may levy, coupled with the permitting of strategic alliances, not only removes the
capacity constraint but also affects both the demand and supply curves for trans-
atlantic air travel. The ability of airlines to more effectively feed their transatlantic
routes and coordinate their activities, through the restructuring of their business
and networks, will reduce the average cost of carriage to C2 in the figure. The
effect is often reinforced due to downward pressure on costs because, although
not strictly part of the Open Skies framework, the wider competitive environment
within Europe, and the privatization of many carriers, by heightening commercial
pressures, reduces the amount of both static and dynamic X-inefficiency in the
Capacity
$ constraint C1

C2
F1

F *1
F2

D2

D1
0 Q1 Q*1 Q2

Figure 14.6 The economics of Open Skies policies

BUTTON 9781786435668 PRINT.indd 544 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 545

airline industry. In other words, there is the combined pressure of both free airline
markets across the Atlantic and within the two feeder markets at either end.
The Open Skies policy also has stimulation effects on the demand side.
By allowing more effective feed to the long-haul stage of transatlantic services
through the concentration of traffic at international hub airports, it increases the
geographical market being serviced and generates economies of scope and scale.
The larger physical market demand, combined usually with the improved quality
of the ‘product’ that accompanies more integrated services, such as code sharing,
interchangeable frequent flier programs, common lounges, and through baggage
checking, pushes out the demand for international air services to D2 in the figure.
The outcome of the lowering of costs and the outward shift in demand is
that the number of passengers traveling increases to Q2 and, because Open Skies
allows price flexibility, the fare falls to F2 in the way our example is drawn. It
should be noted that fares might not actually fall; indeed, they may rise as the
result of the freer market conditions. The reason for this is that the outward
shift in demand reflects a better ‘quality’ of service – for example, more conveni-
ent flights, transferability of frequent flier miles, and seamless ticketing – and
that, on average, potential travelers are willing to pay more for this than the
generic portfolio of features that were found under the old bi-lateral air service
structure. (The shift out in demand may counteract the fall in costs resulting in
F*1 < F2.)
What does become pertinent, however, is the extent to which the fare struc-
ture is influenced by the market power of the airlines. The analysis presented in
Figure 14.4 assumes that, in the Open Skies environment, fares are set to recover
costs; in other words, competition and mergers policy can effectively fulfil the
role of regulation. This raises issues as to the nature of a market served by a rela-
tively small number of network carriers. A degree of competition exists between
the various alliances for the trunk hauls market, and there is also competition
at either end of routes with many other, including low-cost, carriers competing
for passengers in overlapping feeder and origin–destination traffic to interna-
tional hub airports. There are also theoretical reasons derived from game theory
suggesting that the outcome in a market with three players approaches that of
competition. Nevertheless, each alliance, by dint of product differentiation (for
example, they serve different airports), inevitably enjoys some degree of monop-
oly power. This could lead to fares higher than F2 and a smaller output than Q2
with consequential reductions in consumer surplus.
The effects of a full Open Aviation Area – a genuine open market – can be
considered an extension of this framework. Free capital markets, together with
the ability to have more flexible feeder networks owned by the truck carrier at
both ends of transatlantic services, would further lower costs and may generate
additional economies of market presence, although this latter effect is unlikely to
be large. The ability to invest across national boundaries provides for short-term
support in situations of local market fluctuations and more integrated long-term
planning of infrastructure; it would in effect produce air networks akin to those
enjoyed by United States railroads, which can move investment funds across

BUTTON 9781786435668 PRINT.indd 545 27/04/2022 11:07

 546 TRANSPORT ECONOMICS, 4TH EDITION

states rather than have separate rail companies each limited to intra-state opera-
tions. In terms of Figure 14.5, it would mean lower fares and larger air traffic
volumes with concomitant increases in society benefits.

14.6 Studying Regulatory Reform

The impacts of reforms in the economic regulation since the late 1970s have been
extensively researched and debated, but the impossibility of accurately determin-
ing counterfactuals – what would have happened in the absence of regulatory
change – leaves much room for debate. Furthermore, many markets have not yet
reached long-run equilibrium. While some sectors such as the United Kingdom’s
bus industries and the United States’ domestic aviation industries were liberalized
some time ago, many international transport sectors are still in the process of
changing, merging, and shaking out. Infrastructure regulations have also gener-
ally been more recent and, because of the physical nature of the assets involved,
take longer to impact.
Experience so far, however, provides some insights on the working of trans-
port markets and the benefits and costs of regulation, but it is not complete,
and infrastructure often remains heavily regulated and some of it is still state-
provided. Also, there have been problems with some of the reforms initially intro-
duced, as for example with rail track in the United Kingdom, and thus there have
been subsequent modifications that make assessment murky.
Overall, however, simply looking at the crude data indicates that regulatory
liberalization during this period conferred significant benefits on the users of
transport services, as efficiency gains from ‘deregulation’ and privatization were
passed on to consumers in more competitive markets. A simple indication of
this is seen in the pattern of European air transport costs throughout the 1990s,
when reforms were taking place (Figure 14.7); fares and rates fell, to the benefit
of customers.
One important issue in the working of transport markets, as we have seen
in Section 14.3, is that of contestability. Evidence so far indicates strongly that
actual competition is considerably more effective in reducing market power than
is potential competition. But even those skeptical of the contestability of trans-
port markets nevertheless find evidence that the market power of firms in these
markets is nowhere near strong enough to justify regulation. That is, it can be
strongly argued that unregulated markets in transport function quite well by rea-
sonable measures of market performance, compared with other sectors of market
economies, and more efficiently than they did under regulation.
There is, in fact, a strong body of evidence that regulatory reform has
substantially enhanced economic efficiency in the United States, the United
Kingdom, and many other inter-city transport markets, and that the benefits of
that efficiency have been passed on to consumers, as we shall summarize below.
The evidence on more limited experiments in deregulating markets in urban
transport is not so clear, but is still encouraging.

BUTTON 9781786435668 PRINT.indd 546 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 547

100
Passenger
Cargo
95
Index (1991 = 100)

90

85

80

75

70
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Year

Figure 14.7 A
 ir fares and cargo rates for European airlines pre and post the enactment of
the Third Package

Although almost all studies trying to analyse the effects of regulatory liberaliza-
tion for inter-city transport in the United States and the United Kingdom find
positive benefits from change, there is nevertheless some controversy around the
proper methods for evaluating the costs of excessive regulation. Traditionally,
regulation is seen to affect efficiency by imposing static dead-weight loss, encour-
aging excess capacity, and stifling technological change. The implications of
regulation, however, extend beyond these considerations. For example, regulatory
change can have a dynamic effect on productivity change, rather than just a one-
shot effect on static efficiency. Further, regulatory change will likely redistribute
income across different groups and industries, and affect geographical areas dif-
ferently, as well as changing national productivity.
As regards dynamic shifts in productivity, there have been far fewer studies
than of static efficiency. Nevertheless, some studies have been done, including for
airlines and for American trucking. In both contexts the studies concluded that
deregulation caused accelerated improvement of productivity, at least for a period
(Button and Keeler, 1993).
Concerning redistribution, studies clearly indicate that the effects of dereg-
ulation in this area are substantial. Early work in the United States, for
example, indicated that deregulation of United States aviation benefitted the
user by some $6 billion per year (1977 prices) with profits rising to $2.5 billion.
Subsequent mergers and structural changes altered these figures, but not signifi-
cantly (Morrison and Winston, 1995). In the area of freight, American deregula-
tion seems to have had powerful effects on efficiency and distribution. Clifford
Winston et al. (1990) found $20 billion (1988 prices) in annual benefit to shippers
and their customers. Yet investors in trucking are estimated to have lost over $5
billion per annum, with losses of similar magnitude to labor employed in the
industry. On the other hand, investors in railroads appeared to have gained from

BUTTON 9781786435668 PRINT.indd 547 27/04/2022 11:07

 548 TRANSPORT ECONOMICS, 4TH EDITION

deregulation. In some cases, such as with the United Kingdom’s buses and United
States railroads and airlines, there were, however, overall short-term declines in
wages from deregulation.
While the privatization of transport supplying industries has been somewhat
more recent than many other measures of liberalization, the evidence seems to be
that it has enhanced efficiency. In the case of air transport, simple comparisons of
productivity between the privately owned major United States carriers in 1987 and
the mainly state-owned European carriers indicated the former to be significantly
more efficient (Ng and Seabright, 2001). The privatization of British Airways
unquestionably improved the company’s productivity vis-à-vis its nationally owned
European rivals. Equally, the privatized United Kingdom bus companies increased
productivity. The application of more rigorous analysis to those transport activi-
ties still under public ownership tends to indicate that levels of subsidy and inter-
vention in management decision-making leads to technical inefficiency.
In nearly all cases of liberalized regulation, there were concerns on the part
of some observers that regulatory change would come at a cost to society in
terms of safety and externalities, because the need of firms to minimize costs
under a competitive market situation would require them to pay less attention
to these considerations. Although it is always difficult to distinguish between
long-run trends and short-run fluctuations in transport safety, considerable time
has nevertheless now passed since regulatory reform, especially in the United
States. Virtually all evidence regarding deregulation and safety in the United
States goes in one direction: there has been little change, involving some small
improvement.
In the case of airlines, where perhaps the greatest concern of many was
safety, not only did deregulation not increase fatalities per passenger-mile, but it
saw a continuation of increased safety that was the pattern prior to the Airline
Deregulation Act of 1978. Of course, to the extent that deregulation induced
inter-modal shifts in traffic, the effects could be complicated. But even here, the
most evident effect is positive: low airline fares have induced passengers away
from a less-safe mode such as the private car to the safer mode of air transport.
The effects of transport deregulation on the wider environment are, at best,
ambiguous. Greater freedom of entry and exit has removed much excess capac-
ity from markets everywhere; this goes for trucking, airlines, and railroads, in the
European Union and the United States. Fuller planes, lorries, trains, and railway
track should be positive in their effects on the environment. Further, to the extent
that freedom of the railroads to compete in the United States takes traffic from
roads, that could be thought to be an environmental improvement as well. The
major polluter in most contexts is the private car, and little if anything has been
done to liberalize the market for car use.
What the phase of liberalization revealed is that, although less regulation
may be beneficial, there is still a need for government intervention. A degree of
‘re-regulation’, or modifications to the liberalized structures, has been considered
by some, and adopted by a few (Bettini and Oliveira, 2008). In practice, however,
the process is one of tidying up, tweaking the various regimes that are in place,

