Research in Transportation Business & Management 6 (2013) 19–30

Contents lists available at SciVerse ScienceDirect

Research in Transportation Business & Management

Efficiency, profitability and welfare gains from the Canadian National

Railway privatization☆
Anthony E. Boardman a,⁎, Claude Laurin b, Mark A. Moore c, Aidan R. Vining c
a
Sauder School of Business, University of British Columbia, 2053 Main Mall, Vancouver, BC, Canada V6T 1Z2
b
HEC Montréal, 3000 Chemin de la Côte-Ste-Catherine, Montréal, Québec, Canada H3T 2A7
c
Beedie School of Business, Simon Fraser University, 500 Granville Street, Vancouver, BC, Canada V6C 1W6

a r t i c l e i n f o a b s t r a c t

Article history: This article describes and analyzes the privatization of Canadian National Railway (CN), a large railroad
Received 19 June 2012 privatization. First, it reviews the theory and evidence concerning railroad privatizations. Second, it presents
Received in revised form 21 November 2012 a brief history of CN and the regulatory environment prior to and after CN's privatization. Third, it uses data
Accepted 22 November 2012
from 1990 to 2011 to compare CN's post-privatization operating performance with its pre-privatization
Available online 23 December 2012
performance. Fourth, it uses cost–benefit analysis to estimate the social welfare gains from the privatization
Keywords:
and the distribution of those gains. The overall results demonstrate that CN performed substantially better
Privatization following privatization, both from an operational perspective and from a broader social welfare perspective.
Canadian National Railway We find statistically significant increases over the long term (16 years following privatization) in sales, capital
Operational performance investment, assets, profit, profitability, productivity, dividends and corporate taxes paid. There was little
Social welfare change in the capital structure of CN and a significant decrease in employment. Using Canadian Pacific
Railway as a basis for the counterfactual, we estimate that CN's privatization generated social welfare gains
of approximately $25 billion in 2011 Canadian dollars. The Canadian government received almost half of
these gains, while CN's shareholders (most of whom were non-Canadian) captured the rest.
© 2012 Elsevier Ltd. All rights reserved.

1. Introduction an existing organization, while the British and Japanese privatizations

involved either separation of infrastructure and rolling stock (vertical
The privatization of Canadian National Railway (CN) in 1995 was separation) or geographic disintegration (horizontal separation). Third,
by far the largest transportation privatization in Canadian history. the topographical footprint and conditions of the CN system are quite
Alongside the privatizations of the Japanese National Railways and different to the other two systems. Fourth, both before and after pri-
of the British Rail, the CN privatization represents one of the largest vatization, CN faced a direct competitor over much of its network,
transportation privatizations globally. Both the Japanese and British which was not the case in Japan or Britain.
railroad privatizations have been controversial, especially in the Our main purpose is to determine whether the privatization of
British Rail case (Mathieu, 2003; McCartney & Stittle, 2008; Smith, CN was beneficial. We take two approaches. First, we use data from
2006; Yvrande-Billon & Ménard, 2005). As we discuss later, the evi- 1990 to 2011 to compare CN's post-privatization operating perfor-
dence on the performance outcomes of the railroad privatizations is mance with its pre-privatization operating performance. We examine
mixed, although freight privatization appears to have done better than changes in output, capital expenditures, assets, employment, profit,
other privatized railroad businesses, including passenger services. productivity, profitability, capital structure, dividends and corporate
The privatization of CN differs from the Japanese and British taxes. The findings are important from a business strategy perspective,
railroad privatizations in a number of important ways. First, the CN a shareholder perspective and a government perspective. Second, we
privatization primarily involved freight transport rather than passen- use data from the 1981–2008 period to estimate the change in social
ger travel, which is the case in both Japan and Britain. Second, the CN welfare from the CN privatization and the distribution of these bene-
privatization consisted of a one-time share issue sale of an integrated fits and costs between the Canadian government and (Canadian and
business (including track and rolling stock) and the maintenance of non-Canadian) shareholders of CN. We use cost–benefit analysis
(CBA) to conduct this assessment. A key feature of this CBA is that
we use cost data from Canadian Pacific Railway (CP), a direct com-
☆ The authors would like to thank Leo Fankhänel and Robert Boardman for research petitor of CN, to compute the counterfactual, that is, what would have
assistance.
happened in the absence of privatization. These findings are most
⁎ Corresponding author. Tel.: +1 604 822 8484; fax: +1 604 822 8477.
E-mail addresses: [email protected] (A.E. Boardman), claude. directly relevant for the government of Canada and policy analysts in-
[email protected] (C. Laurin), [email protected] (M.A. Moore), [email protected] (A.R. Vining). terested in the welfare consequences of privatization in general and,

