Journal of Critical Realism

ISSN: 1476-7430 (Print) 1572-5138 (Online) Journal homepage: https://www.tandfonline.com/loi/yjcr20

Cost-Benefit Analyses of Transportation

Investments
Neither critical nor realistic

Petter Næss

To cite this article: Petter Næss (2006) Cost-Benefit Analyses of Transportation Investments,
Journal of Critical Realism, 5:1, 32-60, DOI: 10.1558/jocr.v5i1.32

To link to this article: https://doi.org/10.1558/jocr.v5i1.32

Published online: 21 Apr 2015.

Submit your article to this journal

Article views: 419

View related articles

Citing articles: 5 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=yjcr20
 COST-BENEFIT ANALYSES OF
TRANSPORTATION INVESTMENTS
Neither critical nor realistic1

by

PETTER NÆSS

Abstract. This paper discusses the practice of cost-benefit analyses of trans-

portation infrastructure investment projects from the meta-theoretical per-
spective of critical realism. Such analyses are based on a number of untenable
ontological assumptions about social value, human nature and the natural
environment. In addition, main input data are based on transport modelling
analyses based on a misleading ‘local ontology’ among the model makers. The
ontological misconceptions translate into erroneous epistemological assumptions
about the possibility of precise predictions and the validity of willingness-
to-pay investigations. Accepting the ontological and epistemological assumptions
of cost-benefit analysis involves an implicit acceptance of the ethical and polit-
ical values favoured by these assumptions. Cost-benefit analyses of transportation
investment projects tend to neglect long-term environmental consequences and
needs among population groups with a low ability to pay. Instead of cost-
benefit analyses, impact analyses evaluating the likely effects of project alter-
natives against a wide range of societal goals is recommended, with quantification
and economic valorisation only for impact categories where this can be done
in an ontologically and epistemologically defensible way.
Keywords: cost-benefit analysis, critical realism, ontology, epistemology, valori-
sation, politics

1. Introduction
This paper discusses the use of cost-benefit analyses of transportation infra-
structure investment projects from the metatheoretical perspective of criti-
cal realism. Besides criticising the ontological and epistemological foundations
of cost-benefit analyses within the transportation sector, the paper also

1 Thanks to the following persons who have given valuable comments on a previous

version of this paper: Kjell Harvold, Hubert Buch-Hansen, Peter Nielsen and Arvid
Strand.

© Koninklijke Brill NV, Leiden, 2006 JCR 5.1, 32-60

Also available online – www.brill.nl
 cost-benefit analyses of transportation investments 33

draws attention to the ethical/political assumptions on which such analyses

are usually based. In the final part of the paper, an alternative approach
to the evaluation of transport infrastructure projects will be outlined.
Transportation is a sector where the use of cost-benefit analysis as a
main tool for project evaluation is widespread. On the other hand, the
predictions about the impact of such projects (both on the benefit and cost
side) are often problematic, judged from a philosophy of science perspective.
To a great extent, the benefits consist of time savings, whereas various
types of environmental impacts, some of them long-term, are important
elements on the cost side in addition to the pure construction costs. Neither
the time savings nor the environmental impacts are easy to predict accu-
rately, and they are even more difficult to value in economic terms.
Moreover, the transportation sector is interesting because national gov-
ernmental goals have been changed in the 1990s from facilitating projected
growth in road traffic to channelling needs for transportation into more
environmentally friendly modes of travel than the private car, combined
with land use planning aiming to reduce the need for transport.2 Unbiased
and reliable project evaluations are important preconditions for such a
change of course. However, according to some critics, cost-benefit analyses
systematically tend to downplay long term effects and social concerns not
reflected in the willingness-to-pay of individuals.3 The transport modelling
tools used in order to estimate key positive as well as negative impacts
have also been accused of legitimating extensive road construction at the
cost of alternative strategies.4 The use of expert knowledge in the form of

2 Cf., inter alia, Ministry of the Environment (Norway), Governmental Policy Provisions

on Coordinated Land Use and Transport Planning, Oslo: Ministry of the Environment, 1993;
Ministry of Transport (Denmark), Trafik 2005. Problemstillinger, mål og strategier, Copenhagen:
Ministry of Transport, 1993; OECD/ECMT, Final Report of the Joint OECD/ECMT.
Project Group on Urban Travel and Sustainable Development, OECD/European Conference of
Ministers of Transport, 1993; Department of Transport, A New Deal for Transport: Better
for Everyone. The Government’s White Paper on the Future of Transport, HMSO: London, 1998;
J. Hine and J. Preston, ‘Introductory overview’, in J. Hine and J. Preston, eds, Integrated
Futures and Transport Choices, Aldershot: Ashgate, 2003, pp. 1-9.
3 E.g. B. Orland, Economic Evaluation: Cost-Benefit Analysis, St. Louis: East St. Louis

Action Research Project,1998; F. Ackerman and L. Heinzerling, Priceless: On Knowing the

Price of Everything and the Value of Nothing, New York: New Press, 2004.
4 J. Kenworthy, ‘Don’t shoot me—I’m only the transport planner’, in P. Newman,

J. Kenworthy and T. Lyons, Transport Energy Conservation, Perth: Murdoch University,

1990; B. Moen and A. Strand, ‘Når kapasitetsproblemer i vegnettet oppstår, skal andre . . .’,
Report 2000:1, Oslo: Norwegian Institute for Urban and Regional Research, 2000;
34 petter næss

cost-benefit analyses of transportation projects thus touches on important

issues of accountability, democracy and the implementation of national
goals of sustainable development.
For readers of this journal, the philosophy of critical realism hardly needs
any general presentation. In my discussion, I draw particularly on critical
realism’s conception of the world as a differentiated and stratified system
consisting of different layers, and its understanding of the nature of causal
relationships in non-closed systems and the relationship between social struc-
tures and agents.5

2. Cost-Benefit Analysis. Purpose, Techniques and Areas of Use

The purpose of cost-benefit analysis (CBA) is to estimate and total up the
equivalent money value of the benefits and costs to the community of proj-
ects in order to establish whether they are worthwhile. According to its
proponents, cost-benefit analysis is a powerful, widely used and relatively
easy tool for deciding whether to make a change. The method imposes an
accounting framework that prescribes classes of benefits and costs to con-
sider, means to measure them, and approaches for aggregating them. Even
though the approach originated as an analogy to private studies of invest-
ment, and thus to calculate a ‘go-no-go’ decision, the technique is flexible
and is used in public sector planning as a tool for choosing among a range
of alternatives, and for comparison between projects of differing lengths.6
In a cost-benefit analysis, monetary values are attributed to all impacts
of a project. If the calculated aggregated value of all impacts of an ini-
tiative is positive, the initiative is considered profitable for society. The

A. Tennøy, Bidrar bruk av transportanalyser i byplanleggingen til vekst i biltrafikken?, paper for
the conference ‘Trafikdage på Aalborg Universitet’, August 25-26, 2003.
5 R. Bhaskar, A Realist Theory of Science, Hassocks: Harvester, 1978; R. Bhaskar, The

Possibility of Naturalism: A Philosophical Critique of the Contemporary Human Sciences, Hassocks:

Harvester, 1989; B. Danermark, M. Ekström, M. Jacobsen and J. C. Karlsson, Explaining
Society: Critical Realism in the Social Sciences, London/New York: Routledge, 2001;
R. Bhaskar and B. Danermark, ‘Metatheory, interdisciplinarity and disability research:
a critical realist perspective’, paper for the symposium ‘Transcending conventional bound-
aries: contemporary challenges for social science and world peace’ at the University of
Tromsø, Norway, 2005; A. Sayer, Method in Social Science: A Realist Approach, London:
Routledge, 1992; M. Archer, Being Human. The Problem of Agency, Cambridge: Cambridge
University Press, 2000.
6 D. Bjornstad, Cost-Benefit Analysis, http://www.ncedr.org/tools/othertools/costbenifit/

lead.htm, accessed June 6, 2005.