BUTTON 9781786435668 PRINT.indd 548 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 549

and adjusting to new circumstances. Specifically, it is concerned with removing

serious market imperfections.
With a few notable exceptions, the liberalization of transport in the twenty-
first century has tended to focus on operations. Transport infrastructure invest-
ment has in contrast continued to be controlled by, and in many cases owned by,
government. How to achieve optimal utilization of this infrastructure and, specif-
ically, the development of appropriate user-charging mechanisms that minimize
distortions in liberalized markets has attracted considerable attention. At one
level there are matters relating to investment coordination and policy appraisal,
especially in the international market, which have attracted little attention from
economists.
At another level, there is the matter of efficient use and financing. The sepa-
ration of infrastructure from operations, as has been the tradition with road and
air transport and is being considered by more countries in the context of railways,
is seen as one approach to achieving greater efficiency. In North America, recent
history has seen monopoly passenger services (such as VIA Rail in Canada as
well as Amtrak) hiring track capacity from freight companies. An alternative
approach, however, is simply to offer available capacity to the highest bidder
along the lines suggested by Harold Demsetz, although this can raise a variety of
technical problems. Sweden, nevertheless, went some way in this direction in the
early 1980s, and so did the United Kingdom.
An alternative is to develop economic charging regimes. Important in this is
the question of cost attribution. In terms of rail transport, for example, consider-
able advances began to be seen in cost allocation in the United Kingdom during
the 1980s. Equally, at airports, our improved understanding now provides a more
rational basis for determining landing fees. With respect to road track cost allo-
cation, studies in the United States (Small et al., 1989) and the United Kingdom
(Newbery, 1988) provide a similar foundation for more rational user-charging
regimes. Indeed, in New Zealand, the Road User Charges system is on a broad
user-costs basis. The foundation is, therefore, now laid for more rational charging
for infrastructure use.
Anti-trust and mergers have posed serious challenges for regulators both
at the domestic and the international level, and have brought forth a signifi-
cant body of analysis. Where mergers have not been prohibited entirely, ex post
assessment has often within legal–economic work deployed structure, conduct,
and performance measures. This involves analysing the broad structure of the
industry (for example, the degree to which it is inherently monopolistic), the ways
in which firms behave within it (for example, in terms of fixing prices), and the
performance of the industry (for example, in terms of whether consumers benefit
from the merger). This is inevitably a rather subjective exercise, but it does allow
a multi-dimensional approach to the subject.
Econometric and programming approaches of the type set out in Chapter 5
have been deployed in academic work. Table 14.6 offers a brief summary of some
of the analysis on airline alliances under which carriers coordinate some of their
activities. This provides a more quantitative assessment of the implications of

BUTTON 9781786435668 PRINT.indd 549 27/04/2022 11:07

 550 TRANSPORT ECONOMICS, 4TH EDITION

Table 14.6 Studies of the effects of strategic airline alliance

Study Alliances Period Findings

Gellman Research BA/US Air, 1994 Profits increased for all parties with
Associates KLM/NW BA and KLM gaining more than
(1994) their partners
Youssef & Hansen Swissair & SAS 1989–91 Increases in flight frequency;
(1994) variations in fare levels; the
strongest service levels had the
lowest fare increases
US General KLM/NW, 1994 All carriers enjoyed increased
Accounting USAir/BA, revenues and traffic gained at
Office (1995) UAL/ competitors’ expense, not industry
Lufthansa, growth
UAL/Ansett,
UAL/BMA
Dresner et al. Continental/ 1987–91 Mixed successes with traffic
(1995) SAS, Delta volumes; in general alliances did
Swissair, KLM/ not benefit partners
NW
Park (1997) KLM/NW, 1990–94 Traffic increases at the expense of
Delta/ Swissair/ rivals; complementary alliances
Sabena lowered fares while parallel
alliances increased fares
Oum et al. (2000) Star Alliance, 1992–94 Increased traffic on alliance routes
oneWorld
Skyteam, KLM/
NW
Brueckner & US 1999 Fares are 18–20% lower on
Whalen (2000) international international alliance, inter-lining
alliances routes

Source: Button (2009). This paper contains the full references to the studies cited.

collusion, although it inevitably misses out some of the less tangible or quantifi-
able elements of supplying, in this case, air transport services.
In summary, the period since the 1970s has seen considerable change in the
way that economic regulations of airlines are viewed and how they are applied,
and with this have come innumerable studies analysing what has happened. Given
the differences in the timing of changes, the sectors affected, and the nature of
the reforms, these studies have often sought to discover the extent to which the
experiences in one country, or relating to one mode, are transferable to other
counties or modes. For example, the early moves towards lighter regulation in the
United States was an opportunity for other countries to learn lessons (Button and
Swann, 1992).
The degree to which there has been, or can be, a successful learning experi-
ence is not always, however, clear. While it is interesting to speculate over whether,
for example, the European Union approach (with its emphasis on gradualism)
was optimal, in practice it is very difficult to reach any firm conclusions. Just

BUTTON 9781786435668 PRINT.indd 550 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 551

taking air transport as an example, there was and remain significant differences
between the American and European markets (Table 14.7). These differences in
the geography, history, structure, and use of the network can make it hard to
discern what lessons are relevant, and to decide the extent to which transfers have
been successful.

14.7 Disruptive Innovation

As with all sectors of the economy, new technologies and organization practices
are continually emerging in transport. Many of these changes act to comple-
ment the existing transport services being offered. Intelligent transportation
systems, for example, generally complement existing traffic management prac-
tices and can often be integrated with existing traffic management technologies.
Others disrupt the status quo by rapidly taking up a significant market share and
replacing large parts of the market’s existing business practices. These interlop-
ers are often described as disruptive technologies or innovations. One can think
in terms of Henry Ford’s introduction of assembly line techniques of mass
production together with a standard, simple-to-use product (the Model T) as an
example.
A disruptor may target low-end customers who do not need the full perfor-
mance valued by customers at the high end of the market, or be a new-market
disruptor which targets customers who have needs that were previously unserved
by existing incumbents.
When confronted by disruption, incumbent suppliers often call for regula-
tion to protect their position, or to slow down the penetration of the innovative
intruder. Transaction costs arguments are often used. Uber, for example, has
often appeared in the media, labeled a disruption innovation because of its effects
on existing modes of transportation, and most notably on conventional taxi-cabs
and public transit. Local transit and taxi-cabs have sought protection by support-
ing the regulation of Uber, or simply banning Uber from operating. In the 1930s,
railroads in many countries called for protection from trucking. In the United
States, the 1935 Motor Carrier Act was partly intended to do this.
Offering a precise definition of disruptive innovation is not easy. Christensen
et al. (2015) are very clear about what disruptive innovation involves, and
‘[i]n our experience, too many people who speak of “disruption” have not read a
serious book or article on the subject’. They see Uber and other transportation
network companies (TNCs) as a special case within the context of the taxi-cab
industry, and more of a genuine disruptor to the limousine business with its
Uber-SELECT product. This is because disruptive competition involves entry
by a small company (which Uber was) into a market where there are established
incumbents (which there were) who were focusing on providing products for their
most demanding and profitable customers. The disruptive entrant would target
neglected customers by offering more suitable functionality and usually lower
prices.

BUTTON 9781786435668 PRINT.indd 551 27/04/2022 11:07

 552 TRANSPORT ECONOMICS, 4TH EDITION

Table 14.7 
Differences between the United States and European air transport markets
affecting deregulation

• Domestic/international traffic split. Most European carriage and historically its airlines
have largely operated in regulated international markets, and the natural inclination
for rent-protecting may take longer to erode than occurred in the United States.
• The non-scheduled market. European aviation traditionally involved a large non-
schedule, component. In 1999, for example, the revenue from passengers by non-
scheduled services in the United States was $12 billion out of $772 billion –
comparable figures for Europe were $120 billion out of $337 billion. The European
charter market has declined significantly since then, mirroring the decline in all-
inclusive holidays and the rise of scheduled low-cost carriers such as Ryanair, EasyJet
and Norwegian.
• Market size. The average route length in Europe is 720 kilometers; in the United
States it is 1,220 kilometers. The average number of passengers per scheduled route is
about 100,000 in Europe but virtually double that in the United States.
• Airline size. The scale of the European market is also reflected in the size of the
European Union’s airlines. The traditional carriers, such as Lufthansa (with assets of
€42.7 billion in 2019), Air France/KLM (€30.7 billion), and British Airways (€24.4
billion) are some of the largest in the world. The largest passenger airlines and cargo
carriers as measured by conventional parameters are American-based.
• Ownership of airlines. While the United States commercial airline industry has always
been in private hands, most major European carriers have traditionally been state-
owned and several carriers remain so or have a significant state involvement even after
measures of privatization in many countries.
• Approaches to bankruptcy. Chapter 11 arrangements in the United States are less
stringent than most European bankruptcy regimes and allow for existing
management, under supervision, to restructure a company’s finances rather than have
the assets of the undertaking realized.
• Subsidies. America had no tradition of explicitly subsidizing airlines until 2001, except
in limited cases of social air services. This situation changed during the Covid-
pandemic with short-term assistance provided. There has been a tradition of
subsidizing European flag carriers, but Union policy on state subsidies has now been
tightened and criteria made explicit under which subsidies may be given.
• Inter-modal competition. There is substantial inter-modal competition in Europe,
ranging from high-speed train services to good-quality highways. Anything
approaching high-speed rails services are extremely limited in the United States.
• Infrastructure availability. United States air traffic control is centralized, whereas air
traffic control in Europe is a national concern. The EU system comprises a patchwork,
while the United States has less than a third the number of en route facilities and
standardized mainframe computers. Both markets are moving to satellite control
systems – NextGen in the United States and Single European Sky in Europe. Airport
capacity in Europe is rapidly being reached at many major terminals, whereas this is a
more limited problem in America. More than 50 percent of traffic in Europe passes
through 24 airports and, although figures are somewhat subjective on the issue of
airport capacity, all of these have reached their technical capacity or are very close to it.
• Advantage of hindsight. European policy-makers and airlines have benefitted from the
United States’ experiences when liberalizing air transport markets, and avoided some
of the transition problems.

BUTTON 9781786435668 PRINT.indd 552 27/04/2022 11:07

 553
POLITICAL ECONOMY AND TRANSPORT REGULATION ­

Performance
Disruptive innovation

Most demanding use

High-quality use

Medium-quality use

Low-quality use

0
Time

Figure 14.8 Penetration of a disruptive innovation over time

Over time, the disrupting innovator takes over increasing amounts of the tradi-
tional suppliers’ markets. In the example depicted in Figure 14.8, the new disrup-
tive entrant comes in at the lower quality of service which is not being provided
by existing transport suppliers, and then begins taking increasing shares of the
higher-priced, better-quality markets. In practice, a disruptor can enter at any
level of performance that incumbent suppliers are not serving well.
Response from incumbents is initially lax because they are focusing on differ-
ent markets, and this allows the newcomer to move up the market, taking main-
stream business from the incumbents (which Uber, Lyft, and other TNCs have).
However, Uber’s ability to take business from taxi-cabs was more to do with
regulations governing price and market entry over the latter’s activities, shack-
ling their ability and motivation to innovate, than with being part of consciously
moving into more profitable markets. Uber was initially a rent-seeker circum-
venting regulation over very similar products, and in particular over taxi drivers.
Further, Uber’s approach, once in the market, unlike a disrupter, has been, as we
argue below, that of a sustaining innovator seeking to expand markets and to
improve on traditional taxi services.
Another transport market that has claimed to have experienced disruptive
innovation involves the arrival of low-cost air carriers. The argument is that small
airlines, such as Southwest in the United States, and later Ryanair and EasyJet in
Europe, filled a basic, low-fare, one-class market neglected by full service, incum-
bent carriers. These carriers subsequently moved successfully into more network-
based services, offering various ‘bundles’ of services – the so-called ‘upmarket
march’. Raynor (2011) puts emphasis on the low-cost airlines’ use of standard
aircraft (for example, the Boeing 737 series in the case of Southwest) in neglected
markets, combined with continually updating the series model allowing penetra-
tion of mainstream markets.
But, again, while some of the features of disruptive innovation are present,
the incumbent airlines were limited by regulations in virtually all markets from

BUTTON 9781786435668 PRINT.indd 553 27/04/2022 11:07

 554 TRANSPORT ECONOMICS, 4TH EDITION

price discrimination and the networks they could develop. Charter services did
offer lower fares but were tied by regulations that in some cases required the
combined purchase of an airline ticket and a hotel room, or passengers being
members of a ‘club’ of some kind. Disruptive innovation, as with the TNCs,
was of a heavily distorted market and not one where there was a great deal of
competition.