2210-5395/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.rtbm.2012.11.011
20 A.E. Boardman et al. / Research in Transportation Business & Management 6 (2013) 19–30

more specifically, of railroad privatization. It also offers lessons to other North America, that this approach has been able to deliver a degree
governments considering railroad privatization or institutional reform. of competition with relatively little regulatory intervention” (OECD
In sum, we find that the privatization of CN was beneficial to Secretariat, 2006, 71). Evidence suggests that the presence of more
shareholders, to the government of Canada and to the overall welfare than one railroad competitor reduces tariffs anywhere between 3%
of Canadians. The most important results can be summarized as and 25% (Gomez-Ibanez, 2010; Grimm & Winston, 2000; Karikari,
follows. In the short run (which we consider to be the 5-year period Brown, & Nadji, 2006; Winston, Maheshri, & Dennis, 2011). Further-
following privatization), profit (net income), profitability (return on more, both CN and CP faced increasing inter-modal competition from
assets and return on sales), and productivity (measured by sales trucking and shipping. In an important article, Caves and Christensen
per employee and net income per employee) increased significantly (1980) refer to Canadian railroads as operating in a competitive envi-
relative to the (5-year) pre-privatization period, employment experi- ronment. Although the Canadian railroad industry was a duopoly
enced a statistically significant drop and there was no statistically on the long haul freight routes, this collective evidence suggests that
significant change in sales revenue, capital investment, assets, capital it was “reasonably competitive” and, consequently, one might expect
structure, dividends or corporate taxes. In the long run (which we that the privatization of CN would lead to efficiency improvements
consider to be the 16-year period following privatization), there and welfare gains.
were statistically significant increases in sales, capital investment, The evidence on previous rail privatizations has been mixed.
assets, profit, profitability (return on assets and return on sales), Only in Japan is the evidence clearly positive: privatization increased
and productivity (sales per employee and net income per employee). efficiency and profitability, especially for freight rail (Mizutani,
There were also statistically significant increases in dividends and 1999; Mizutani & Nakamura, 1996, 1997; Thompson, 2003). The
corporate taxes, demonstrating that the most significant stakeholders privatization also improved safety (Evans, 2010; East Japan Railway
in the privatization – government and shareholders – benefited from Company Management Planning Department, 2008) and travel
the privatization. Capital structure (debt-to-assets) did not change times (East Japan Railway Company Management Planning Department,
substantially. There was, however, a significant reduction in employ- 2008).
ment, although we argue that employees were not adversely affected. In Britain, the evidence is much less positive. The restructuring
Turning to the welfare results, CN's privatization generated social and privatization of British Rail (BR) took place during 1993–1996,
welfare gains of almost $25 billion in 2011 Canadian dollars (which when John Major's conservative government was in power. Thompson
we use hereafter unless explicitly stated otherwise). The Canadian (2003, 347) argues that it “has been the most contentious of all railway
government received almost half of these gains, while CN share- system restructuring efforts”. There was significant political and eco-
holders (over half of whom were non-Canadian) captured the rest. nomic opposition to BR's privatization and significant disagreement as
The article is organized as follows. Section 2 briefly reviews the to how it should be privatized. Eventually, BR was fully privatized and
theory about privatization and summarizes the academic evidence resulted in over 100 different private firms. Many of these privatiza-
concerning the outcomes of previous railroad privatizations in Japan, tions, including that of the Railtrack, which owned all of the track
Britain, Australia, New Zealand and Chile. Section 3 presents a brief and the stations, were carried out rapidly (some would argue hasti-