 cost-benefit analyses of transportation investments 35

main valuation principle on which cost-benefit analyses are usually based

is that the monetary value of a positive impact is defined as equal to the
amount of money that the population is willing to pay to obtain it. Thus,
when a project is assessed as profitable for society, this means that the
population at large is willing to pay as least as much for the project as it
actually costs. When a project is analysed, there will be a need to compare
costs and benefits appearing at different times. This is accounted for by
discounting future monetary benefits and costs to a reference date (usually
the time of decision to implement the project), using a given annual dis-
counting rate.7
Based on ideas tracing back to the French engineer Jules Dupuit in the
mid-nineteenth century, formal cost-benefit analysis was first used in prac-
tical planning in the USA in the 1930s as a result of the impetus provided
by the Flood Control Act of 1936. In order to control federal spending,
this Act specified that the U.S. Corps of Engineers could carry out projects
for the improvement of the waterway system only ‘if the benefits to whom-
soever they may accrue are in excess of the estimated costs’. The engineers
of the Corps did this without much, if any, assistance from the econom-
ics profession. It was not until about twenty years later (in the 1950s) that
economists attempted to develop a rigorous, consistent set of methods for
measuring benefits and costs and deciding whether a project is worthwhile.8
Along with the increasing influence of neoclassical economics on public-
sector decision-making, there has been an increasing pressure in a number
of countries to employ cost-benefit analyses to decide whether or not to
implement governmental initiatives.9 Examples of projects subjected to cost-
benefit analysis include waste incineration plants, hydro-electrical development,
training programs, health care systems, and, not least, transportation infra-
structure investments.
Within the latter category, the main impacts on the benefit side are usu-
ally time savings10 and any safety and local environmental improvements,

7 Ministry of Finance, Norway, Veiledning i samfunnsøkonomiske analyser, Oslo: Ministry

of Finance, 2000.
8 L. Lohmann, Cost-Benefit Analysis: Whose Interest, Whose Rationality?, London: The

Corner House, 1997; T. Watkins, Introduction to Cost Benefit Analysis, San Jose: San Jose
Sate University, Economics Department, 2005.
9 Ackerman and Heinzerling, Priceless (Note 3, above); Bjornstad, Cost-Benefit Analysis

(Note 5, above).
10 The question of whether or not impacts on the regional economy should also be

included as a separate item in the analysis is contested. Many economists hold that such
36 petter næss

whereas construction costs (notably construction materials and labour power)

and operating costs are obvious items on the cost side. Depending on the
context and type of project, impacts on safety and local environment may,
however, also be negative (e.g. due to encroachments into landscapes and
demolition of houses, or increased number of accidents due to additional
traffic) and will then occur on the cost side of the calculation. This also
applies to long-term and global impacts such as the impacts of the project
in terms of greenhouse gas emissions, where some projects (e.g. road pricing
or improved public transport in cities) may contribute to reduce these emis-
sions while other projects (e.g. increased highway capacity) may lead to
increased emissions.
Following the general principle of discounting, benefits as well as costs
of transportation projects occurring in the future are discounted to their
present value using a stated discounting rate. This rate varies somewhat
across nations. For instance, the recommended discounting rate for trans-
portation projects is six per cent annually in the United Kingdom, seven
per cent in the United States and Australia, eight per cent in Norway, ten
per cent in Canada.11 For projects evaluated by the World Bank, a dis-
counting rate of 12 per cent annually is used.12 The discounting rates also
apply to environmental impacts, positive as well as negative. In the liter-
ature, there is some debate on the reasonableness of using the stated dis-
counting rates for very long-term environmental impacts, for example, the
contribution of the projects to global climate change.13 The recommendation
of national governments on how to deal with such effects in cost-benefit
analyses of transportation projects therefore also vary, as does the treatment
of such impacts in actual project evaluations. So far, most cost-benefit

effects are in principle derivative from time savings already accounted for, and a sep-
arate inclusion of these effects would therefore imply a double counting. However, this
assumption holds true only under conditions of ‘perfect competition’ in the economy as
a whole. (World Bank, Notes on the economic evaluation of transport projects, Transport Note
TRN 5, New York: The World Bank, 2005; SACTRA, Transport and the Economy, London:
Standing Advisory Committee on Trunk Road Appraisal, 1999).
11 Department of Transport (UK), Guidance on the Methodology for Multi-Modal Studies,

London: Department of Transport, 2000; K. Small, ‘Project evaluation’, in J. A. Gómez-

Ibanez, W. B. Tye and C. Winston, eds, Essays in Transportation Economics and Policy: A
Handbook in Honor of John R. Meyer, Washington D. C.: Brookings Press, 1999; Ministry
of Finance, Norway, Veiledning (Note 7, above).
12 World Bank, Notes (Note 10, above).

13 See, e.g., C. D. Kolstad and M. Toman, The Economics of Climate Policy, Discussion

Paper 00-40REV, Washington D. C.: Resources for the Future, 2001.

 cost-benefit analyses of transportation investments 37

analyses of transportation projects have omitted impacts on the global cli-

mate in their economic valuation of the projects.
As mentioned above, time savings make up an important item on the
benefit side of cost-benefit analyses of transport infrastructure investment
projects, and in many cases these savings are the dominating positive eco-
nomic value. In order to assess this value, the value of each time unit
saved must be calculated, as well as the number of travellers affected and
the average number of minutes (or seconds) saved by each traveller. The
latter two components are normally estimated using transport model com-
putations, whereas the value of each time unit saved is usually based on
willingness-to-pay investigations. Typically, different values per minute saved
are used for travellers using different modes of transport, reflecting differences
in perceived inconvenience when making the trips as well as differences in
average income levels.14
Transport model computations also make important inputs into the assess-
ments and valuations of safety and environmental impacts. For example,
based on transport modelling, estimates are made of the total amount of
traffic, its distribution on the road network and between different hours of
the day, and the average speed levels. From these estimates, assessments
are made of, among other things, energy use and greenhouse gas emis-
sions, local air pollution, noise, and the number of traffic accidents. A
translation of the estimates measured in physical entities into monetary
terms is then carried out, using some willingness-to-pay-based assumptions
about the economic value per physical unit of change. Willingness-to-pay
is also used as the principle for assessing the social value of non-traffic
impacts like encroachments and fragmentation of natural and outdoor recre-
ation areas, cultural heritage, etc.

3. Ontological Assumptions of Cost-Benefit Analysis

Cost-benefit analysis is based on a certain set of ontological assumptions
that are far from uncontested. As mentioned above, the social value of the
impact of a project is assumed to be equal to the population’s willingness
to pay for obtaining or avoiding this impact. In other words, social values
are considered to be the aggregate of individuals’ willingness to pay (which
in its turn is a combination of the preferences of these individuals and

14 Small, Project Evaluation (Note 11, above); Ministry of Finance (Norway), NOU.

1997: 27 Nytte-kostnadsanalyser, Oslo: Ministry of Finance, 1997.

38 petter næss

their ability to pay). Such a way of conceiving of social value is an exam-

ple of what Margaret Archer terms upward conflation. Although Archer refers
to upward conflation as a position where social structures are considered
as nothing but the aggregate preferences of individual, instrumentally rational
agents, this notion appears to be relevant also to the corresponding con-
ception of social value.15 This purely individualist position, expressed in a
condensed form in previous British prime minister Margaret Thatcher’
famous phrase, ‘there is no such thing as society’, contrasts fundamentally
with a critical realist view, according to which both social structures and
agents have particular properties and causal powers.16 More fundamentally,
individuals and society make up different strata of reality. According to crit-
ical realism, reality is stratified and differentiated, consisting of different
‘layers’ where new causal powers and properties that are irreducible to
lower layers emerge when moving from a lower to a higher layer.17 The
value of something for society is thus not reducible to the aggregate pref-
erences of individuals, let alone their willingness (or acceptance, cf. the sec-
tion on epistemological assumptions) of individuals to pay for it.
This insight is reflected in planning theory on the functions of and
justification for intervention into the market mechanism in the form of
public-sector planning. According to Richard Klosterman, two of the main
reasons for maintaining such planning are the need to provide public goods
and to counteract ‘externalities’.18 These tasks are more adequately addressed
through political deliberation and decision-making than through the count-
ing of atomistic individual preferences.19
Moreover, for cost-befit analysis to produce accurate results, conditions
of perfect market competition must exist. These conditions include, among
other things, the requirement that humans act as fully informed, utility-
maximising consumers. However, actual human behaviour hardly conforms

15 Archer, Being Human, p. 6 (Note 5, above).

16 Archer, ibid.; Bhaskar, The Possibility (Note 5, above).
17 Bhaskar, ibid.; Danermark et al.; Explaining (Note 5, above).

18 R. Klosterman, ‘Arguments for and against planning’, Town Planning Review

vol. 56, no. 1, 1985, pp. 5-20.