14.8 Coordination Via the Market, or by Direction?

Several very important facts emerge from the previous sections. First, government
in virtually all countries finds the transport sector extremely difficult to handle;
this is perhaps most clearly seen if one reflects on the fact that major pieces of
transport legislation appeared on the United Kingdom’s statute books every
seven or eight years during one half-century period (that is, 1930, 1933, 1947,
1953, 1962, 1968, 1974, 1980, 1985, 2000, 2008, and 2020). These have been sup-
plemented by numerous pieces of minor legislation. There have been substantial
shifts both in the way that transport has been viewed and in the type of policy
approach pursued.
Second, the changes in policy have exhibited systematic swings between
attempts at making greater use of market forces and a more comprehensive level
of central control or direction. Intervention has always existed, but the degree and
nature of this intervention has differed according to whether confidence was felt
in market processes. Recent changes in policy, from the late 1970s, have reflected
a greater market orientation while the nationalization of the immediate post-war
period followed philosophies of direction. The respective merits of market-
orientated policies vis-à-vis ones of planned allocation of resources are subjects
of considerable debate. While much of the discussion is political (often doctrinal)
there is, nevertheless, an important area of economic controversy involved.
It is quite possible that in theory a variety of approaches with quite sig-
nificant differences in their degree of market orientation could achieve the basic
objectives of transport policy, but the standard economic problem of integra-
tion or ‘coordination’ remains difficult in practice. (Coordination is here used in
its economic context and is best defined by G.S. Peterson, 1930: ‘Coordination
is the assignment by whatever means of each facility to those transport tasks
which it can perform better than other facilities, under conditions which will
ensure its fullest development in the place so found’.) In practice, reliance is never
placed entirely on the market mechanism to achieve the desired coordination, or
upon direction. The issue is, rather, one of degree and emphasis. It seems useful,
however, to look initially at some of the arguments that are advanced for adop-
tion of the extreme position.
The price mechanism is the main instrument of the market and offers an
obvious method of coordination, each consumer being able to purchase trans-
port services at the lowest cost. Arguments that transport is a public utility, on
a par with street lighting or the police force, are dismissed by advocates of this

BUTTON 9781786435668 PRINT.indd 554 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 555

school, who point out that transport, unlike genuine utilities, would be provided
even if government did not exist.
There may be cases (for example, quasi-public goods) where government
must act as the supplier, if provision is to be optimal, but providing the rules
of marginal cost pricing are pursued, no difficulties arise. If externalities exist,
then these can be handled adequately by means of ensuring that property rights
are allocated according to accepted rules, while lump-sum transfers could tackle
social difficulties associated with hardship. (At the extreme, one may argue that
income differences are themselves the result of market forces and should, there-
fore, be left.)
Insurance markets would handle the longevity of transport infrastructure,
and the associated risks. Safety would be ensured by travelers/consignors select-
ing operators with good safety records, or alternatively they may prefer a higher
risk but at a lower charge for the trip. Perfection in the ‘safety market’ may
require government intervention to ensure that users are cognizant of the dangers
involved – in this sense information becomes the only ‘merit good’ to be provided
in the transport sector.
Advocates of market approaches to coordination point to the automatic
mechanisms involved, and to the freedom of the system from political manipu-
lation. John Hibbs (1982) suggested that the direction of resources by some
over-riding body is likely to be less efficient at coordination because ‘[t]he admin-
istrative mind is not likely to possess the qualities of imagination and flair that are
necessary if the consumer’s interest is to be served’. Cooter and Topakin (1980)
produce evidence from the Bay Area Rapid Transit System in the San Francisco
Bay Area that provides tentative confirmation of this type of bureaucratic
hypothesis; namely, that the technostructure places its interest before that of cus-
tomers. The validity of this view is, however, debatable, and it has been suggested
that the managers of any large undertaking, irrespective of the type of purpose
they are charged to pursue, will be motivated by their own self-interest – they
will attempt to maximize their power and security. Self-interest at the expense of
customer interest, therefore, seems to be a function of the scale of management
rather than of ownership or the objectives that are set.
But even if there are potential managerial problems associated with the
central direction of resources, these may well be outweighed by the possible
benefits. Strictly, planned resource allocation involves the direction of traffic
as well as factors of production employed in providing transport services. In
general, however, United Kingdom transport policy has seldom attempted to
direct traffic, leaving the consumer free to choose his/her mode, route, service, etc.
(There are exceptions such as one-way streets, barriers to trucks, etc., but these
are rather outside of the main thrust of the debate.)
The most important case, where some degree of direction was intended, con-
cerned the proposed introduction of quantity licensing into long-distance truck-
ing under the Transport Act, 1968. The broad argument here was that it was to
the consignor’s benefit to be directed to rail in certain instances because of his/her
own misperception. (‘Inertia and habit will play their part and some consignors

BUTTON 9781786435668 PRINT.indd 555 27/04/2022 11:07

 556 TRANSPORT ECONOMICS, 4TH EDITION

may not even be aware of the advantage to them of the new rail services, nor of
the true economic cost of their present arrangements’, according to early behav-
ioral economics lines of reasoning by the UK Ministry of Transport, 1977). The
difficulty with this line of argument, and possibly one of the main reasons the
system was never implemented, is that the administrators, in making their alloca-
tion, may misperceive the priorities and needs of the consignor.
On the more central issue of service provision, it has been suggested that
without directed coordination it is often impossible or prohibitively wasteful
for many of the wider goals to be fully achieved. Direct income transfers, for
example, may rectify differences in the spending power of households, but it is
often a subgroup within a household (for example, housewives) whom one is
trying to assist, and direct transfers may not reach them. Further, with so many
operators in the transport market, there are suggestions that technical coordina-
tion of services would be less efficient (for example, bus services would not act as
local distributors to trunk rail services). Private firms may not be willing to under-
take substantial capital projects because, unlike government, they cannot spread
risk adequately, even where insurance markets do exist. In more simple terms,
advocates of direction feel that it is likely to prove overall to be more efficient than
an optimally maintained market environment.
Quite clearly, the substance of these lines of argument is likely to vary
among transport sectors. It is not surprising, therefore, that in general the alloca-
tive mechanism favored for inter-urban transport (especially freight) is that of the
market and of price, while intra-urban transport tends to be subjected to consid-
erable planning and control. The widespread occurrence of externalities, the more
immediate distribution issues, the interaction of an imperfect transport market
with that of an imperfect land market, etc., make it difficult to remove the imped-
ance that exists to the efficient functioning of a pure market for urban transport.
One exception, which should be noted, to the dominance of the market
in allocating inter-urban resources is the public provision of roads. Here, as we
have seen elsewhere, there are important differences between the way road and
rail track costs are passed on to users. Comparable pricing policies are the clear
market solution, but increasingly economists have taken the argument further
and have suggested that both sets of tracks should be publicly owned and then
users should pay on an identical basis for the services rendered – a situation that
exists with parts of Sweden’s rail network and one that has been actively pursued
in the United Kingdom. Open access and economic charging for rail infrastruc-
ture is also part of the European Union’s transport policy.
Finally, it is becoming increasingly apparent that the tools of direction or
control are, in effect, so numerous, sophisticated, and subtle that the distinction
between market-oriented policy and control is rapidly ceasing to be a meaningful
one. By manipulating price, licensing, operating laws, and work conditions, the
government is in effect directing resources and ultimately influencing the traffic
patterns that evolve. The tools of policy outlined earlier are all, in effect, tools of
direction, but at the same time they may operate as tools to improve the work-
ings of the market. Would road pricing, for example, be extending the market

BUTTON 9781786435668 PRINT.indd 556 27/04/2022 11:07

 POLITICAL ECONOMY AND TRANSPORT REGULATION ­ 557

to embrace congestion or would it be a direction of traffic? In this context, the

extremes of market versus direction cease to be helpful; one moves to a rather
more basic debate concerning the desirability of goals and the relative merits of
different policy tools for achieving them.

References

Averch, H. and Johnson, L. (1962) Behavior of the firm under regulatory constraint,
American Economic Review, 52, 1053–69.
Baumol, W.J. and Willig, R.D. (1986) Contestability: developments since the book, Oxford
Economic Papers, 38, 9–36.
Baumol, W.J., Panzar, J., and Willig, J. (1982) Contestable Markets and the Theory of
Industry Structure, Harcourt Brace Jovanovich.
Bettini, H.F.A.J. and Oliveira, A.V.M. (2008) Airline capacity setting after re-regulation:
the Brazilian case in the early 2000s, Journal of Air Transport Management, 14, 289–92.
Bowman, A. (2015) An illusion of success: the consequences of British rail privatisation,
Accounting Forum, 39, 51–63.
Button, K.J. (1989) The deregulation of US interstate aviation: an assessment of causes
and consequences, Part 2, Transport Reviews, 9, 189–215.
Button, K.J. (2009) The impact of EU–US ‘Open Skies’ agreement on airline market
structures and airline networks, Journal of Air Transport and Management, 15, 59–71.
Button, K.J. and Keeler, T.E. (1993) The regulation of transport markets, Economic
Journal, 103, 1017–28.
Button, K.J. and Swann, D. (1992) Transatlantic lessons in aviation deregulation: EEC and
US experiences, Antitrust Bulletin, 37, 207–55.
Christensen, C.M., Raynor, M.E., and McDonald, R. (2015) What is disruptive innova-
tion? Harvard Business Review, December, 44–53.
Cooter, R. and Topakin, G. (1980) Political economy of a public co-operation: pricing
objectives of BART, Journal of Public Economics, 13, 299–318.
Demsetz, H. (1968) Why regulate utilities? Journal of Law and Economics, 11, 55–65.
Finn, B. and Mulley, C. (2011) Urban bus services in developing countries and countries
in transition: a framework for regulatory and institutional developments, Journal of
Public Transportation, 14, 89–107.
Gwilliam, K.M. (1979) Institutions and objectives in transport policy, Journal of Transport
Economics and Policy, 13, 11–27.
Hibbs, J. (1982) Transport without politics? Hobart Paper 95.
Keeler, T.E. (1984) Theories of regulation and the deregulation movement, Public Choice,
44, 103–45.
Meyer, J.R. and Tye, W.B. (1981) The regulatory transition, American Economic Review,
Papers and Proceedings, 71, 268–72.
Moore, T.G. (1986) US airline deregulation: its effects on passengers, capital and labor,
Journal of Law and Economics, 29, 1–28.
Morrison, S.A. (2001) Actual, adjacent, and potential competition: estimating the full
effects of Southwest Airlines, Journal of Transport Economics and Policy, 35, 239–56.
Morrison, S.A. and Winston, C. (1995) The Evolution of the Airline Industry, Brookings
Institution.
Newbery, D.M. (1988) Road user charges in Britain, Economic Journal, 90, 161–76.
Ng, C.K. and Seabright, P. (2001) Competition, privatisation and productive efficiency:
Evidence from the airline industry, Economic Journal, 111, 591–619.
Peterson, G.S. (1930) Transport Coordination: meaning and purpose, Journal of Political
Economy, 38, 660–81.