history of CN and of the regulation of Canadian railroads prior to pri- ly) towards the end of Major's term.
vatization. It also describes some key events following privatization. Pollitt and Smith (2002) found evidence of significant operating
Section 4 analyzes both the short-run and the long-run operational cost savings in the first few years following privatization. Further-
performance impacts of privatization. Section 5 presents our estimate more, Cowie (2009) found productivity gains in passenger rail on
of the overall welfare change attributable to the privatization of CN by the order of 3–4% per annum over the first 4 years of privatization.
using cost–benefit analysis, and of the distribution of these changes In addition, there was substantial growth in passenger and freight
among different stakeholders. Section 6 summarizes the main results traffic. However, there have been many criticisms of this privatiza-
and identifies some of the factors that led to CN's success. It also offers tion, including accusations of lower quality service (perceived
lessons for private sector and government management, and includes reduced punctuality and overcrowding), a worse price-to-quality
some suggestions for future research. ratio and safety concerns (Mathieu, 2003), even though the evidence
suggests that safety actually improved (Evans, 2007; Thompson,
2. Theory and evidence about railroad privatizations 2003). Following a huge cost escalation for the upgrade to the West
Coast Main Line and the Hatfield derailment in 2000, the government
The weight of the empirical evidence is that privatization of decided to place Railtrack into administration and eventually
businesses in reasonably competitive markets increases firm com- replaced it with Network Rail, a not-for-profit company, owned by
petitiveness, improves productivity and generates positive welfare its members (Crompton & Jupe, 2007). Subsequently, annual industry
gains (Boardman, Laurin, & Vining, 2002; Boardman & Vining, 1989; cash costs rose by 47% and unit costs rose by 40%. Smith (2006)
Boubakri & Cosset, 1998; Megginson & Netter, 2001; Shirley & Walsh, argues that this cost increase was largely due to an “excessive” concern
2001). However, the long-haul Canadian railroad industry has been with safety.
and is effectively duopolistic. While the direct overlap of CN's and CP's Under the provisions of the Railways Act 1993, BR freight was split
networks has been estimated at between 25% and 50%, the effective into seven companies, each of which was separately offered for private
competitive overlap, given feed-in complementors that have options, sale. In the end, however, five were sold to a single purchaser, English,
is considerably larger (McNish, Jang, & Silcoff, 2012). Unfortunately, Welsh and Scottish Railway (EWS), controlled by Wisconsin Central.
the theory and evidence concerning the impact of privatization in Following privatization, freight traffic grew 42% between 1994 and
highly oligopolistic or duopolistic markets is not resolved (Chirwa, 2000 (Mathieu, 2003; Thompson, 2003). However, operators have
2004; De Fraja, 1991; Willner & Parker, 2007). only been marginally profitable (Fowkes & Nash, 2004). Furthermore,
Both Canada and the United States have been unusual in enjoying a despite mandating open access to freight companies, which was an im-
fairly high degree of direct competition in rail freight. The Secretariat portant feature of the British restructuring, there has been very little
of the OECD has noted: “competition in-the-market between new entry into the rail freight business (Cowie, 2010).
vertically-integrated rail companies …requires the existence of at In Australia there has also been extensive privatization of parts of
least two separate rail infrastructures capable of providing substitute the railroad system. Here, the record appears mixed. Williams, Greig,
rail services …this is the predominant form of competition in rail and Wallis (2005) note that the privatization of freight railroads has
freight services in North America” (OECD Secretariat, 2006, 71). Fur- allowed consolidation across state boundaries and argue that, as a
thermore, the Secretariat argues: “experience shows, at least in result, the industry is markedly stronger than in the past, although
 A.E. Boardman et al. / Research in Transportation Business & Management 6 (2013) 19–30 21