19 J. Mønnesland, P. Næss and A. Strand, ‘Land use planning and cost effectiveness’,

in J. Røhr and A. Hagen, eds, Strategisk planlegging—mote eller metode?, Working Paper
20/1996, Lillehammer: Forum for utdanning i samfunnsplanlegging and Høgskolen
i Lillehammer, 1996, pp. 77-88; T. Sager, Democratic Planning and Social Choice Dilemmas:
Prelude to Institutional Planning Theory, Aldershot: Asghate, 2002, pp. 74-80 and 158-9.
 cost-benefit analyses of transportation investments 39

to this requirement.20 Along with instrumental rationality, human actions

are guided by social rationality as well as communicative rationality and
influenced by traditions and habits.21 They are also guided by ethical con-
siderations which may well differ from the principle of individual utility
maximising. Thus, the method is based on a reductionist and flawed assump-
tion about human nature. Neither is the presupposition of fully informed
individuals met in real life. In particular, this applies to the ramifications
of the various environmental impacts of the alternatives. For example, very
few, if any, individuals are likely to be fully informed about the environ-
mental changes and related impacts on societies and individuals resulting
from greenhouse gas emissions. I will discuss this in more detail in the sec-
tion on the epistemological assumptions of cost-benefit analyses.
As mentioned above, traffic forecasts constitute an important base for
the calculation of significant items on the benefit side (notably time sav-
ings) as well as on the cost side (notably traffic accidents, energy use, air
pollution and noise) of cost-benefit analyses of transportation investment
projects. These forecasts are usually carried out by engineers based on a
variant of a four-step gravity-based transport model.
However, conventional transport models are to a great extent insensitive
to the influences of land use and transport infrastructure investments on
the modal split between car and other modes of conveyance. The most
commonly used transport models therefore have a tendency to underestimate
the possibilities of influencing traffic development through transport policy
measures and coordinated land use and transport planning. Moreover, the
traffic-generating effect of increased road capacity in situations with congestion
on the road network is to a great extent ignored.22 However, theoretical

20 Ministry of Finance (Norway), Nytte-kostnadsanalyser (Note 14, above); Lohmann,

Whose Interests (Note 7, above); Archer, Being Human, pp. 51-85 (Note 5, above).
21 P. Diesing, Reason in Society: Five Types of Decisions and Their Social Conditions, Urbana,

Illinois: University of Illinois Press, 1962; J. Habermas, The Theory of Communicative Action,
Volume 1: Reason and the Rationalisation of Society. London: Heinemann, 1984 and Volume
2: Lifeworld and System: A Critique of Functionalist Reason, Boston, MA: Beacon Press, 1987;
P. Bourdieu, Distinction: Critique of the Judgment of Taste, Cambridge: Harvard University
Press, 1984; A. Giddens, Central Problems in Social Theory, Berkeley: University of California
Press, 1979.
22 For example, in the Environmental Impact Assessment (EIA) of the proposed Third

Link across the Limfjord in Aalborg, Denmark, the presupposed traffic growth rate was
the same (two per cent annually) in the alternatives, where a new motorway was con-
structed, as in the zero alternative, where no road capacity increase took place. Similarly,
in the EIA of a proposed new link of the E18 highway through Kristiansand, Norway,
40 petter næss

as well as empirical research has demonstrated the existence of mecha-

nisms by which transport infrastructure may lead to ‘induced travel’ as
well as changes in the distribution between different travel modes.23 There
is also ample evidence of the influence of land use patterns on travel behav-
iour in urban areas.24
The forecasts are worked out by experts who belong to a particular pro-
fessional ‘culture’. In modern society, different sectors represent different

traffic growth within the planning area was presupposed to be almost identical, regard-
less of whether or not a considerable increase in road capacity was implemented. Both
in Aalborg and Kristiansand, ‘fixed matrixes’ (i.e. exogenously given total future amounts
of traffic, where the traffic in the analysed alternatives differ only in their distribution
between different parts of the road network) were used in spite of the fact that the
same analyses showed a breakdown of traffic flows on large proportions of the road
network in a decade or two if the road capacity was not increased. However, if con-
gestion really grew that bad, many peak-period motorists would choose to go by pub-
lic transport, bike or foot instead of driving (see references in Note 23, below). These
two examples are by no means anecdotal. According to Cowi (Paper dated October
10, 2004 in connection with the consideration of the Naturklagenævnet on a complaint
about the EIA for the Third Limfjord Crossing), the analysis of traffic consequences of
the proposed Third Link across the Limfjord ‘has methodically [. . .] been in accor-
dance with the approach generally followed when preparing Environmental Impact
Assessments. [. . .] The method used in the assessments of induced traffic is the method
generally used for new infrastructure construction in Denmark.’ See also, e.g., Kenworthy,
Don’t Shoot (Note 4, above); Tennøy, Bidrar (Note 3, above); O. A. Nielsen and
M. Fosgerau, Overvurderes tidsbenefit af vejprojekter, paper for the conference Trafikdage på
Aalborg Universitet, 2005; Moen and Strand, Når kapasitetsproblemer (Note 4, above);
N Arge, T. Homleid and A. Stølan, Modeller på randen . . . Bruk av transportmodeller i norske
byområder. En evaluering, Oslo: Civitas, 2000.
23 See, e.g., M. J. H. Mogridge, ‘The self-defeating nature of urban road capacity

policy: a review of theories, disputes and available evidence’, Transport Policy vol. 1, no.
1, 1997, pp. 5-23; P. Næss and S. L. Sandberg, Choosing the fastest mode? Travel time and
modal choice in two transport corridors of Oslo, Report 1998:15, Oslo: Norwegian Institute
for Urban and Regional Research, 1998; S. Cairns, S. Atkins and P. Goodwin,
‘Disappearing traffic? The story so far’, Municipal Engineer vol. 151, March 2002, pp.
13-21; R. B. Noland and L. L. Lem, ‘A review of the evidence for induced travel and
changes in transportation and environmental policy in the US and the UK’, Transportation
Research Part D, vol. 7, No. 1, 2002, pp. 1-26; see also SACTRA, Transport and the Economy.
24 See, e.g., D. Stead and S. Marshall, ‘The relationships between urban form and

travel patterns: an international review and evaluation’, European Journal of Transport

Infrastructure Research vol. 1, 2001, pp. 113-41; P. Næss, ‘Urban structures and travel
behaviour: experiences from empirical research in Norway and Denmark’, European
Journal of Transport Infrastructure Research, vol. 3, 2003, pp. 155-78; P. Næss, ‘Residential
location affects travel behaviour—but how and why?’, Progress in Planning vol. 63/2, 2005,
pp. 167-257.
 cost-benefit analyses of transportation investments 41

‘cultures’ with varying values, attitudes and perceptions of the situation.

Within each sector, established rules, standard operating procedures and
routines govern the actions of the agents.25 Arguably, a ‘local ontology’
prevails among transport analysts, assuming that neither the total amount
of travel within a region nor the distribution of travellers between different
modes of transport is influenced by changes in the road capacity.26 This
inevitably produces unreliable results in transport analyses, and accordingly
also unreliable results for the cost-benefit analysis based on these traffic
forecasts. In particular, these errors are serious in situations where there
are politically adopted aims of dealing with congestion problems in ways
other than through road development facilitating traffic growth (e.g. through
land use policy, improved public transport and road pricing).27
As mentioned above, environmental impacts are main items of the cost
side of transportation investment projects (apart from the mere construc-
tion costs). A final, highly problematic ontological assumption of cost-benefit
analyses of such projects is the belief that nature is fully or at least to a
very great extent, substitutable with human-made capital, with an unlimited
capacity to support growth in consumption and production. Without this
assumption, the discounting of long-term negative environmental consequences
to negative present values much smaller than the originally calculated