BUTTON 9781786435668 PRINT.indd 557 27/04/2022 11:07

 558 TRANSPORT ECONOMICS, 4TH EDITION

Raynor, M.E. (2011) Disruptive innovation: the Southwest Airlines case revisited, Strategy
and Leadership, 39, 31–4.
Small, K.A., Winston, C., and Evans, C.A. (1989) Road Works: A New Highway Pricing
and Investment Policy, Brookings Institution.
Spiller, P.T. (2013) Transaction cost regulation, Journal of Economic Behavior and
Organization, 89, 232–42.
Stigler, G.J. (1971) The theory of economic regulation, Bell Journal of Economics and
Management Science, 2, 3–19.
UK Department of Transport (1977) Transport policy, Cmnd 6845, HMSO.
Vickrey, W. (1961) Counter-speculation, auctions, and competitive sealed tenders, Journal
of Finance, 16, 8–37.
Winston, C., Corsi, T.M., Grimm, C.M., and Evans, C.A. (1990) The Economic Effects of
Surface Freight Deregulation, Brookings Institution.

BUTTON 9781786435668 PRINT.indd 558 27/04/2022 11:07

 Name index

Abe, M.A. 254, 278 Becken, S. 293, 318

Abukanlil, Y. 318 Beckman, M. 20
Ackerman, K.B. 372, 395 Beesley, M.E. 114, 130, 239, 278
Acutt, M.Z. 105, 130 Bekhor, S. 169
Akerlof, G. 127 Bell, M.G. 365
Ahmed, Y. 486, 513 Ben-Akiva, M. 460, 465
Aifadopoulou, G. 279 Bendtsen, P.R. 90, 130
Akerman, J. 219 Bennathan, E. 235, 278
Al-Braizat, E. 318 Berechman, J. 144, 182, 452, 465
Al-Rafayah, S. 318 Berhofen, D.M. 145
Albers, S. 47 Beuther, M. 379, 396
Albert, G. 344 Biehl, D. 477, 512
Alexander, K.J.W. 1, 2, 20 Bitzan, J.D. 178, 183
Alfredsson, M. 380, 396 Blasé, J.R. 107, 130
Allais, M. 2, 6 Blinder, A.S. 20
Allen, D.W. 66, 83, 179, 182 Bloom, D.E. 318
Alonso, W. 76 Bly, P.H. 462, 464, 465
Alperovich, G. 102, 130 Boardman, A. 411, 437
Anciaes, P.R. 214, 233 Bonnafous, A. 188–9, 233, 508, 512
Anderson, M.L. 40, 54, 470, 513 Bonnieux, F. 199
Arango Alves, J. 213, 233 Bookbinder, J.H. 182, 183
Arnott, R. 332, 363 Börjesson, M. 328, 363, 364
Arvis, J-.F. 48, 57 Bos, S. 273, 278
Aschauer, D.A. 477, 479, 512 Botham, R.W. 507, 512
Atkinson, A.B. 412, 436 Botzen, W.J.W. 197
Autor, D.A. 81, 83 Bould, M. 130
Averch, H. 523, 557 Bowen, J. 257, 278
Bowley, A.L. 17
Bai, Y. 396 Bowman, A. 161, 183, 539, 557
Baker, S.H. 154, 182 Bradford, D.F. 252, 278, 431, 436
Bakker, D. 130 Breul, M. 68, 84
Balanguer, J. 480, 512 Brewer, A.M. 372, 395, 372
Banister, D. 109, 130 Brooks, M.R. 394, 396
Barro, R.J. 475, 512 Brown, A.J. 504
Barzel, Y. 230, 232 Brown, R.H. 148, 183
Basso, L.J. 130 Brown, S. 20, 92, 130
Bates, J. 124, 130, 465 Broz, J. 396
Baum, H.J. 300, 318 Brueckner, J.K. 350, 352, 353, 364
Baumol, W.J. 6, 18, 20, 36, 54, 89, 132, 243, Buchanan, C. 442
252, 278, 289–90, 318, 376, 378, 396, Buchanan, J.M. 2, 6, 17, 20, 325
431, 437, 462, 466, 528, 557 Buhl, S. 452, 466
Baur, P. 13, 20 Bulfin, R. 365
Baxter, R.D. 470, 512 Button, K.J. 3, 7, 20, 64, 83, 102, 104, 123,
Bayliss, B.T. 139, 183 130, 144, 158, 182, 183, 184, 204, 234,
Becker, G. 391, 396 257, 262, 279, 302, 318, 324, 331, 332,

BUTTON 9781786435668 PRINT.indd 559 27/04/2022 11:07

 560 TRANSPORT ECONOMICS, 4TH EDITION

335, 342, 351, 353, 363, 364, 365, 376, de Neufville, R. 264, 179
388, 396, 402, 408, 409, 434, 437, 477, de Palma, A. 360, 363, 364
484, 512, 514, 524, 547, 557, 550 De Walle, S.V. 318
de Witte, A. 301, 318
Cairncross, F. 55, 83 Dellucci, M.A. 206, 234
Caldwell, S. 302 Demsetz, H. 7, 528, 549, 557
Cameron, G.C. 65, 83 Devanney, J.W. 492, 512
Cannell, W. 396 DeVany, A.S. 100, 130
Cantavella-Jorda, M. 480, 513 Diamond, D. 64, 83
Carmel, A. 365 Dimitriou, H.T. 424, 436
Carney, M. 436 Disraeli, B. 22
Cascajo, P. 465 Dix, M.C. 463, 466
Caves, D.W. 179, 183 Dixit, A. 241
Cervero, R. 81 Dodgson, J.S. 105, 130, 163, 183
Chen, F. 185 Dolan, P. 127, 132
Chen, J. 90, 132 Domenich, T. 97, 131, 462, 466
Chen, P. 396 Dunn, J.A. 308, 319
Chen, S. 396 Dupuit, J. 4
Cheng, D. 396 Durant, G. 359, 364
Chestnut, L. 208, 233
Chisholm, M. 64, 83 Edlin, A.S. 316, 318
Christensen, C.M. 552, 557 Edwards, S.L. 64, 83, 139, 183
Claffey, P.T. 116, 130 Einstein, A. 22
Clark, B.D. 65, 83 Ekici, A. 387, 396
Clarke, J.B. 4 El-Agraa, A.M. 54
Coase, R.H 9, 20, 283–5, 318, 324, 352, El-Sahli, Z. 144
364, 369–70, 396 Elhorst, J.P. 407, 436
Coffman, D. 396 Elofsson, A. 319, 363, 364
Cohen, M.J. 212, 233 Estrada, M. 279
Coles, O.B. 144, 184 Evans, A.W. 79, 83, 223, 233
Collins, P. 386 Evans, C.A. 184, 547, 558
Comtois, C. 387 Evans, S. 448, 466
Conlon, R.M. 253, 280
Cook, W.R. 63, 83 Fairhurst, H.M. 125, 126, 130
Cooter, R. 555, 557 Fare, R. 182, 183
Corsi, T. 547, 558 Fearnley, N. 105, 131
Costa, A. 158, 182, 183 Fiddis, C. 381, 396
Crandall, R.L. 255 Fillipinni, M. 127, 131
Crandall, R.W. 3, 6, 317, 318 Finn, B. 540, 557
Currie, G. 131 Fischer, L. 65, 83
Fisher, R.C. 479, 512
D’Este, G. 380, 396 Flügel, S. 131
Dalvi, M.Q. 116, 131 Flyvbjerg, B. 450, 451–2, 466
Daniel, J.I. 350, 352, 364 Fogel, R.W. 435, 474, 485, 513
Dargay, J.D. 107, 119, 125, 130, 318 Ford, H. 552
Davis, O.A. 248, 279 Forester, T.H. 310, 318
Davis, R.C. 394, 396 Foster, C.D. 144, 172, 183, 419, 436
Davis, S.G. 396 Fowkes, A.S. 113, 130, 131, 464, 466
Dawson, R.F.F. 117, 130 Franklin, R.S. 508, 513
De Bok, M. 85, 130 Frenking, H. 202, 234
de Goote, J. 80 Friedlaender, A.F. 179, 183
De Haan, J. 480, 513 Friedman, D.D. 258, 279
De Jong, G. 130, 371, 396 Friedman, M. 448, 466