governments remain the funders of last resort. They also maintain relatively random, but politically salient, factors can affect whether or
that long-distance passenger privatization appears to have been suc- not the privatization is sustained (especially by governments elected
cessful and consider the urban and rural passenger rail concessions subsequent to the privatizing government). For example, a single
a “qualified success”. train crash that kills 10 people has greater political saliency than 10
New Zealand Rail Limited (later re-named Tranz Rail) was road crashes that each kills one person.
privatized as a single integrated entity. In this case, the privatization Our primary purposes are to consider whether the privatization
appears to have been a failure on most dimensions (Clark, 2010; of CN has improved operating performance or social welfare. The
Wilson, 2010). New Zealand Rail was purchased in 1993 by a consor- net effect of the interplay of all the above factors over time in any
tium of financiers and Wisconsin Central who jointly put up about particular privatization is impossible to predict ex ante. The answer
one-third of the money and financed the rest by debt. In 1994, New lies in the empirical pudding.
Zealand Rail sold its stake in Clear Communications and, the next
year, it made a capital repayment to the consortium members of 3. Overview of CN and the Canadian freight rail
$100 million (current Australian dollars), only slightly less than the regulatory environment
initial amount of cash they had invested. In 1996, Tranz Rail issued
new shares to the public, equivalent to 25% of the company, and it CN was formed as a government-owned corporation through the
was listed on the New Zealand stock exchange. Despite significant amalgamation of several financially troubled, privately owned rail-
investments in freight operations, the required volumes did not ma- roads between 1917 and 1923. Since then, CN and CP have dominated
terialize and Tranz Rail lost money. In 2000, a new Board replaced railroad freight services in Canada. During much of the period of
the former management team and sold off many of the non-core public ownership, CN was in various forms of financial stress. It is
assets. However, Tranz Rail continued to perform poorly and the very difficult to determine the size of direct federal assistance to CN.
stock price dropped. In 2003, Toll Holdings made a takeover offer Over the period 1927–1991, direct subsidy payments to the Canadian
for the company and, eventually, Toll became the operator and the railroad sector amounted to almost $12 billion (1995 Canadian
government acquired the rail network. The company was renamed dollars), although over half of this amount was to subsidize passenger
Toll NZ and its financial situation continued to worsen. In 2008 it services (Bonsor, 1995). Much of the “subsidy” to CN was by way of
was re-nationalized by the federal government. Although the Tranz recapitalizations: the Canadian government swapped CN's debt for
Rail privatization was a failure on most dimensions, it was highly equity or recapitalized it in 1937, 1952 and 1978.
profitable for the initial consortium investors. With the final recapitalization in 1978, the government imposed a
In Chile, railroad reforms resulted in the privatization of the entire hard budget constraint and mandated that CN should pay a 20% divi-
Northern Railroad (sold to FERRONOR) and a state-owned Southern dend on any profits earned. After 1978, CN appears to have operated
Railroad (EFE), which owned and maintained the track and carried more efficiently. Between 1978 and 1989, it was profitable in 9 out of
passengers, but allowed private freight concessionaires (FEPASA and the 11 years and remitted $237 million (current Canadian dollars) to
TRANSAP) to use the track (Soto, 2010). FERRONOR acquired the the federal government (Bruce, 1997, 15).
Northern Railroad in a competitive bid in 1995. It operated as an In the early 1970s CN was highly diversified. However, over time, it
unregulated monopoly, concentrating on more profitable segments has focused more on railroad freight transportation. Around 1977, Air
and eliminating others. From 1997 to 2000, freight volumes almost Canada was spun off as a separate Crown Corporation (state-owned