25 A. Strand and B. Moen, Lokal samordning—finnes den? Report 2000:18, Oslo, Norwegian
Institute for Urban and Regional Research, 2000; J. P. Olsen, ‘Analysing institutional
dynamics’, Statswissenscaften und Staatspraxis, vol. 2, 1992, pp. 247-71.
26 The roots of this ‘local ontology’ are somewhat obscure, however it seems to be

related to the ‘upward conflation’ prevalent in neoclassical economics, according to

which structural influences on human behaviour are ignored or downplayed. A neglecting
of the influence of our physical environment on human actions also has long roots in
sociological traditions; for a critique, see, e.g., R. E. Dunlap and W. R. Catton Jr.,
‘What environmental sociologists have in common (whether concerned with ‘built’ or
‘natural’ environments)’, Sociological Inquiry vol. 53, 1983, pp. 113-35; and T. Benton,
‘Why are sociologists naturephobes?’, In J. Lopez and G. Potter, eds, After Postmodernism:
An Introduction to Critical Realism, pp. 132-45, London/New York: The Athlone Press,
2001. The prevalence of the ‘local ontology’ among transport modelling analysts may
perhaps also reflect power relations: by underestimating the possibilities of influencing
traffic development through transport policy measures and coordinated land use, the
models produce results legitimating more road construction (which may in its turn lead
to more jobs for the consulting firms making the forecasts). See P. Næss, B. Flyvbjerg
and S. Buhl, ‘Do road planners produce more “honest numbers” than rail planners?
An analysis of accuracy in road traffic forecasts in cities vs. peripheral regions’, Transport
Reviews, vol. 26, 2006 (forthcoming).
27 Tennøy, Bidrar (Note 4, above).
42 petter næss

negative values would not be possible. The diminution of future conse-

quences by means of discounting has partly been justified by a psychological
phenomenon among individuals (impatience and uncertainty about the pos-
sibility of utilising consumption opportunities if they are postponed), and
partly by expected economic growth increasing the general possibilities for
consumption in a society. However, according to Flemming Møller et al.,
the psychological time preference related to impatience and uncertainty is
hardly relevant for decision-making at a social level, albeit understandable
when seen from the perspective of individuals with their limited lifetimes.
The justification for discounting future consequences therefore rests with
the assumption of continuously rising consumption levels.28
How realistic, however, is this assumption in the case of investments in
physical structures with a long-term durability and even longer-term envi-
ronmental consequences? Among the negative environmental impacts, emis-
sion of greenhouse gases is one of the most important. The amount of
greenhouse gases emitted per kilometre travelled by one person varies con-
siderably among different means of transport. In addition, growth in the
amount of transport contributes, other things being equal, to increasing
the emissions. It is therefore crucial for the ranking of alternatives that the
costs resulting from these emissions are calculated in an unbiased way.
However, given an annual discounting rate of seven per cent, as rec-
ommended in, among other countries, USA and Australia, a climate dis-
aster occurring in 150 years causing damage of ten trillion dollars has a
discounted cost today of only 391 million dollars, i.e. an amount twenty-
six thousand times smaller than the non-discounted amount.29 In other
words, the importance of the contribution of alternatives to climate change
is dramatically reduced through the discounting procedure. Needless to say,
this places alternatives generating high greenhouse gas emissions in a much
more favourable light than what would otherwise be the case, and may
lead to preference of such solutions rather than lower-emission alternatives.
In principle, similar bias will also occur in relation to other long-term envi-
ronmental consequences such as loss of biodiversity and biological productivity
due to local pollution and the fragmentation of natural areas represented
by transportation infrastructure construction.

28 F. Møller et al., Samfundsøkonomisk vurdering af miljøprojekter, København: Miljø- og

Energiministeriet, Danmarks Miljøundersøgelser, Miljøstyrelsen og Skov- og naturstyrelsen,

2000.
29 Small, Project Evaluation (Note 11, above). If instead the 12 percent discounting rate

recommended by the World Bank had been used, the calculated present value would
be 24 million times smaller than the non-discounted value!
 cost-benefit analyses of transportation investments 43

Admittedly, in some countries, including Norway, the government has

advised against trying to value long-term environmental consequences in
monetary terms when doing cost-benefit analyses.30 However, in several
other countries, such impacts are made subject to economic valuation and
discounting,31 if they are at all taken into consideration. Notably, cost-
benefit analyses of greenhouse gas emissions have exerted considerable
influence on governmental decision-making in countries like USA and
Denmark.
As mentioned above, the justification for discounting future consequences
is based on expectations of continuous economic growth: if we are ten
times richer in the future, paying a cost of 1,000 dollars will impact our
welfare much less than paying the same amount of money today. However,
in the example above, we are not talking about getting ten times richer,
but twenty-six thousand times richer on average (cf. the justification for
the discounting rate). Is such an increase in wealth at all possible? A number
of studies indicate that the present level of consumption of natural resources
and environmental pollution is already higher than what can physically be
sustained in the long term.32 A continuous growth in the consumption of
natural resources and the ‘throughput’ of these resources into waste and
pollution is therefore not environmentally sustainable.33 The answer of

30 Ministry of Finance, Norway, Veiledning (Note 7, above).

31 See, e.g., T. Odgaard, Current Practice in Project Appraisal in Europe, paper for the
conference Trafikdage på Aalborg Universitet, 2005; W. D. Nordhaus, Managing the
Global Commons: The Economics of Climate Change, Cambridge, MA: MIT Press, 1994;
Kolstad and Toman, The Economics of Climate Policy; B. Lomborg, The Sceptical Environmentalist:
Measuring the True State of the World, Cambridge: Cambridge University Press, 2001;
W. A. Pizer, Climate Change Catastrophes, Discussion Paper 03-31, Washington D. C.:
Resources for the Future, 2003. Based on such calculations, Nordhaus (ibid.) hypothesises
that a 3ºC rise in the global average temperature results in a 1.33% loss in world output.
32 See, e.g., B. Commoner, The Closing Circle: Nature, Man and Technology, New York:

Alfred Knoph, 1971; World Commission on Environment and Development (WCED),

Our Common Future, Oxford/New York: Oxford University Press, 1987; R. Goodland,
‘The case that the world has reached limits: more precisely that current throughput
growth in the global economy cannot be sustained’, in Goodland, Daly, El Serafy and
von Droste, eds, Environmentally Sustainable Economic Development: Building on Brundtland, New
York: UNESCO, 1991; M. Wackernagel and W. Rees, Our Ecological Footprint: Reducing
Human Impact on the Earth, Philadelphia, USA: New Society, 1996.
33 T. Haavelmo and S. Hansen, ‘On the strategy of trying to reduce economic inequal-

ity by expanding the scale of human activity’, in Goodland et al, Environmentally Sustainable
Development (Note 32, above).
44 petter næss

neoclassical environmental economists to this dilemma is that economic

growth must (and can) be ‘decoupled’ from natural resource consumption
and environmental load.34 Such a ‘dematerialised’ growth requires a mas-
sive shift to more ‘eco-efficient’ ways of production and consumption, i.e.
reducing the environmental load per unit produced of a commodity or a
service. If, for example, the resource consumption and environmental load
per unit can be reduced to one fourth, i.e. by ‘factor four’, the negative
impact of nature and the environment can be halved while doubling the
production.35 This line of thought is the base for the so-called ‘factor four’
objectives stated by several environmental agencies since the early 1990s.36
There is evidence that improved eco-efficiency has taken place in sev-
eral sectors of the economy, and in some cases (e.g. the production of beer
cans, energy-saving lamps and electronic equipment) the factor of improve-
ment is high. However, for the economy as a whole the degree of improve-
ment is modest, and reduction of the environmental load per unit has most
often been offset or outweighed by increases in the volume of production.37
In certain sectors of the economy, such as housing construction and trans-
port, a reduction of the per unit environmental load by factor four will
be politically highly controversial although technically possible.38
However, when discounting long-term environmental consequences, we
are not assuming a ‘factor four’ dematerialisation of the economy, but (in
the case of a seven per cent discounting rate over 150 years) a ‘factor
twenty-six thousand’ dematerialisation (and even much more if the World
Bank discounting rate of 12 per cent is used). The absurdity of such assump-
tions is probably obvious to everyone except neoclassic economists.39
The separation of neoclassical economic theory from any material con-
text exemplified above illustrates the ‘autistic’ character of this theory, where

34 See, e.g., WCED, Our Common Future, p. 51 (Note 32, above).

35 E. von Weizsäcker et al., Factor Four: Doubling Wealth, Halving Resource Use. The New
Report to the Club of Rome, London: Earthscan, 1997.
36 E.g. Nordic Council of Ministers, Factors 4 and 10 in the Nordic Countries: the Transport

Sector, the Forest Sector, the Building and Real Estates Sector, the Food Supply Chain, TemaNord
199:528, Copenhagen: Nordic Council of Ministers, 1999.
37 J. H. Spangenberg et al., Towards Sustainable Europe, Luton: Friends of the Earth,

1995.
38 For a discussion, see K. G. Høyer and P. Næss, ‘The Ecological Traces of Growth’,

Journal of Environmental Policy and Planning, vol. 3, 2001, pp. 177-92.