BUTTON 9781786435668 PRINT.indd 560 27/04/2022 11:07

 NAME INDEX ­ 561

Friesz, T.L. 367, 396 Handy, S. 66, 84

Frye, H. 364 Hanly, M. 318
Fu, X. 132 Hanson, I.G. 411, 437
Fujii, E.T. 92, 131 Haralambildies, H. 356, 364
Harback, K.T. 352, 364
Galbraith, J.K. 9, 259–60, 448 Harberger, G. 174, 183
Gallen, T.S. 480, 513 Harrington, W. 234, 319
Gammoch, H. 318 Harris, C.M. 440
Garcia-Martinez, A. 466 Harrison, A.J. 97, 131
Garcia-Sierra, M. 126, 131 Hart, O. 408, 437
Gargeya, V.B. 369, 396 Hartgen, D.T. 414, 466
Garreau, J. 39, 54 Hatemi, A.-J. 480, 513
Gilbert, C.L. 105, 131 Hawkins, H.F. 185
Gillen, D.W. 132, 179, 183, 205, 234 Haynes, K. 188, 234
Giuliano, G. 119, 130 Heaver, T.D. 142, 183
Glaister, S. 91, 105, 131, 336, 364 Heggie, I.G. 118, 131, 445, 466
Golay, J. 172, 187 Hensher, D.A. 84, 117, 118, 127, 131, 132,
Goldstein, G.S. 510, 512, 513 234, 319, 344, 364, 365, 395, 396, 397,
Gómez-Ibáñez, J.A. 6, 20 466
Gong, P. 396 Hertz, S. 380, 396
Good, H. 133 Hibbs, J. 555, 557
Goodhue, R. 184 Hicks, J.R. 413, 437, 438
Goodwin, P.B. 90, 100, 106, 131, 165, 183, Hill, M. 426, 437
301, 306, 318, 336, 364 Hillman, M. 111–12
Gordon, P. 451, 466 Hills, G.J. 185
Graham, D.J. 91, 131, 335, 364 Hine, J. 130
Gramlich, E.M. 479, 513 Hirschleifer, J. 269, 279
Green, J. 396 Hirschman, A.O. 260–61, 486, 513
Greenhutt, M.L. 63, 83 Holland, S.P. 216, 234
Gregrerson, F.A. 131 Holm, M. 451, 466
Grimm, C. 547, 558 Hooper, P. 184
Grimme, W. 319 Hotelling, H. 172, 183
Gronau, R. 92, 131 Hough, J. 41, 54
Grosskopf, S. 183 Howell, D.R. 396
Gruenspecht, A.K. 318 Howrey, E.P. 462, 466
Gu, Z. 329, 364 Hubacek, K. 396
Guan, D. 329, 364 Hue, J. 396
Guasch, J. 411, 437 Hunter, H. 470, 513
Gudgin, G. 64, 84 Huskinsson, W. 315
Gunasekaran, A. 397
Gundlach, G.T. 241, 480, 513 Ippolito, R.A. 100, 131
Gunn, H.F. 121, 131 Isaac, J.K. 107, 131
Gwilliam, K.M. 6, 20, 101, 131, 396, 535,
557 Jackson, C.C. 319
Jacobs, J. 38, 54
Hadley, T. 157, 183 Jacobson, M.R. 212, 234
Hägerstrand, T. 463, 466 Jadaan, K. 318
Hahn, R.W. 287 Jalilian, H. 105, 131
Haig, R.M. 74 Johnson, K. 204, 234
Hall, J.V. 302 Johnson, L. 523, 557
Hallegatte, S. 396 Johnston, K. 139, 183
Hamilton, C.J. 363, 364 Jones, P.M. 233, 329, 364, 464, 466
Hancock, K.L. 384, 396 Jones-Lee, M.W. 210, 211, 234

BUTTON 9781786435668 PRINT.indd 561 27/04/2022 11:07

 562 TRANSPORT ECONOMICS, 4TH EDITION

Jordan, W. 7, 527 Lehe, L. 327, 364

Joy, S. 144, 183, 430, 437 Leibenstein, H. 127, 174, 184
Juan, E.R. 276, 279 Leinbach, T.R. 257, 279
Jung, J.M. 92, 131 Leitham, S. 512
Jung, M. 319 Leeper, E.M. 482, 513
Leonard, D.P. 213
Kahneman, D. 127 Lerman, S. 460, 465
Kain, J.F. 58–9, 84 Levin, L. 92, 132
Kaldor, N. 413, 414, 437, 438 Lewis, D. 105, 131
Karlaftis, M. 148, 183 Lewis, M.S. 92, 132
Kasarda, J.D. 370, 396 Li, S. 396
Keeler, T.E. 142, 178, 183, 184, 317, 319, Li, Z. 127, 132, 319, 364
527, 537, 547, 551 Liang, X. 396
Kemp, M.A. 267, 300, 319 Lin, H. 365
Kennedy, C. 250 Lindbolm, C.E. 412, 437
Kenworthy, J.R. 271, 279 Lindsey, R. 363
Keynes, J.M. 1 Linneker, B.J. 507, 513
Kilani, M. 364 Lipsey, R.G. 247, 279, 364
Killi, M. 131 Little, I.M.D. 487, 513
Kindleberger, C.P. 469, 513 Littman, T. 89, 132
King, J. 344, 364 Liu, D. 155, 179, 182
Kleit, A.N. 318 Löbmann, R. 319
Kneller, L. 144 Loomes, G. 210, 234
Knight, F.H. 391, 396 Lösch, A. 63, 70
Koopsman, T. 18 Lu, X. 396
Korattyswaroopam, N. 54
Koshal, R.K. 144 Ma, X. 141, 185
Koster, H.R.A. 80 Maaskant, R.H. 499, 513
Kraft, G. 97, 131 MacDonald, J. 396
Kreutzerberger, F. 379, 396 Macharis, C. 318
Kroes, E.P. 463, 466 Machnes, Y. 102, 130
Krüger, H.-P. 319 Mackay, B. 293, 318
Krugman, P. 18, 20, 54 Mackie, P.J. 101, 131, 279, 402, 437
Kuhn, N. 309, 318 Mackinder, I. 448, 466
Kumar, N. 131 Madden, J.F. 82, 84
Kwon, C. 367, 396 Mahalel, D. 343, 364
Maler, K.G. 296, 319
Laffont, J.-J. 411, 437 Mankiw, N.G. 18, 20
Lago, A.M. 107, 131 Marks, J.V. 184
Lall, S. 5, 84 Marks, P. 131
Lancaster, K. 6, 247, 279, 461, 466 Marshall, A. 11, 17
Lancioni, R. 64, 84 Marteau, J.F. 48, 54
Langer, A. 226, 235, 359, 365 Martin, D.J. 213, 245, 382, 396
Lannoy, P. 318 Mansur, E.T. 234
Larcon, S. 128 Mathijssen, M.P.M. 316, 319
Larsson, J. 294, 319 Mattson, J. 41, 54
LaTourrette, T. 390, 396 Mayer, C. 350, 352, 364
Laurits, R. 183 Mayworm, P.D. 131
Lave, C.A. 90, 131, 317, 319 McCarthy, P. 148, 183, 215, 317, 319
Lave, L.B. 317, 319 McCubbin, D.R. 206, 234
Lay, M.G. 439, 466 McDonald, R. 557
Leape, J. 328, 341, 364 McDougall, G. 402, 436
Lee, N. 116, 131, 144, 179, 184 McEnroe, J. 131

BUTTON 9781786435668 PRINT.indd 562 27/04/2022 11:07

 563
NAME INDEX ­

McFadden, D. 4, 449, 462, 466 Newman, P.G. 271, 274

McGinnis, M.A. 107, 132 Newell, R.G. 283
McGuire, C.B. 20 Ng, C.K. 547, 557
McIntosh, P.T. 90, 132, 170, 184 Ngai, E.W.T. 397
McKinnon, A.C. 382, 396 Ngoe, N. 130
McLean, M. 368 Nijkamp, P. 365
McNally, M.G. 463, 466 Nilsson, J.-E. 247, 279
McNown, R.F. 318 Noble, S.E. 132
McQuaid, R.W. 512 Noland, R.B. 363, 365
Meixell, M.J. 369, 396 Nordhaus, D. 18
Metcalfe, R. 127, 132 Nythi, M. 154, 184
Meurs, H. 130
Mew, K. 436 O’Connor, A.M. 12, 20, 109, 183
Meyer, J.R. 5, 6, 20, 163, 184, 542 O’Donnell, K. 144
Micaels, G. 471 Oates, W.E. 289–90, 318
Millard, R.S. 489, 513 Oort, C.J. 499, 513
Milne, D. 4, 20 Oosterhaven, J. 407, 436
Miraller-Guash, C. 131 Ortiz, C. 396
Mirrlees, J.A. 6, 487, 513 Orznent, J. 139, 183
Mishan, E.J. 248, 279, 415, 424, 437 Oum, T.H. 90, 96, 97, 101, 130, 132, 182,
Mittal, N. 54 183, 184
Mitchell, A. 382, 397 Owen, W. 57, 84
Mogridge, M.J.H. 90, 102, 132
Mohring, H. 76, 84, 164, 184, 432, 437 Pagonis, W. 368
Mokhtarian, P.L. 113, 132, 313, 319 Pan, Q. 78, 84
Moore, T.G. 528, 557 Pan, X. 141, 185
Morlok, B.K. 270, 279 Panzar, J.D. 528, 557
Morrison, J.C. 480, 513 Papadopoulos, T. 387
Morrison, S.A. 331, 351, 352, 364, 365, Pareto, W. 413
528, 547, 557 Parker, P. 423
Moses, L.N. 6, 8, 84, 510, 512, 513 Parry, I.W.H. 252, 234, 306, 319
Mosher, D.E. 396 Paulley, N.J. 462, 466
Mueller, J. 390 Pearce, D.W. 201, 234
Mulley, C. 540, 557 Pearman, A.D. 130, 376
Munby, D. 15, 20 Peck, M.J. 4, 20
Mural, Y. 92, 97, 132 Peer, E. 129, 132
Musgrave, R.A. 251, 279 Pegrum, D.F. 5
Mutti, J. 92, 97, 132 Pels, E. 349, 365
Myrdal, G. 476 Peltzman, S. 7, 317, 319
Peng, Z.-R. 54
Namri, N.K. 363, 365 Perl, A. 308, 319
Narayan, P.K. 480, 513 Peterson, G.S. 544
Nash, C.A. 131, 149, 160, 183, 184, 272, Phifer, S.P. 464, 466
279 Pickrell, D.H. 449, 466
Nelson, J.D. 512 Pigou, A. 4, 288, 319, 324
Nelson, J.P. 203, 204, 234 Pirog, S.R. 64, 84
Nesbit, T.M. 317, 319 Pirrong, S.C. 276, 279
Neto Paiva, F. 233 Pitfield, D.E. 184
Neuberger, H.L.I. 169, 184, 419, 436 Plotch, P.M. 434, 437
Neutze, G.M. 223, 234 Polain, C. 318
Neven, A.J. 464, 466 Poole, E.C. 163, 184
Newbery, D.M. 226, 227, 234, 549, Porter, M.E. 14, 20
557 Poschmann, F. 422, 437