quintupled, revenues tripled, labor productivity tripled and profitabil- enterprise), CN's passenger services subsidiary became another sepa-
ity increased (see Soto, 2010, 16). On the Southern Railroad, FEPASA's rate Crown Corporation, later called Via Rail Canada, and CN Marine,
freight volumes, revenues and productivity did not increase immedi- which operated ferries in Atlantic Canada, also became a separate
ately following privatization, but did increase markedly after a decade Crown Corporation. CN sold CN Route (a trucking business) to private
of private ownership (see Soto, 2010, 18). Volumes rose partially investors in 1986. CNCP Telecommunications and a chain of hotels
due to the fact that prices were about 40% lower than prior to pri- were sold to CP in 1988, along with telecommunications divisions in
vatization (Thompson, Budin, & Estache, 2001), but despite this, prof- the Northwest Territories and Newfoundland. CN also owned oil and
itability remained low. At this time, it is hard to reach a definitive gas assets, non-rail real estate, and a subsidiary that manufactured
conclusion on rail privatization in Chile, although the results appear transport equipment. Most of these businesses were divested prior
positive in terms of freight. to privatization, and the manufacturing subsidiary was finally sold in
This review demonstrates that railroad privatization has been 1996 (Canadian National Railway Company, 1995, 1996).
conducted quite differently in different countries. Also, the outcomes The 1967 National Transportation Act significantly reduced the
have been very different, ranging from clearly beneficial to disastrous. scope of rate regulation over both CN and CP, with the exception of
In general, it appears that the evidence for freight privatization is grain transportation. But, Timur and Ponack (2002, 538) note that
considerably better than for the privatization of other railroad busi- the 1967 Act still “emphasized cooperation between CN and CP and
nesses, including passenger services. Apart from that result, given the …the two companies exchanged cost information and set common
relatively small number of railroad privatizations, and the diversity of freight rates.” In 1983, the Western Grain Transportation Act shifted
situations and outcomes, it is difficult to generalize about the key the burden of the so-called Crow Nest Pass rates for grain (essentially
success factors. However, the following factors appear to affect the price controls) from the railroads to direct government subsidies
efficiency and welfare outcomes of railroad privatizations: (1) how (Heaver & Waters, 2004). In 1987, the National Transportation Act
the privatization process was conducted, most importantly whether (NTA) allowed shippers located on only one of the railroads' lines
through a share issue or direct sale; (2) the regulatory framework greater access to the other's line, allowed confidential negotiation of
following privatization; (3) whether the railroad was privatized as an rates, encouraged reliance on market forces and arbitration rather
integrated network or whether there was separation of ‘wheel and than on the regulation of most non-grain rates, and it allowed both
rail’; (4) the market structure following privatization: the degree of di- railroads somewhat greater freedom to abandon uneconomic branch
rect competition, as well as the openness of the market and barriers to lines. The 1996 Canada Transport Act extended the 1987 NTA provi-
entry; (5) the price/quality characteristics of substitutes (especially sions and allowed both railroads to eliminate low-density lines,
road and sea); (6) the geography of the terrain over which the railroad resulting in the formation of more independent short-line railroads.
operates, and economies of scale and density; (7) the degree and nature More importantly, the 1987 National Transportation Act initiated
of technological change; (8) government policy following privatization, a transition from the era of tacit collusion to one of more direct
especially the extent of any government failures. Moreover, a myriad of intra-modal competition (Bonsor, 1995; Timur & Ponack, 2002).
22 A.E. Boardman et al. / Research in Transportation Business & Management 6 (2013) 19–30