39 For a broader discussion including a political economy perspective, see P. Næss,

‘Unsustainable growth, unsustainable capitalism’, forthcoming in Journal of Critical Realism,

vol. 5, no. 2, 2006.
 cost-benefit analyses of transportation investments 45

the axiomatic assumptions are so to speak immune to any corrective inputs

from other sciences.40 In critical realist terms, the neoclassical assumptions
about discounting and economic growth disregard the dependence of the
economic processes on the natural environment, which consists of strata
below the strata of individual actions and social structures. Thus, neoclas-
sical economics fail to realise the fact that reality is hierarchically stratified,
where each higher-level stratum includes the powers and mechanisms of
the lower strata, in addition to the new powers and mechanisms emerg-
ing at the higher stratum. While social life cannot be reduced to the pow-
ers and mechanisms of lower strata, it still by necessity involves the powers
and mechanisms of, among other things, the objects of physics, chemistry
and biology. The neoclassical assumptions about infinite economic growth
(and infinite dematerialisation) represent a particular sort of ‘downward
conflation’ where the powers and mechanisms of the natural world are con-
sidered totally controllable by humans as if they were mere epiphenomena of
the human world.

4. Epistemological Assumptions of Cost-Benefit Analysis

Many of the impacts dealt with in cost-benefit analyses of transportation
investment projects are expected changes in human behavioural patterns
resulting from the projects. This applies, for example, to time savings and
changes in traffic accidents, emission levels and noise. Economic valuation
of these effects presupposes that it is possible to predict each of these out-
comes with a high degree of accuracy. However, for several reasons, such
precise predictions will hardly be possible.
Above, the inadequate ‘local ontology’ prevalent among transport mod-
elling engineers was discussed. However, even if the current transport mod-
elling tools were replaced by models representing state-of-the-art knowledge
about relevant mechanisms influencing travel behaviour, precise prediction
about the future amount of traffic and its distribution between modes of
conveyance would not be possible.
Several theorists of science, including a number of prominent critical
realists,41 have rejected the possibility of making any scientifically based

40 E. Fullbrook, ‘Broadband versus narrowband economics’, in E. Fullbrook, ed., A

Guide to What’s Wrong With Economics, London: Anthem Press, 2004.

41 E.g. Sayer, Method (Note 5, above); T. Lawson, Economics and Reality, London/New

York: Routledge, 1997; Danermark et al., Explaining Society (Note 5, above); A. Sayer,
Realism and Social Science, London/Thousand Oaks/New Delhi: Sage Publications, 2000.
46 petter næss

predictions at all about events and activity patterns in society.42 Elsewhere,

I have argued against this highly pessimistic assumption, which would have
dramatic implications for planning and political government if it were true.43
However, there are indeed important limitations regarding the exactness
of possible predictions. The theoretical and empirical knowledge relevant
for assessments of travel behavioural impacts of transportation infrastructure
projects is seldom context-independent. If such knowledge is to be used in
impact assessments, it must therefore be adapted and modified according
to the actual context. Both for this reason and because society is constantly
changing, any predictions about the impacts of transport investment proj-
ects on travel behaviour will always be crude. Society is not a closed sys-
tem, but a system where many different causal mechanisms are at work,
combined in a more or less contingent way. Although the contexts of the
relationships between transport infrastructure, urban form and travel may
be considered ‘pseudo-closed’ systems rather than completely open systems,
the event regularities that can be identified will always be context-depend-
ent—in space as well as in time. Therefore, any generalisation to other
geographical contexts or to the future must include careful, qualitative
assessments of possible differences between the original space/time context
and the context to which we want to generalise the research results.
Regarding the generalisation across time, this implies a discussion of pos-
sible traits of development that may reproduce, weaken or strengthen exist-
ing relationships.
The predictions that are possible to make on the basis of social science
studies (such as studies of the relationships between urban land use, trans-
port infrastructure and travel) are therefore of the crude ‘rule of thumb’
type. They are limited to aggregate-level predictions of the impacts of the
causal factor, not of future events or states (the latter being much more
difficult and uncertain due to the impossibility of knowing the future com-
bination and development of a multitude of other causal mechanisms
influencing travel behaviour). They are also necessarily non-exact, i.e. lim-
ited to the direction and order of magnitude of the impact.44 These limitations
have their important implications regarding the levels of measurement—

42 See also O. Brox, Praktisk samfunnsvitenskap, Oslo: Universitetsforlaget, 1989;

B. Flyvbjerg, Making Social Science Matter: Why Social Inquiry Fails and How It Can Succeed
Again, Cambridge: Cambridge University Press, 2001.
43 P. Næss, ‘Prediction, regressions and critical realism’, Journal of Critical Realism,

vol. 2, No. 2, 2004, pp. 133-64.

44 For a more thorough discussion, see ibid.
 cost-benefit analyses of transportation investments 47

ordinal, interval or ratio—that can reasonably be used when assessing

behavioural consequences of transport investment projects.
Not only will the assessment of travel behavioural impacts necessarily
be crude, but the economic valuation of these impacts is also inevitably
uncertain and contested. Traditionally, the valuation of time savings has
been based on a comparison with the most profitable alternative use of
the time. Here, the value of travel time has usually been set as zero,
whereas the alternative time use has been the value of working hours. An
underlying assumption for this way of assessment is that the seconds or
minutes saved when new transportation infrastructure is introduced will by
and large translate into a higher number of working hours at an aggre-
gate level.45 Obviously, several objections may be raised against this assump-
tion, for example, in situations with unemployment. People may also choose
to utilise their time saving for leisure purposes instead of using the time
to earn more money.
During recent years, this method has been supplemented by an alter-
native approach based on willingness-to-pay studies, either in the form of
‘revealed preferences’ (investigations of how much people actually pay in
order to travel by faster modes of transport or on faster transport infra-
structure, for example, toll roads) or as ‘stated preferences’ (respondents’
answers to hypothetical questions about which among alternative combinations
of travel time and travel expenses they would prefer). However, here too
considerable uncertainty is involved. Valuation of time varies widely among
population subgroups and probably depends critically on individual cir-
cumstances.46 For example, possibility to work while travelling may influence
strongly on the perceived cost of travel time. The time value of different
parts of a trip (e.g. walking to public transport stops, waiting time, in-
vehicle time etc.) has also been found to vary significantly, as has the value
of time spent on car driving under different driving conditions. Moreover,
people with strong humanistic and environmental values may place less
emphasis on high incomes and may have lower ability-to-pay.47
A more fundamental question is whether it is at all meaningful to add
up tiny increments of time savings (e.g. thirty seconds due to the straight-
ening of bends previously necessitating speed reduction on a particular
stretch of a road) to an aggregate economic value for society. It should

45 Ministry of Finance, Norway, Nytte-kostnadsanalyser (Note 14, above).

46 Small, Project Evaluation (Note 11, above).
47 Orland, Economic Evaluation (Note 3, above).
48 petter næss

also be noted that predicted travel-time savings may sometimes be offset

by behavioural shifts induced by the new infrastructure. In particular, this
is likely to happen in urban areas where the new infrastructure relieves
congestion. In such situations, people previously deterred by congestion
from driving may shift from public transport to car, thus contributing to
increase the congestion level again.48 This latter source of exaggerated
assessment of the value of time savings is, however, related to the traffic
forecasts (cf. above) and not to the price set per minute of travel time
saved.
The economic valuation of the environmental consequences of transport
investments projects is probably even more problematic. Apart from the
serious problems related to the discounting of long-term effects (cf. the sec-
tion on ontological assumptions), important epistemological objections can
also be made to the valuation of environmental impacts at a given time.
As mentioned earlier, the translation of physical impacts into monetary
figures is usually made by means of willingness-to-pay investigations. Here,
the ‘stated preferences’ approach dominates, although travelling expenses
have sometimes been used as a ‘revealed preferences’-based measure of
how much people are willing to spend in order to experience a certain
natural area. (Obviously, the latter method is not appropriate when the
consequences affect a large geographical area rather than a specific, delimited
location.)
Willingness-to-pay investigations of the ‘stated preferences’ type are also
encumbered with several validity problems. First, people often do not have
sufficient knowledge about the impact of a project to be able to state in
a meaningful way how much they are willing to pay in order to make a
contribution to avoid this impact. For example, it may be uncertain whether
what you measure is the willingness to pay for the entire environmental
good (e.g. an outdoor recreation area) or to avoid the negative impact of
a proposed project on this area (e.g. in the form of fragmentation, noise
and loss of a certain part of its space). The latter would be the relevant
measure, but answering this would require quite detailed information—and
could the respondents be expected to possess this level of information?
Even more problematic, of course, is the use of willingness-to-pay state-
ments as a basis for valuation of a project’s contribution to global climate