BUTTON 9781786435668 PRINT.indd 563 27/04/2022 11:07

 564 TRANSPORT ECONOMICS, 4TH EDITION

Pratt, J.B. 154, 182 Sass, T.R. 317, 319

Premiti, A. 492, 513 Schafer, A. 103, 132
Prest, A.R. 419, 423, 437 Schank, J.L. 270, 279
Preston, J. 419, 437, 464, 466 Scheelhaase, J. 294, 319
Prettner, K. 318 Scholvi, S. 68, 84
Proost, S. 232, 234 Schwing, R.C. 291, 292, 319
Prud’homme, R. 477, 513 Seabright, P. 548, 557
Pucher, J. 47, 54 Segal, M. 82, 84
Selod, H. 386, 397
Qu, W.W. 183 Sharp, C.H. 291, 319, 335, 365
Quandt, R.E. 6, 89, 132, 462, 466 Sheldon, P.J. 463, 466
Quarmby, D.A. 97, 116, 131, 132, 171, Sherman, R. 97, 132
184 Shiftan, Y. 169
Quddus, M.A. 335, 365 Shiller, R.J. 3, 20
Quinet, E. 232, 235 Shires, J. 279
Quixiu, C. 329, 364 Shoup, D. 323
Shreiber, C. 239, 279
Raballand, G. 48, 54 Shultz, G.P. 82, 84
Rainelle, P. 199 Sickles, R.C. 183
Rakowski, J.P. 5, 6, 8, 20 Silva, L.T. 233
Ramsey, F. 252, 279 Singh, K. 43, 184
Rassenti, S.J. 353, 365 Simon, D. 402, 437
Rauch, F. 128 Simon, H.A. 127, 424, 437
Raynor, M.E. 552, 553, 557 Sinai, T. 350, 352, 364
Redmond, L.S. 113, 132 Singell, L.D. 318
Reaves, D. 304 Skitovsky, T. 414–15, 437
Rees, A. 82, 84 Slack, B. 397
Remoaldo, P. 233 Small, K.A. 142, 159, 160, 184, 336, 365,
Revilla Diez, J. 68, 84 549, 577
Reza, A.M. 102, 132 Smith, A. 237, 399
Ricci, M. 312 Smith, D. 109
Rietveld, P. 365 Smith, L.V. 200, 235, 353
Rigakos, G. 395 Smith, M.G. 90, 132
Rindt, C.R. 463, 465 Smith, V.K. 2, 198, 235
Roberts, M. 130 Sobel, R. 317, 319
Robertson, G.E. 66, 84 Solow, R.W. 20, 472, 513
Rodrigue, J.-P. 382, 397 Sonmahoro, S. 386, 397
Rogers, K. 287 Sorensen, P.A. 322, 365
Röller, D.H. 183 Southwirth, B.W. 319
Rostow, W.W. 470, 513 Spence, N.A. 64, 83, 507, 513
Rothenberg, J.G. 188, 235 Spiller, P.T. 519, 558
Rowe, R. 396 Spiro, H.M. 102, 132
Rowney, M.J. 397 Srinivasan, S. 131
Rudloffe, C. 464, 465 Stafford, H.A. 66, 84
Ruijgrok, C.J. 70, 84 Stanason, J. 20
Rundshagen, V. 47 Starkie, D.N.M. 278, 279, 430, 437
Steedman, I.W. 144, 178, 184
Saberi, M. 229, 364 Stern, N. 207, 235
Salanovaa, J.M. 242, 279 Sterner, T. 319
Sala-i-Martin, X. 475, 512 Stevenson, R.L. 55, 84
Salomon, I. 313, 319 Stewart, M.G. 390
Samsom, T. 251, 279 Stigler, G.J. 7, 261, 279, 523, 558
Samuelson, P.A. 18, 20, 414, 437 Stopford, M. 140, 184

BUTTON 9781786435668 PRINT.indd 564 27/04/2022 11:07

 NAME INDEX ­ 565

Stopher, P.R. 445, 466 Vary, D. 227, 235

Stough, R. 54 Vega, H. 257, 279, 332, 364
Straszheim, M.R. 97, 132 Vegel, T. 130
Straub, M. 465, 466 Verhoef, E.T. 223, 235, 336, 337, 342, 349,
Straub, S. 411, 437 365
Streenberghen, T. 318 Vickerman, R.W. 485, 513
Stromquist, U. 470, 513 Vickrey, W. 26, 227, 235, 325, 529, 558
Sturm, J.E. 480, 513 Viner, J. 5
Sturmey, S.G. 241, 279 Vining, A. 411, 437
Sullivan, D. 396 Vinod, H.D. 376, 378, 396, 462, 465
Sunstein, C.R. 127, 132 Vollrath, M. 316, 319
Swann, D. 2, 20, 515, 550, 557 von Thünen, J.H. 74
Sweet, M. 220, 235
Wabe, J.S. 144, 184
Talequani, A.R. 41, 54 Wagon, D.J. 185
Talley, W.K. 213, 235 Walker, T.B. 482, 513
Talvitie, A.P. 462, 467 Walles, H. 319
Tangney, T. 265 Wallis, I.P. 148, 184
Tanner, J.C. 121, 132, 142, 184 Walls, M. 234
Taplin, J.H.E. 102, 132 Wang, G. 386, 397
Taussig, F.W. 4, 163, 184 Wang, J. 396
Tay, R. 215, 316, 319 Walters, A.A. 142, 144, 184, 236, 324,
Taylor, B.D. 322 356, 365
Teal, R.F. 148, 184 Wardman, M. 98, 104, 116, 119, 131,
Thaler, R. 2, 127, 132 133
Thatcher, M. 324 Warne, J. 397
Thomas, S. 119, 132 Warner, L. 165
Thomson, J.M. 5, 16, 21, 55–6, 84, 222, Warren, G. 130
235 Waters, W.G. 115, 117, 132, 133, 278,
Timothy, D. 82, 84 395, 397
Tinbergen, J. 18 Watkiss, P. 279
Tolifari, S. 145, 184 Watling, D. 4, 20
Toner, J. 121 Weber, A. 60, 84
Topakin, G. 555, 557 Weber, W.E. 317, 319
Tretheway, M.W. 183 Webster, F.V. 453, 467
Treyz, G. 235 Wegener, M. 59, 84
Trice, M. 83 Wei, F. 90, 133
Truelove, P. 253, 279 Weiner, E. 440, 467
Tsai, Y-.H. 81, 83 Weisbrod, G. 227, 235
Tullock, G. 17, 20 Wells, S.J. 18, 20
Turner, M.A. 359, 364 Wesseling, B. 130
Turner, R.K. 201, 234 Wheaton, W.C. 82, 84, 427, 437
Turvey, R. 265–6, 279, 419, 423, 437 Whebell, C.F.L. 68, 84
Tye, W.B. 542, 557 Whiffen, A.C. 213
Tyson, W.J. 269, 279 Whinston, A.B. 248, 279
White, P.R. 98, 133
Van Bekker, P. 130 Wildavsky, A. 423, 437
Van Benthem, A.A. 212, 234 Williams, M. 144, 184
Van Buseck, C.R. 319 Williams, T. 128
van den Berg, J.C.J.M. 131 Williamson, H. 4
Van Dender, K. 233, 234 Williamson, O.E. 9, 21
van Es, J. 70, 84 Willig, R.D. 528, 557
van Ommergen, J. 80 Wilson, A.G. 165, 185

BUTTON 9781786435668 PRINT.indd 565 27/04/2022 11:07

 566 TRANSPORT ECONOMICS, 4TH EDITION

Wilson, G.W. 486, 513 Yang, S.-C.S. 482, 513

Wilson, W.W. 465, 467 Young, W. 342, 365
Winsten, C.B. 20 Yu, C. 182, 184, 395, 397
Winston, C. 1, 4, 21, 139, 184, 185, 226,
235, 353, 359, 365, 480, 513, 547, Zahavi, Y. 87, 102, 133
557 Zerby, J.A. 253, 280, 492, 513
Wiseman, J. 245, 280 Zhang, L. 132, 141, 185
Zhu, Y. 54
Xu, B. 396 Zimmerman, P.R. 317, 319
Xue, Q. 396 Zwick, C.J. 20

BUTTON 9781786435668 PRINT.indd 566 27/04/2022 11:07

 Subject index

accessibility 55–60, 65–6, 452 Northeastern Rail Reform Act 1981

accidents 116–27, 209–11, 303–4, 309, (US) 537
315–18, 376, 535, 554 Bus Regulatory Reform Act 1982 (US)
valuation of life 211 537
activity analysis 463–5 Transport Act 1985 (UK) 270, 539
acts (by date) Single European Act 1986 (EU) 496
Railway Act 1844 (UK) 531 Federal Highways Act 1987 (US) 316–17
Interstate Commerce Act 1887 (US) National Transportation Act 1987
531 (Canada) 110, 538
Railway Act 1903 (UK) 532 National Highway Designation Act 1995
Road Traffic Act 1930 (UK) 532 (US) 310, 471
Road and Rail Traffic Act 1933 (UK) Clean Air Act 1996 (US) 314
532 America Recovery and Investment Act
Canadian National–Canadian Pacific 2009 (US) 400
Act 1933 (Canada) 533 African Continental Free Trade Area
Motor Carriers Act 1935 (US) 552 492
Transport Act 1938 (Canada) 533 air traffic control 176, 194, 220, 345–6,
Town and Country Planning Act 1947 438, 535, 551
(UK) 441 EUROCONTROL 176, 221, 345
Transport Act 1947 (UK) 533 NATS 304, 403, 526
Road Traffic Act 1953 (UK) 109 NAV Canada 403
Transport Act 1962 (UK) 534 airlines 2, 4, 10, 34, 36, 58, 92, 108–9, 256,
Transport Act 1968 (UK) 253, 299 263, 273–4, 277, 290, 295, 310, 320,
Town and Country Planning Act 1968 348–9, 386, 398, 451, 501–2, 595,
(UK) 442, 534–5 526–7, 528–9, 552, 537
Clean Air Act Amendment 1970 (US) alliances 11, 152, 385, 502, 544, 549–50
286 charter 243, 270, 551, 554
Civil Aviation Act 1971 (UK) 535 flag carriers 248–9
Local Government Act 1972 (UK) 442, freight 26–7, 47, 49, 369, 371
535 long–haul 95, 97
Energy Highway Emergency low cost 65, 262, 264–5, 553
Consolidation Act 1974 (US) 310 regulation 11, 523–7
Railways Act 1974 (UK) 535 security 388–90
Railroad Revitalization and Regulatory subsidies landing slots 110
Reform Act 1976 (US) 537 Airport Council International 28
Air Cargo Deregulation Act 1977 (US) airports 2, 15, 16, 27, 28, 321, 399, 482–3,
537 504
Transport Act 1978 (UK) 535 congestion 344–56
Airline Deregulation Act 1978 (US) 10, delays 350–2
100 economic development 65, 482–3, 508
Energy Tax Act 1979 (US) 307 flight paths 303
Transport Act 1980 (UK) 535–6, 538 landing slots 2, 27, 47, 173, 344–6, 549
Staggers Rail Act 1980 (US) 537 low-cost terminals 264
Motor Carriers Deregulation Act 1980 ownership 344–5, 435–6
(US) 537 security 387–90