And, as mentioned above, both CN and CP faced increasing inter- each variable during the 5-year (SR) post-privatization period and
modal competition from trucking and shipping. during the 16-year (LR) post-privatization period to the average
Paul Tellier was appointed CEO of CN in 1992 with the mandate to value in the pre-privatization period (denoted SR post–pre and LR
make CN more profitable and efficient prior to an ultimate privatiza- post–pre, respectively). Finally, we present the Mann-Whitney U statis-
tion. Tellier began to run CN on a more commercial basis (Canadian tic (also known as the Wilcoxon rank-sum test), which tests whether
National Railway Company, 1995; Bruce, 1997). The Canadian federal the post-privatization observations come from the same distribution
government privatized a number of Crown corporations between 1985 as the pre-privatization observations.
and 1997. Boardman and Vining (2012) provide a comprehensive In the following sub-sections we discuss changes in four major
discussion of the aggregate effects of these privatizations. The main operational performance categories: overview (sales, capital invest-
reasons for the federal privatizations were the desire for more revenue ment, assets and employment); profit, profitability and productivity
or to reduce subsidies in the face of large deficits and government debts, (net income, return on assets, return on sales, sales per employee
and because it was consistent with the conservative government's and net income per employee); capital structure (debt-to-assets
ideological preferences (Boardman, Laurin, & Vining, 2003). Unlike ratio); and payments to stakeholders (dividends and corporate taxes
many railroad privatizations in other countries, CN was privatized in paid). In total, these measures provide a comprehensive picture of
its entirety and the infrastructure was not separated from operations. the effects of privatization on operational performance.
All of the large Canadian privatizations, including the CN privatization,
were carried out through fixed-price share issue privatizations (SIPs). 4.1. Overview
These firms were in at least somewhat competitive industries and were
expected to improve their efficiency under private ownership. CN was CN's assets, measured in real 2011 Canadian dollars, were about
privatized in November 1995 in one tranche, with proceeds amounting $10 billion in the years prior to privatization. They fell by more than
to just over $2 billion in current Canadian dollars. Only the privatization 10% in the year of privatization, but soon rose rapidly. Assets increased
of Petro-Canada in 1991 raised more money for the Canadian government by almost $14 billion from 1997 to 2001, due partly to the Illinois Cen-
(about $9 billion in aggregate in current Canadian dollars). tral (1998) and Wisconsin Central (2001) acquisitions. Since 2001, as-
One issue of interest is whether the government left “money on sets have continued to increase, albeit at a slower rate. By 2011, assets
the table.” Laurin, Boardman, and Vining (2004) found that CN was were over 150% higher than during the 5-year pre-privatization period.
underpriced by about 25%. Based on a sample of 104 SIPs across 25 As shown in Table 1, the increase in assets from the 5 years prior to pri-
countries, they also found that SIP underpricing averaged 15.2% in vatization versus the 16 years after privatization was statistically signif-
other developed countries, but only 6.3% in Canada. This comparison icant at the 1% level.
suggests that the Canadian government did leave significant money CN's real sales did not increase as much as real assets during the
on the table with CN's privatization, unlike its other SIPs. post-privatization period. Like assets, they were relatively flat prior to
Since its privatization, CN has made a number of important acquisi- privatization. Following privatization they decreased slightly. They
tions that have substantially expanded its North American network. picked up in 1998 and then grew slowly through 2008. In 2011, real
Most significantly, it increased its presence in the U.S. with the acqui- sales were approximately 50% higher than during the 5-year pre-
sition of the Illinois Central Railroad in 1998 for $2.4 billion (current privatization period. Similar to assets, sales were not significantly
U.S. dollars). After this acquisition, Hunter Harrison of Illinois Central higher in the 5-year post privatization period than in the 5-year pre-
became CEO of CN. CN later acquired Wisconsin Central for $1.2 billion privatization period but they were significantly higher in the 16-year
(current U.S. dollars) in 2001, Great Lakes Transportation for post-privatization period than in the pre-privatization period.
$380 million (current U.S. dollars) in 2003 and BC Rail for Employment at CN was declining prior to privatization and con-
$1 billion (current Canadian dollars) in 2003. When CN acquired tinued to decline during the first 3 years following privatization. By
Wisconsin Central it also acquired interests in EWS Railways in the 1998, employment had been reduced by 1/3 compared to the average
UK, Tranz Rail in New Zealand, and TasRail in Tasmania. CN quickly of the 3 years prior to privatization. Thereafter, employment levels
sold off these holdings. remained stable until 2011. The decrease in employment following
As of 2012, the increased efficiency of CN has had an impact on the privatization was statistically significant for both the 5-year and
management and ownership of CP. In 2012, private equity share- 16-year post-privatization periods.
holders of CP (most importantly, Pershing Square) won a proxy battle
to have CP hire CN's retired CEO, Hunter Harrison, to run CP (Deveau, 4.2. Profit, profitability and productivity
2012; McNish et al., 2012).
CN experienced losses (negative net income) in most years imme-
4. Operational performance changes at CN diately prior to privatization. It took a major write-down (of over
$1 billion in constant dollars) in 1992, the year Paul Tellier was
This section adopts the methods used by D'Souza and Megginson appointed CEO and lost over $1 billion in 1995, the year of privatiza-
(1999), Boardman et al. (2002) and others to examine the impacts tion. Following privatization, CN's net income (NI) was always posi-
of the privatization on operational performance. One key difference tive and increased consistently, reaching over $2 billion in 2006 and
is that this research uses data for a much longer period of time — 2007 (in both real and nominal terms). Profits fell slightly in 2008
from 5 years prior to privatization to 16 years post-privatization. and 2009 during the recession, but in 2011 profits reverted to their
The main results are presented in Table 1. Data are drawn from the 2007 level. The differences in NI between both the 5-year and
annual reports, as reported by Compustat. All financial data are 16-year post-privatization periods, relative to the pre-privatization
converted to real (inflation adjusted) 2011 Canadian dollars using period, were both statistically significant.
the Canadian Consumer Price Index — All Items (CPI). In the 5-year post-privatization period, CN's return on assets (ROA)
Given that Table 1 contains the raw data for each variable, one can averaged 3.6% and its return on sales (ROS) averaged 8.2%, which were
clearly see how each performance measure has changed over time. substantially better than prior to privatization. Thereafter, ROA in-
Table 1 also contains the average value of each performance variable creased slightly while ROS continued to increase substantially. This
during the 5-year period prior to privatization (denoted Pre-Priv), long, consistent increase in profitability is particularly impressive
during the short-run (5-year) period after privatization (denoted SR compared to other Canadian SIPs, where profitability gains were usu-
post-priv) and during the long-run (16-year) period after privatiza- ally achieved in the first 3 years and did not increase thereafter
tion (denoted LR post-priv). It also compares the average value of (Boardman & Vining, 2012). Both ROA and ROS showed statistically
 A.E. Boardman et al. / Research in Transportation Business & Management 6 (2013) 19–30 23