48 A. Downs, ‘The law of peak-hour expressway congestion’, Traffic Quarterly vol. 16,

1962, pp. 393-409; J. M. Thomson, Great Cities and Their Traffic, London: Gollancz, 1977;
Mogridge, Self-Defeating (Note 23, above); Small, Project Evaluation (Note 11, above).
 cost-benefit analyses of transportation investments 49

change. Here, most people (and scientists too!) have far too little knowledge
of the various direct and indirect impacts to make meaningful economic
statements.
Secondly, for several reasons, individuals’ statements about their will-
ingness to pay do not necessarily reflect their actual valuation of an envi-
ronmental good. Even in the most commoditised societies, practices of
assigning prices to a wide range of items are likely to be unfamiliar, cir-
cumscribed, irrelevant or disallowed. In that case the question, ‘How much
would you be willing to pay?’, is unlikely to elicit responses satisfactory to
surveyors.49 Some people may, for example, consider the protection of an
outdoor recreation area to be something that should be subject to politi-
cal decision-making, for which they on grounds of principle should not
pay. Such a person may perhaps tick for ‘zero’ as their answer to the
question of how much they are willing to pay in order to maintain the
opportunity of experiencing the area with the same landscape qualities as
today—even if the person is an outdoor life enthusiast and strongly opposed
to technical encroachments in the area. In the cost-benefit analysis, how-
ever, the answer would be interpreted as if the person were little con-
cerned about protecting the area.
Thirdly, if the establishment or protection of collective goods like, for
example, a publicly accessible outdoor recreational area or clean air is to
be left to the willingness-to-pay of individuals, each individual is free to
under-communicate their real preferences regarding these goods in the
hope that others will pay for them.50 On the other hand, it is non-
committal for people to state a high willingness to pay. Hardly anyone will
be made economically responsible for having stated in a questionnaire that
she is willing to pay twice as much to finance an environmental good as
the person would have accepted in a real-life subscription. According to
some economists and psychologists, what people state their willingness to
pay for in such studies is a good conscience, not their own valuation of the
good itself.51
Fourth, in a willingness-to-pay investigation, affluent people are much
more likely to show high willingness to pay than those who can afford less
(provided that the answers about willingness to pay are honest, cf. above).

49 Lohmann, Whose Interest (Note 8, above).

50 Klosterman, Arguments (Note 18, above).
51 D. Kahneman and J. Knetsch, ‘Valuing public goods: the purchase of moral sat-

isfaction’, Journal of Environmental Economics and Management, vol. 22, 1992, pp. 57-70.
50 petter næss

It is a fallacy to believe that the sum of the amounts of money that indi-
viduals are willing to pay is a good measure of society’s total valuation of
a good. If one dollar more or less is of greater importance to some people
than to others, aggregated willingness-to-pay is, according to mainstream
economic theory, not a precise measure of aggregated utility.52
Fifth, a side effect of a project (e.g. increased possibilities for commer-
cial or industrial development along a new highway) may sometimes be
beneficial to some people while disadvantageous to other people. Since
willingness-to-pay investigations usually do not record negative willingness-
to-pay, but ask only how much people are willing to pay (with zero as the
lowest value), a willingness-to-pay investigation in connection with a pre-
sumptive positive side effect will only record the valuation among those
respondents who have positive or indifferent preferences in relation to the
side effect in question.53 (Correspondingly, for presumptive negative side
effects, the willingness-to-pay investigation will only record positive or zero
willingness-to-pay for the environmental good affected negatively by the
project, while those who possibly have an opposite valuation do not count.)54
To base decisions about environmental goods on willingness-to-pay inves-
tigations is problematic in relation to democratic ideals. This will be
addressed in the next section, where ethical and political aspects of the
use of cost-benefit analyses of transportation investment projects will be
discussed.
Most of the above-mentioned validity problems of willingness-to pay
investigations can be traced back to two of the false ontological assump-
tions mentioned in the previous section, viz. the reduction of social struc-
tures and social value to aggregated individual preferences, and the false
assumptions about human nature (a single-minded homo oeconomicus). This
illustrates the point made by Danermark et al. on the need to anchor the

52 M. Nyborg, Miljø og nytte-kostnadsanalyse. Noen prinsipielle vurderinger, Report 5/2002,

Oslo: Frisch-senteret, 2002.

53 Recording negative preferences would require a separate investigation of the will-

ingness to pay in order to avoid the negative side effects of the side effect originally
held to be positive.
54 The latter example may appear somewhat hypothetical, but might occur, for exam-

ple, in connection with the demolition of old, historically interesting buildings. Some
people might be less concerned about cultural heritage and consider it an advantage to
have such buildings replaced with new, modern buildings.
 cost-benefit analyses of transportation investments 51

chosen research methods in an adequate meta-theoretical understanding.55

Without this, there is a risk that the research methods will be used in a
less than fruitful way—and sometimes even directly erroneously.

5. Ethical/Political Assumptions of Cost-Benefit Analysis

Above, I have argued that cost-benefit analyses as practised in the trans-
portation sector are based on a number of untenable ontological and epis-
temological assumptions. These problematic assumptions imply that the
analyses are biased in several ways. These biases have, in their turn, cer-
tain social and environmental effects which involve ethical and political
values. Accepting the ontological and epistemological assumptions of cost-
benefit analysis thus involves an implicit acceptance of the ethical and polit-
ical values favoured by these assumptions. A critical discussion of the
practice of cost-benefit analysis on the transportation sector should there-
fore also uncover these values. I consider such a normative critique to be
an important part of a critical realist discussion of the method.56 Uncovering
the ethical and political values served by the method may also help in
understanding why cost-benefit analyses are used to such a great extent in
spite of their considerable validity problems.57
Fundamentally, cost-benefit analysis of transportation investment proj-
ects serves to legitimise a market-based development within a sector where
the adopted political goals imply that the hitherto dominating, demand-
led ‘predict and provide’ approach should be replaced—or a least supple-
mented to a greater extent than currently—by strategies meeting accessibility
needs in other ways than through facilitating more traffic. More generally,
cost-benefit analysis promotes a deregulatory agenda under the cover of
scientific objectivity.58 Several researchers have pointed to the fact that
transportation is a sector where deviations between official goals and imple-
mented measures are particularly high.59 To a great extent, the official

55 B. Danermark, M. Ekström, M. Jacobsen and J. C. Karlsson, Att förklara samhället,

Lund: Sudentlitteratur,1997, p. 258.

56 Cf., e.g., Sayer, Realism (Note 41, above).

57 Following Bhaskar and Danermark, Metatheory (Note 5 above), the latter would be

a critique at the level of normative kinds of mechanisms, types of context and charac-
teristic effects.
58 Ackerman and Heinzerling, Priceless (Note 3, above).

59 B. Flyvbjerg, Rationality and Power: Democracy in Practice, Chicago: University of Chicago

Press, 1998; A. Strand, ‘Evaluering av TP10-arbeidet: Miljøalternativene var best, men

52 petter næss

goals reflect the need to obtain a more environmentally sound and acci-
dent-preventing development in the transportation sector. Such needs are
not captured in an adequate way in cost-benefit analyses: long-term envi-
ronmental consequences are systematically and dramatically underestimated,
and willingness-to-pay investigations are hardly able to account for the
multi-faceted social value of environmental qualities such as climatic con-
ditions, clean air, or the combined function of a particular area as a beau-
tiful landscape, an outdoor recreation area, a rich ecosystem, and the
habitat of particular species (some of which may be rare or threatened by
extinction). Neglect of the difference between demand and societal needs
may be part of the reason why ‘decisions on transport policy is a field
where a substantial gap between intentions and realities is common both
at national and local level’.60 The political consequence of this are devel-
opments within the transportation sector whereby increasing mobility, in
particular by car and airplane, is given priority at the cost of environ-
mental concerns.
Moreover, cost-benefit analysis is blind to the consequences of the proj-
ects in terms of the distribution of benefits and burdens between different
population groups. Ackerman and Heinzerling show a grotesque example
of this from a cost-benefit analysis of proposed restrictions on the use of
cell phones while driving. Researchers have found that people who are
talking on cellular phones while diving are four times more likely to get
into car accidents than people who are not. However, some of USA’s most
influential economists have concluded that restrictions on the use of cell
phones while driving would be a bad idea, because the people who are
talking while driving are willing to pay a lot to talk on the phone—more
than many people who face deadly risk are willing to pay to avoid the
risk of being killed.61
While democratic majority decisions are based on the principle of
one vote per person, the number of ‘votes’ available for each person to
influence which alternative will obtain the highest willingness-to-pay are

vegalternativene vant!’, Samferdsel, no. 8, 1992; S. L. Larsen, T. Lerstang, P. K. Mydske,

P. G. Røe, T. Solheim, M. Senstadvold and A. Strand, TP10—en miljøvernpolitisk snuop-
erasjon? TP10 som prosess: hvilke forhold lokalt og sentralt har vært bestemmende med hensyn til
organisering, prosess og virkemåte, Samarbeidsrapport 3/93, Oslo: Norsk institutt for by- og
regionforskning, 1993.
60 Ministry of the Environment, Norway, Miljøbyrapport. Vedlegg a Oslo: Ministry of

the Environment, 2001.