BUTTON 9781786435668 PRINT.indd 567 27/04/2022 11:07

 568 TRANSPORT ECONOMICS, 4TH EDITION

slot allocation 353–5 Commission of the European

Third London 6, 13, 196, 202–3, 303, Communities 498, 501
420–23 Commission on the Third London Airport
two-till regulation 277–8 6, 196, 303, 420–24, 498, 501
American Automobile Association 211 commuting 38, 66, 80–83, 106, 118–20,
American Community Survey 41 126, 168, 313, 323–4, 480, 498, 501
American Economic Association 3 reverse 80–82
Amoco Cadiz sinking 198–9 competition 2, 237, 348, 351, 356, 387,
Armitage Committee 65, 303 416–17, 517
Association of American Railroads for the market 352–4, 529
Association of European Airlines 345 excessive 272–6, 496–7, 517
auctions 262, 285, 353–5, 528–9 imperfect 261, 349, 351–5
automobiles 34, 47, 219, 359 non-price 491–2
carpooling 336, 363, 461 perfect 237–8, 240, 298–9, 439
forecasting 119–26, 215 potential 241, 254, 261, 528, 546
ownership 39–40, 45, 47, 58–9, 79–80, see also contestable markets
119, 121–6, 306, 399 congestion 4, 6, 16, 40, 47, 79–80, 126,
see also congestion 188–90, 221–32, 298, 309, 320–63,
Averch–Johnson effect 523–4 427–8, 479
club good 10, 186, 249, 554
big data 4, 130, 266, 366, 386–7 generic 188
behavioral economics 2, 22, 89, 126–30, speed–flow relationship 221–5, 355
463 types 227–8
habit 106–7, 166, 464, 555 consumer surplus 4, 244–6, 263, 346,
hysteresis 126, 465 404–6, 415, 419, 428–30, 545
inertia 67, 106, 126, 174, 465, 555 containers 25, 367–9, 388, 534
nudges 127 contestable markets 7, 237, 240, 261, 528,
see also satisficing; X–inefficiency 537, 546
benchmarking 176, 450 COP26 UN Climate Change Conference
benefit–cost analysis see cost–benefit 215
analysis cost–benefit analysis 5, 6, 12, 323, 324,
bid-rent curves 72, 74–7, 508, 510 341, 367, 389–90, 392, 403–6, 447
British Airports Holdings 264 cost-effectiveness analysis 393
British Railways Board 312–13 COBA 13, 119, 120, 407, 430, 423
buses 37, 42, 102, 103–4, 128, 150–51, 159, NATA 13, 420
170, 172, 212, 246, 268, 328, 341, 359, theory 412–16
360–62, 402, 510, 515, 526, 537 costs
bunching of service 170–71, 471 average 225–6, 229, 238, 242
lanes 288, 283 back-haul 155, 267–8, 383
see also subsidies common 155–7, 161–2, 172, 267–8
decreasing 249–59
canals see inland waterways fixed 16, 249–50, 259, 268, 273–4
car ownership see automobiles generalized 90, 165–7, 170, 225–6, 325,
cartels see airline alliances; shipping 358, 369, 432, 452–3, 456, 511
conferences joint 155–7, 267–8
Channel Tunnel 19, 410, 420, 485 long-run 267–9, 401–2, 527
circular-and-cumulative causation 476 marginal 97, 192, 225–6, 229, 238,
climate change 186, 196–7, 206–7, 281 242, 243–6, 249–55, 258–9, 262,
Stern review 207 268, 254–5, 285–91, 296–7, 325–6,
cliometrics 44, 486 392–3, 401, 428–30, 431–2
Cobb–Douglas function 101, 473, 477, operating 110, 119, 124, 147–8, 251, 269,
478 272, 273, 312, 327, 341, 357, 373,
collective goods 269 411, 426

BUTTON 9781786435668 PRINT.indd 568 27/04/2022 11:07

 SUBJECT INDEX ­569

opportunity 113, 135, 164–5, 203, 330, standardization 494, 495

341–2, 366, 377 see also network economies
perception 97 elasticity
short-run 268–9, 401–2, 536 demand 63, 73, 75, 89, 190–96,
stranded 27, 507, 541–3, 528–9 103–7, 239, 248–9, 243–4, 256,
sunk 16, 27, 258, 507, 528, 537, 542, 572 267
see also congestion; cost–benefit cross 67, 84, 89, 91, 103–5, 106, 248–9,
analysis, externalities 252, 269, 298–300, 344, 360–61,
Council for Mutual Economic Assistance 414, 419
45 distance 127
countervailing power 259–65 employment 35–6
Covid–19 see pandemics fuel price 90–91, 306
cycling 4, 22, 311–12, 359 generalized cost 170
Boris bikes 311–12 income 108–90
cycle lanes 15 service quality 107
see also micromobility substitution 178
supply 73
data envelopment analysis 4, 110–2, 264 trip purpose 96–7
dead-weight loss 169, 174, 226–7, 231, 300, electric vehicles 215
362, 495, 547 empty core problem 273–76
demand 19, 85–130, 239 see also excessive competition
abstract modes 461–2 energy 215–19, 304–14, 383, 521
derived 15, 224–5, 465, 510 renewable 215
Cobb–Douglas function 101, 178, 472, see also electric vehicles
477, 478 Engels’ curve 43
fluctuating 85, 266, 402 environment 2, 5, 16, 17, 19, 190–93,
modeling 4, 6, 19 281–318, 323, 383
path dependence 107 acid rain 207–8, 292
peak 39, 86, 105, 252, 267–73 carbon dioxide 206–7, 292, 307, 338,
see also forecasting 341, 383
difference-in-differences analysis 92, 128 community severance 213–14
discount rate 196–7, 204, 404–5 lead 205, 286–7, 291–2, 295, 305, 521
disruptive innovations 550–53 sustainability 2–3, 219
particulate matter 205–6
e–commerce 51–2 pollution 188–90, 205–7, 281, 338, 371,
economic gateways 67 429, 517
economic growth theory sustainability 2–3, 219, 282, 539–40
balanced growth 486, 488 vibration 213, 303
circular-and-cumulative causation 476 visual intrusion 190, 211–12
economic convergence 475–6 water contamination 212, 283
new economic growth theory 476, 488 environmental policies
economies of CAFE standards 94, 286, 294, 307–9
agglomeration 1, 56, 76, 227, 379, 481 carbon off-sets 293–4
density 352, 375, 381, 385, 409, 506 carbon sinks 283
experience 153–4 emission standards 283, 294–8
final 60, 215, 483 emissions trading 294
fleet size 139, 142, 269, 401 environmental justice 282
market presence 12, 154, 302, 352, 385, polluter–pays principle 296
388, 545 tradable permits 283, 296
scale 16, 69–70, 71, 73 valuation 193–201, 463
scope 12, 36, 42, 63, 70, 154, 156, 172, European Coal and Steel Community
256, 352, 357, 381, 385, 388, 409, 495
516, 545 European Free Trade Area 44

BUTTON 9781786435668 PRINT.indd 569 27/04/2022 11:07

 570 TRANSPORT ECONOMICS, 4TH EDITION

European Union 10, 31, 36, 44, 53, 212, expansion 358–9
235, 313, 314, 345, 353, 385, 407, 475, user charges 296
492–504, 516, 536, 550, 556 see also investment appraisal; tolls
Commission of the European inland waterways 34, 56, 187, 213, 403,
Communities 497, 501 474, 531
Court of Justice 503 institutional economics 9, 187, 231, 305,
Emissions Trading Scheme 294 356
European Single Market 31–45, 385–6, new institutional economics 9
497 insurance 127, 213, 215–16, 391, 394
EUROSTAT 2, 3, 53 intelligent transport systems 127, 320, 326,
Trans-European Networks 400, 407, 336, 552
439, 497, 499 Intergovernmental Panel on Climate
experimental economics 353, 393, 464–5 Change 207–8
externalities 10, 18, 25, 32, 187–8, 193, intermodal transport 70, 366, 379, 434,
218, 233, 342, 346, 382, 394, 402, 490–91
416, 444, 486, 517, 519, 522, 534, International Air Transport Association
548, 555 54, 254, 260, 346
benefits 11–12 International Civil Aviation Organization
costs 275–6 10, 26, 47, 54, 293–4
Coasian 285, 296, 353–4 International Maritime Organization 10,
network economies 11, 67 54
pecuniary 187–8 International Road Confederation 54
Pigouvian 10, 283, 288–92, 295, 353 International Road Transport Union 23
Marshallian 188 International trade 4, 8, 22, 23, 24–41, 45,
technological 187–8 70, 143, 366, 394, 470–71, 493, 516
spatial concentration 63 International Union of Railways 54
valuation 193–201 investment appraisal 398–438, 487–9
see also agglomeration; congestion; goal achievement matrix 426
environment multicriteria 424–6
multiplier effects 481–6
fuel efficiency 107–8, 215, 286, 294, 305–9, network effects 417–9
313, 318, 383 planning balance sheet 424–5
project impact matrix 435–6
game theory 172, 200, 236, 261, 264, rule-of-half 417, 429–30
283–7, 352, 388, 391 see also cost–benefit analysis
global distribution systems 3, 362
global warming see climate change just-in-time production 7, 14, 34, 108, 367,
globalization 2, 7, 8, 44, 368, 379–80, 400, 374, 470, 482, 517
516
Keynesian economics 1, 174, 468, 518
hedonic price index 80, 195, 203, 422
high-technology industry 65–6, 68–9, 108, land-use/transportation planning 6, 19, 58,
482–3, 516 60, 79, 98, 304, 443, 453, 465
highways see roads blue-print planning 439–40
household expenditure on transport 42–3, location theory 55–83
45, 101–2 logistics 2–4, 6, 11, 14, 18, 19, 21, 57, 70,
hub-and-spoke operations 46, 141, 143, 84, 132, 367–95, 399, 470
150–53, 263–4, 349, 353, 375, 493, consolidation 141, 372, 375–6, 378, 383,
501–2, 504 494
FedEx 369, 381
inferior good 39–40, 43, 101, 111 freight forwarding 257–8, 272, 370, 376,
infrastructure 7–8, 19, 27, 37, 45, 296, 367, 532
371, 398–438, 480, 495 green 382–3