61 Ackerman and Heinzerling, Priceless (Note 3, above).
 cost-benefit analyses of transportation investments 53

determined by their level of affluence. If a majority prefers alternative A

to alternative B, but the minority has a high willingness to pay for alternative
B, the latter project may well be the one which obtains the highest total
willingness-to-pay.62
Generally, market mechanisms are not able to secure a socially accept-
able distribution of burdens and benefits. Admittedly, according to main-
stream economic theory, perfect market competition will distribute resources
in such a way that they cannot be redistributed to the benefit of some
individuals without at the same time reducing the benefits of others (Pareto
optimality). However, the markets of the real world are far from perfect,
as there are a number of deviations (e.g. the existence of public goods,
externalities and not-fully-informed consumers) from the ideal precondi-
tions of a perfect market. Moreover, even if the market had fulfilled these
preconditions, a Pareto optimal situation warrants neither that the original
nor the final distribution of resources is optimal. If, for example, there is
a great extent of agreement in society that the inhabitants in all parts of
a region—including persons without a car and with a low income—should
be secured basic transport opportunities, a public transportation service
with affordable fares will be required even in the local communities where
pure market considerations would conclude that the population base was
too low. Such a service would typically be justified on the ground that a
need for public transport existed among the affected population groups. If
the need for such a service is defined as identical to willingness to pay,
this need would not be recorded and acknowledged. Considering the sum
of the individual’s willingness to pay as an adequate indicator of the need
among the population thus implies an underestimation of needs among
low-income groups. Conversely, negative impacts on parties without the
ability to pay are considered insignificant.63
Moreover, due to its complexity and lack of transparency, the calculations
of cost-benefit analyses will often be impossible to penetrate for people
other than a narrow group of experts. Lack of transparency refers both to
assumptions inherent in the valuation of different types of impacts, the way
of discounting, and (especially) the forecasts of traffic consequences of proj-
ects. As mentioned above, these assumptions are metatheoretically highly

62 Nyborg, Miljø (Note 52, above).

63 Apart from the world’s poor, the latter group also includes non-human nature—
cost-benefit analysis is based on a strictly anthropocentric view of the relationship between
humans and nature.
54 petter næss

problematic and lead to a bias reducing the likelihood of reaching envi-

ronmental and social political objectives within the transportation sector.
For politicians, the public at large as well as experts within other fields,
lack of transparency makes it difficult or impossible to evaluate the results
and find a basis for critique and development of alternative solutions.64
The cost-benefit analysis, and in particular the transport model producing
some of its key input data, are ‘black boxes’ with a content that it is not
considered necessary to take into consideration. This is aggravated by the
widespread practice of presenting the conclusions of the analysis at a highly
aggregated level, concentrating on only a few main indicators like the
benefit-cost ratio or the net present value, while omitting the more nuanced
and detailed information collected, let alone the uncertainty associated with
the various estimates. According to Maarten Hajer, ‘black boxing’ (i.e.
enclosing technical analyses and their presuppositions inside a ‘black box’)
is ‘maybe one of the most basic discursive mechanisms. Making things
appear as fixed, natural and absolute conditions is the most effective mean
to avoid potential oppositional forces’.65
Although sophisticated demand models seem objective and hard to manip-
ulate, it is technically easy to tune the models in ways so that ‘plausible’
or ‘desirable’ results are achieved.66 The same can be said about the cost-
benefit analysis as a whole. Any deliberate manipulation is of course not
something that the cost-benefit method in itself can be blamed for. Yet,
lack of transparency (‘black boxing’) makes it more difficult to discover
manipulation, and use of methods of analysis that only a narrow group of
experts have the opportunity to check creates a ‘protective veil’ behind
which those who wish to manipulate may more easily play their game.
Above, a number of problems associated with the quantification and
economic valuation of important impact categories of transportation invest-
ment project were discussed. Among cost-benefit analysts, two different
ways of reacting to this are common: reducing the analysis to impact cat-
egories that most easily lend themselves to economic valuation, or refining

64 T. Sager, Communicate or Calculate: Planning Theory and Social Science Concepts in a

Contingency Perspective, Stockholm: Nordplan, 1990; Lohmann, Whose Interest (Note 8, above).
65 M. A. Hajer, The Politics of Environmental Discourse: Ecological Modernisation and the Policy

Process, Oxford: Oxford University Press, 1995, p. 272.

66 M. Wachs, ‘When planners lie with numbers’, APA Journal, vol. 55, no. 4, 1989,

pp. 476-79; B. Flyvbjerg, N Bruzelius and W. Rothengatter, Megaprojects and Risk: An

Anatomy of Ambition, Cambridge, Cambridge University Press, 2003.
 cost-benefit analyses of transportation investments 55

the assessment and valuation methods in order to measure ‘intangible’

impacts in a better way. However, since the low validity and reliability of
the present ways of trying to put price tags on these impacts is rooted in
ontological misconceptions, improving the surveying methods, for example,
in willingness-to-pay investigations can hardly do more than marginal
improvements. (It would be like correcting the wrong digits to the right of
the comma while leaving the digits to the left of the comma uncorrected.)
Reducing the analysis to the ontologically and epistemologically least prob-
lematic categories therefore seems to be a better option. But what will then
be left of the analysis? As mentioned above, time savings often make up
the main benefit category in cost-benefit analyses of transport infrastructure
projects. However, the assessment of the magnitude of these savings as well
as the valuation of these savings in monetary terms is highly uncertain and
controversial. In some cases (e.g. bypass roads) safety improvements are
also important aspects of the motivation for the projects. However, as men-
tioned earlier, uncertainty and controversy is attached both to the assess-
ment of the influence of the project on the number of deaths, injuries and
other damages, and to the economic valuation of these impacts. On the
cost side, the mere construction cost should, in principle, be possible to
estimate relatively accurately (although what happens in practice is often
serious underestimation of costs).67 However, since the economic valuation
of nearly all the other items of the analysis, both on the benefit and cost
side, is encumbered with uncertainty and contestation, it is difficult to avoid
the conclusion that the entire method of cost-benefit analysis is inappropriate
for the evaluation and comparison of transportation investment projects.

6. What Should Be Done Instead?

Developing a critique of a phenomenon is quite pointless unless we are
able to identify at least the outline of a desirable and feasible alternative.
Without any constructive alternative, the critique will be ‘like criticising
the power of gravity’.68 When presented with one or more of the points
of criticisms raised in this article, proponents of cost-benefit analysis often
point to the necessity of having some sort of criteria for choosing between
alternatives of action and deciding whether or not to implement a proposed

67Cf. Flyvbjerg et al., ibid.

68A. Sayer, Alternatives and Counterfactuals in Critical Social Science, paper for the Thirteenth
World Congress of Sociology, Bielefeld, July 1994, p. 7.
56 petter næss

project. I agree with this, and also with the claim of defenders of the
method that the criteria should be made explicit as much as possible instead
of being concealed in the minds of the individual planners and decision-
makers. However, cost-benefit analysis is not the only way to systematically
evaluate and rank projects.
Instead, an impact analysis should be carried out, with assessment of all
relevant impacts that might be anticipated of the project (including intended,
non-intended, direct as well as indirect effects). According to Bhaskar and
Danermark, every social event must be understood in terms of (a) mate-
rial transactions with nature, (b) social interaction between agents, (c) social
structure proper, and (d) the stratification of embodied personalities of
agents.69 All these categories of impacts should be taken into consideration.
Impacts along the first dimension typically include consequences to the
physical environment, whereas unwholesome noise and air pollution are
examples of impacts belonging to the fourth dimension.70 Ramifications in
terms of changing location of activities and changing travelling patterns
are examples of impacts within category (b). Relevant impacts on the social
structure proper (category [c]) might include consequences to the devel-
opment of the overall transportation system at a regional or national scale,
including any path dependencies.71
The impact categories of the analysis should correspond to a goal hier-
archy reflecting the multitude of needs affected by the project (including
both ‘project-triggering’ needs and needs related to side effects of the project,
and needs among different population groups). The relevant consequence
categories should be identified on the basis of a broad hearing among
experts from different disciplines and representatives of different stakeholder

69 Bhaskar and Danermark, Metatheory (Note 5, above).

70 The stratified personalities of agents may also be influenced by the way the plan-
ning and decision-making process is organised, for example, in terms of effects on their
self-esteem, values, opportunities for personal growth and their motivation for partici-
pating in democratic processes. See, e.g., J. Forester, ‘Critical theory and planning prac-
tice’, Journal of American Planning Association, vol. 46, 1980, pp. 275-86. From a democratic
and humanist perspective, this implies a call for methods of impact analysis where stake-
holder groups and the population at large are actively invited to contribute to the
identification of the categories of impacts to be addressed in the analysis.
71 Cf. also I. Røpke, ‘Hvor mange “lag”?—miljøkonsekvenser i strukturelt pers-

pektiv’ (How many ‘layers’? Environmental impacts in a structural perspective), in

B. Kullinger, and U. B. Strömberg, eds, Planera för en bärkraftig utveckling, pp. 133-44,
Stockholm: Byggforskningsrådet.
 cost-benefit analyses of transportation investments 57

and population groups. The scope of the impact analysis must not be
reduced to the most easily measurable effects or those impacts that may
most readily be valorised in economic terms. The various effects should
be dealt with at a level of measurement appropriate for the specific effect.
The documentation material of the impact analysis should in an easily
accessible way give an account of the criteria, assumptions, weighting and
uncertainties on which the conclusions are based.
Seen in the light of the inevitable and considerable uncertainty associ-
ated with the results of transport modelling analyses (and the substantial
costs such modelling work represents in many planning processes), the pos-
sibility of replacing transport model computations with more qualitative
assessments of factors that may contribute to increase or reduce traffic,
should be seriously considered. Anyhow, transport modelling should only
be included in the impact analysis if the built-in assumptions of the model
have been quality controlled by independent experts from subject fields
covering a wider range than that of the model makers.
Direct construction costs (including costs for land purchases, materials,
construction work, financing costs, etc.) should of course be estimated as
exactly as possible in monetary terms. The same applies to expenses as
compensation for demolished buildings. In order to counteract the risk of
too optimistic or biased cost estimates, the so-called reference class fore-
casting method,72 according to which the proposed project is compared to
the actual economic expenses of a class of similar projects, should prefer-
entially be used instead of the traditional ‘bottom-up’ way of calculating
costs. The economic valorising may also include benefit effects that may
reasonably be expressed in monetary terms (but as mentioned above, in
the case of transportation projects, key effect types like time savings may
not belong to this category). Some other effects (e.g. the conversion of
different types of natural areas, or the size of noise zones and number of
dwellings within these areas) may be quantified without monetarisation.
Effects that are difficult to measure should be made subject to more qual-
itative descriptions, or be assessed from a combination of qualitative and
quantitative indicators. The same applies to effects where what is meas-
ured has no unambiguous relationship with the values affected. In such
cases, too, a certain degree of quantification of the total effect may be pos-

72 Cf. B. Flyvbjerg, M. S. Holm and S. Buhl, ‘How (in)accurate are demand fore-

casts in public works projects? The case of transportation’, Journal of the American Planning
Association, vol. 71, 2005, pp. 131-46.
58 petter næss

sible, for example, by placing the effect on a certain level of a goal achieve-
ment scale.
All quantified effects (with or without price tags) should be stated with
error margins. These margins should not necessarily be based on calcula-
tions of the statistical probability of possible outcomes (which may some-
times not be possible to perform, or too laborious and costly), but could
be indicated as rough estimates where the latter is the only practical option.
Proponents of cost-benefit analysis may argue that the approach pro-
posed here does not provide decision-makers with the answer about whether
or not a project should be implemented, or which among several imple-
mentation-worthy projects should be given priority. However, the belief
that it is possible to give a technical answer to these questions is ‘trying
to do the impossible: to reconcile incommensurable values in one quanti-
tative objective figure [. . .] no amount of technical wizardry will succeed
in absolving us of the need to resolve [. . .] conflict through political
processes’.73
Some analytical tools for helping stakeholders in their process of syn-
thesising disaggregated information to an overall assessment of each alter-
native, and a ranking of alternatives, may still exist. However, because of
the considerable disagreement in society on the importance of different
ethical values, such techniques are only suitable for consensus-building
within relatively homogeneous groups (such as an environmentalist group,
an organisation of local residents, or the chamber of commerce), and not
for arriving at conclusions about the ‘optimal’ solution for society as a
whole. Edwards and Newman’s Multi-Attribute Utility Technology (MAUT)
is a relevant example of a method for synthesising non-monetarised dimen-
sions of project alternatives.74 The method implies that a goal hierarchy is
established, based on inputs from various stakeholder groups. The goal
hierarchy should include all the concerns mentioned by any of the par-
ticipating stakeholder groups, and would thus be a ‘gross’ compilation of
goals. The various impacts are indicated by experts according to a goal
achievement scale (e.g. a five-level scale) for each dimension (operationalised
goal). Different stakeholders are then given the opportunity to express how

73 S. H. Hanke and R. A. Walker, ‘Benefit-cost analysis reconsidered: an evaluation

of the mid-state project’, Water Resources Research, vol. 10, no. 5, 1974, pp. 898-908, cited
in Lohmann, Whose Interest, p. 6 (Note 8, above).
74 W. Edwards and J. R. Newman, Multiattribute Evaluation, Thousand Oaks: Sage

Publications, 1982.
 cost-benefit analyses of transportation investments 59

the different dimensions should be weighted against each other. Separate

processes of evaluation and discussion of the alternatives should be carried
out among each group of stakeholders, based on the importance attached
by each group to different impact categories. (Goals/consequence categories
considered by a stakeholder group to be of no importance at all would be
given a zero weight in the analysis based on the prioritisations of this stake-
holder group.) The total, weighted score of each alternative for non-
monetarised dimensions is compared to each alternative’s economic score
(construction costs and possible other items that may be valorised in an
appropriate way). This sets up a basis for a discussion among the stake-
holder group representatives of the mutual ranking of the alternatives.
Even for processes within relatively homogeneous groups, the synthesis-
ing method would only be suitable as an interactive tool for arriving at a
more reasoned opinion of the merits of the alternative projects, and not
for arriving at firm conclusions. It should also be noted that the group
processes required are time-consuming. On the other hand, such processes
could contribute to more vivid stakeholder participation in the planning
and decision-making process. The costs of carrying out the processes should
also be compared to the considerable costs associated with obtaining the
data material required for a cost-benefit analysis.
Whether or not a synthesising method like the one outlined above is
used, uncertainty analyses and sensitivity analyses should be conducted in
order to give indications about the ways the project alternatives will per-
form under different conditions and for differently affected groups. Based
on such analyses, strategies should be developed, focusing on how different
situations—in particular the most unfavourable ones—could be met.

7. Concluding Remarks
The practice of cost-benefit analysis within the transportation sector is based
on a number of untenable ontological and epistemological assumptions:
– social value is reduced to the aggregated preferences of individual con-
sumers
– humans are considered as fully informed, utility-maximising consumers
– transport modelling analyses based on a misleading ‘local ontology’ among
the model makers make up main input data of the analyses
– nature is dealt with as if it were a mere epiphenomenon of the human
world and is assumed to have unlimited capacity to bear growth in con-
sumption and production
60 petter næss

– precise, quantitative predictions of travel behaviour and other impacts

are considered possible in spite of the necessary crudeness of such forecasts
– willingness-to-pay investigations are encumbered with serious validity and
reliability problems, many of which are rooted in ontological miscon-
ceptions of social value and human nature
Cost-benefit analyses of transportation investment projects tend to neglect
long-term environmental consequences and needs among population groups
with a low ability to pay. Fundamentally, they serve to legitimate a mar-
ket-based development within a sector where adopted political goals imply
that the hitherto dominating ‘predict and provide’ approach should be
replaced by strategies meeting accessibility needs in other ways than through
facilitating more traffic.
Instead of cost-benefit analyses, impact analyses evaluating the likely
effects of project alternatives against a wide range of societal goals is rec-
ommended. In such analyses, quantification and economic valorisation
should take place only for impact categories where this can be done in an
ontologically and epistemologically defensible way. Uncertainty and the pre-
suppositions on which the analysis rests must be made visible in the pres-
entation of the results.