BUTTON 9781786435668 PRINT.indd 570 27/04/2022 11:07

 SUBJECT INDEX ­ 571

international 384–6 North American Free Trade Area 31, 32,

inventories 107, 370–75 44
just-in-time production 7, 14, 34, 108,
367, 374, 470, 482, 517 Official Airline Guide 51, 347
lean 373 oligopoly 93, 263, 352, 408
outsourcing 308, 379, 387 Organisation for Economic Co-operation
reverse 383 and Development 23, 29, 53,
transshipment 207, 367–8, 375–6, 379, 399–400
381, 400, 438, 474, 497, 501 Organization of Petroleum Exporting
urban 380–82 Countries, 219
warehousing 69 polluter–pays principle 287–94
lorries see trucking
pandemics 3, 18, 35, 42, 46–7, 50–51, 86,
market areas 55, 69–3, 470, 506 209, 313, 363, 400
market instability 273–6 parking 79–80, 166, 223, 283, 320, 323,
see also empty core problem 326–8, 340, 342–4, 388, 434–5, 440,
merit goods 554 510
meta-analysis 103–4, 204–5 airport 108–9, 435
metro systems 58, 127, 128 charges 166, 283, 320, 342–4, 511
Metropolitan Standard Area 39 park-and-ride 266, 339, 344, 511
micromobility 3, 362–3 restrictions 323, 324
Bird 363 Pareto optimality 174, 334, 349, 412–13
e-bicycles 356 pipelines 34, 36, 49, 217, 372, 384, 388,
e-scooters 3, 311 433
Lime 363 policing and enforcement 158, 285, 287,
motorcycles 159, 317, 340 305, 332, 394
walking 33, 38–40, 85, 101, 116, 164–5, pollution see environment
304, 362–3, 416, 507 prices 236–78
see also cycles capped 264, 434, 436, 520, 526–7, 541
migration 8, 472–4, 508–9 computer reservation systems 256
mobility 3, 17, 39, 45, 50, 55–8, 69, 111–12, discounts 99, 211, 253, 257, 259, 352
214, 235, 237, 298, 301, 330, 456, 474, discriminatory 249–58, 352–3, 530, 531
476, 487, 504, 535 fuel 90, 93, 94, 107, 146, 152, 283, 306
monopoly 236–7, 239, 241, 244–5, 246–7, marginal cost 225–7, 239, 262, 243–4,
249, 348, 351, 357, 400, 495, 523–9 249–50, 265–6, 268, 274, 283,
anti-trust policies 346, 382, 495, 515, 288–9, 296–7, 298–300, 325–6, 341,
517, 531–2 347–9, 356–7, 358–9, 398, 401, 438
bilateral 26, 346–7 peak-load 267–8, 283, 270, 351
monopsony 262 predatory 240–41, 497, 517
motorways see roads Ramsey 52–3, 346, 351, 352
rate-of-return regulation 163, 520,
nationalization 17, 533–4, 554 523–6, 527
need 16–17, 109–12 sales revenue maximizing 67, 242
neoclassical economics 9–10, 88, 126–7, second best 246–9, 277–8, 282, 320, 321,
173, 196, 260, 387, 395, 408, 469, 336–7, 342–3
472–6, 473–5, 476, 478 willingness-to-pay 251, 342, 347
network economies 3, 11, 67, 523 see also need, road pricing; yield
point-to-point 150–51, 349 management
radial 57, 152–3, 439 principal–agent problem 97
see also hub-and-spoke operations privatization 344, 394, 403–4, 407–12, 433,
new trade theory 24 435–6, 514–5, 538, 548
noise nuisance 3, 190, 201, 231, 282, 289, producer surplus 347–8, 404–6, 429–30
290, 294, 303, 309 see also profits

BUTTON 9781786435668 PRINT.indd 571 27/04/2022 11:07

 572 TRANSPORT ECONOMICS, 4TH EDITION

profits electronic 327, 328–9, 332–3

maximization 67, 237–8, 243, 245, 283, setting of prices 332–4
170–1, 344, 348–9, 352 Smeed Report 6, 332
normal 238–9, 249 use of revenues 335–6
not-for-profits 348–9 value pricing 330
see also rent seeking variable 327, 332–4
property rights 9, 10, 218, 282, 283–5, 287, see also congestion
289, 354–6, 530, 555 roads 6, 26, 29, 33, 148–9, 316–8, 320, 346,
public choice theory 17, 425 430–31, 438, 442, 450–51, 484, 490,
public goods 187, 487, 518, 555 495, 507
public ownership see nationalization build-to-meet-demand 323
public–private partnerships 13, 19, 158, capacity 159, 221, 227, 242, 327, 335,
389, 407–12, 435–6, 514 336, 358–60
contract issues 408–11 construction 8, 16, 17, 49, 142, 157, 406,
franchises 408–10, 435, 515, 539 441, 489, 520
public transport 37, 39, 59, 103–5, 107, Federal Highway System 158, 316, 420,
299–301, 323, 327, 330–32, 415, 442, 400, 471, 476
478, 539, 540 high-occupancy vehicle lanes 313
see also buses; rural transport intersections 39
investment 13, 112, 120, 159, 337, 359,
railroads 2, 4, 17, 29, 34, 35, 47, 57, 62, 69, 406, 420, 424–5, 433, 457, 507
125, 121, 246, 249, 252–3, 290, 315, speed-flow relationship 221–5
372, 374, 399, 403, 410, 430–31, 439, track costs 6, 157, 158, 160–61, 163,
450–51, 461, 470, 474, 485, 489, 519, 556
525, 534, 545, 548 turnpikes 276, 404
Amtrak 29, 33, 136, 151, 161, 163, 217, see also road pricing, tolls
549 route assignment 459
British Rail 161–3, 253, 266, 298, 423, rural transport 6, 41, 111
433, 539
Canadian National 69–70 safety see accidents
freight 69–70, 244 sales revenue maximization 241, 243
high-speed 23, 29, 91–2, 349 satisficing 67, 243
length of haul 85–6, 258–9 seaports 19, 28–9, 236, 321, 342, 356–8,
light 78, 449 399, 408, 434, 501
passenger 33, 42, 244 congestion 356–8
track 15, 16, 144, 546, 556–63, 273, 403 security 4, 18, 35, 45, 67, 109, 236, 243,
railways see railroads 302, 369, 371, 376, 379–80, 387–95,
rapid transit 299, 361, 449, 484, 555 435, 521, 555
Bay Area Rapid Transit System 449, terrorism 335, 388, 390, 391, 393, 394,
555 396
regulatory capture 218, 393, 435 shipping 10–11, 17, 25–6, 27, 212, 218,
remote working 43, 313, 464 248, 309, 388
revealed preference methods 113–4, 195–6, alliances 11, 253, 385
198, 203, 211, 332, 371, 455, 463, 465 bulk cargo 25, 145, 253, 368–9, 492
road haulage see trucking coastal 35, 433, 474, 532
road pricing 6, 183, 230, 283, 299, 320, conferences 11, 241–3, 245, 253–4
321–44, 346, 349, 352, 354–5, 357, consortia 11, 497, 501
360, 511, 556 container 25, 27, 28, 139, 140, 142,
buy-back effect 314 143–4, 357, 368–9, 379, 384–5
collection 332–3 liner 141, 152, 241, 253, 276, 357,
cordon 227–8 390–91, 492
difficulties 329–41 registry 145–6
distribution effects 334–5, 336 subsidies 110

BUTTON 9781786435668 PRINT.indd 572 27/04/2022 11:07

 SUBJECT INDEX ­ 573

tankers 4, 25, 148, 294 travel-time savings 112–13, 115–19, 370,

tramp 152, 253 396
shopping 39, 43, 59, 83, 86, 97, 302, 388, commuting 116–18, 325–6
390–92, 394, 395, 422, 457, 480 equity values 118
malls 388, 390, 392, 393, 394, 395 generalized time costs 170
teleshopping 135, 313 non-work 97, 112, 113–14, 116–20
speed limits 194, 221, 283, 295, 304, 310, valuation 6, 89, 112–19
316–17, 333, 355 work 113
stated preference methods 116–17, trip distribution models
200–201, 211, 344, 371, 385, 445, 448, disaggregate demand models 4, 6,
463–5 100–101
subsidies 58, 248, 270–71, 298–302, 442, gravity models 449, 455–7
511–12, 520, 529, 551 intervening opportunities models 457,
cross 111, 266, 272–3, 296, 317, 532 471
public transport 6, 58, 59, 90, 271–4, trip generation 453–5, 460
298–302, 320, 323, 337, 360–62 trucking 27, 34, 35, 57, 65, 107, 134, 137,
supply chain 1, 7, 11, 10, 19, 112, 358, 144, 148, 179, 186, 212, 227, 290, 295,
366–7, 370–72, 375, 376, 381, 382–3, 297, 303, 309, 321, 359, 371, 379, 382,
386, 388, 391, 393, 395 495, 498, 517, 528, 532–4, 535, 538,
see also logistics 547–8, 552, 555
supply-side economics 7, 223 driving hours 495, 498, 519
forked tariffs 495–6
taste 106 lanes 359
taxes 66, 148, 157, 160, 216, 243, 271–7, less than truckloads 381
283, 298, 306–7, 321, 403–4, 520 rates 90
allowances 283
environmental 277, 283, 290 UK Census of Production 64
fuel 157, 250, 276–7, 306–9 UK Civil Aviation Authority 11, 264, 535,
hypothecation of revenues 336 538
indirect 339 UK Confederation of British Industry
purchase 277 266
vehicle 337 UK Department for Transport 53, 121,
taxi-cabs 4, 33, 39, 86, 149, 170, 207–8, 127, 132, 191, 423, 425
238–9, 241–2, 301–2, 340, 363, 552 UK Department of the Environment 183,
technological progress 36 433, 442
teleworking 45, 51, 57, 313–14 UK Department of Transport 97, 111,
Texas Transportation Institute 22 118, 121, 184, 380, 437, 438
tolls 158, 276, 321, 322, 329, 332, 404 UK Family Expenditure Survey 42–3, 125,
cordon 322, 327–8, 334, 339 445
total factor productivity 180, 447, 479 UK House of Commons Committee of
tourism 15, 56, 87 Public Accounts 449, 467
traffic management 303–4, 335 UK House of Commons Select
traffic calming 295, 304 Committee on Nationalised
see also intelligent transport systems Industries 162, 433
transport corridors 67–70, 441, 483, 499 UK House of Commons Trade and
Transport Research Laboratory 121, 123 Industry Subcommittee 65
transportation network companies 363 UK Ministry of Transport 424, 442, 467,
DiDi 302 556
Lyft 302, 553 UK National Air Travel Services 304
Uber 3, 301–2, 340, 363, 552 UK National Travel Survey 445
Transportation Research Board 45, 54, UK Price Commission 107–8, 184
521 UK Royal Commission on Environmental
travel budgets 87, 102 Pollution 281

BUTTON 9781786435668 PRINT.indd 573 27/04/2022 11:07

 574 TRANSPORT ECONOMICS, 4TH EDITION

UK Trade and Industry Sub-Committee US Federal Urban Mass Transportation

of the House of Commons Administration 421, 437, 445, 467,
Expenditure Committee 65 471
UK Treasury 387 US Highways Trust Fund 157
United Nations 38, 40, 49, 54 US Interstate Commerce Commission 523,
Committee on Trade and Development 531
491, 492 US National Highway Traffic Safety
United States–Mexico–Canada Agreement Administration 308–9, 315, 317
44, 492 US National Travel Survey 125, 445
urban freight movement 267–8 US Office of Management and Budget 41
see also supply-chain US Transportation Security
urban sprawl 38, 47 Administration 389–90
urban wage gradient 80–82
urbanization 37, 40, 47 value chain 14, 259, 261, 263, 266, 272,
US Bureau of Labor Statistics 32, 191, 480 381
US Bureau of Transportation Statistics 32, video conferencing 43, 48, 51, 86
34, 53, 209, 210, 217, 384
US Civil Aeronautics Board 52–3, 525, 536 welfare economics 18, 243, 404
US Congressional Budget Office 405, 437 World Bank 10, 13, 37, 54, 309
US Department of Energy 197 World Commission on Environment and
US Department of Homeland Security 388 Development 219
US Department of Housing and Urban World Trade Organization 23, 25, 26, 31,
Development 77 384
US Department of Transportation 240
US Environmental Protection Agency X–efficiency 127, 271, 298, 323, 393, 434,
286–7, 308, 317, 386–7 498, 506, 526
US Federal Aviation Administration 244, inert areas 174–5
344, 403
US Federal Highway Administration 421, yield management 249, 255–8, 310,
437, 445, 467, 471 312–13
US Federal Transit Administration 111,
449 zero-emission vehicles 310, 312–13

BUTTON 9781786435668 PRINT.indd 574 27/04/2022 11:07