Table 1
CN's pre-privatization and post-privatization operating performance, 1990–2011.

Year Sales Capital Assets Employment Net Return on Return Sales per Net income Debt to Dividends Corporate
(millions) expenditures (millions) income assets (%) on sales employee per employee assets (millions) taxes
(millions) (millions) (%) (thousands) (thousands) (%) (millions)

1990 6236 500 10,749 37,000 12 0.11 0.19 169 0 49.55 86 −3

1991 5827 346 10,085 35,300 −21 −0.20 −0.35 165 −1 49.30 0 −5
1992 5783 477 10,065 33,500 −1435 −14.26 −24.81 173 −43 64.67 111 −18
1993 5889 614 9953 33,000 −111 −1.11 −1.88 178 −3 66.06 0 18
1994 6476 803 10,925 30,770 343 3.14 5.29 210 11 65.98 0 17
1995 5609 446 8483 26,951 −1485 −17.51 −26.48 208 −55 62.79 0 −25
1996 5609 669 8411 24,060 192 2.28 3.41 233 8 61.83 92 15
1997 5772 765 9384 22,800 535 5.70 9.26 253 23 51.70 103 431
1998 5442 649 14,267 21,510 143 1.00 2.63 253 7 60.50 130 8
1999 6790 812 19,046 23,490 777 4.08 11.44 289 33 62.69 172 476
2000 6845 763 19,099 22,460 970 5.08 14.18 305 43 62.50 201 556
2001 6929 782 23,034 22,670 891 3.87 12.86 306 39 66.14 228 481
2002 7326 685 22,690 23,190 685 3.02 9.35 316 30 64.98 222 321
2003 6863 680 20,003 21,490 856 4.28 12.47 319 40 62.22 223 394
2004 7499 1228 22,069 22,470 1485 6.73 19.81 334 66 61.88 254 723
2005 8113 1322 24,863 21,540 1744 7.01 21.49 377 81 58.32 308 875
2006 8480 1426 26,380 21,810 2294 8.69 27.05 389 105 59.07 374 706
2007 8492 1491 25,227 22,700 2321 9.20 27.33 374 102 56.62 449 589
2008 8913 1496 28,078 22,230 1991 7.09 22.34 401 90 60.48 458 683
2009 7721 1469 26,386 21,500 1943 7.36 25.17 359 90 55.38 497 427
2010 8539 1632 25,942 22,300 2165 8.35 25.36 383 97 55.23 518 795
2011 9028 1625 26,026 23,200 2457 9.44 27.22 389 106 58.96 585 889
Pre-priv 6042 548 10,356 33,914 −242 −2.46 −4.31 179 −7 59.11 39 2
SR post-priv 6092 732 14,041 22,864 523 3.63 8.18 267 23 59.85 140 297
LR post-priv 7398 1093 21,306 22,464 1341 6 17 330 60 59.91 301 523
SR post–pre 49 184 3686 −11,050 766 6.09 12.50 88 30 0.74 100 295
SR Mann–Whitney U 10 4 10 0a 2b 2b 2b 0a 2b 10 5 4
LR post–pre 1355 546 10,951 −11,450 1583 8.29 21.27 151 67 0.80 261 521
LR Mann–Whitney U 15b 7a 10b 0a 2a 3a 2a 0a 2a 36 16c 4a

The figures for 1995 are in italics because CN was privatized in this year and these data are not included in the statistical analyses.
a
Statistically significant at the 0.01 level of significance for a two-sided alternative.
b
Statistically significant at the 0.05 level of significance for a two-sided alternative.
c
Statistically significant at the 0.10 level of significance for a two-sided alternative.

significant increases following privatization in both the shorter and debt-to-assets ratio has changed little, although it has been slightly
longer post-privatization periods. lower in recent years.
Although both sales and employment dropped in the first 5 years
following privatization, employment dropped relatively more, which 4.4. Dividends and corporate taxes paid
resulted in an increase in labor productivity following privatization,
measured in terms of real sales per employee. Thereafter, productivity For three of the years during the 5-year pre-privatization period, and
continued to increase. Prior to privatization, CN's sales per employee in the year of privatization, CN paid no dividend. It commenced paying
were about $180,000 (in real dollars). From 2005 to 2011, they were dividends in the first year after privatization and has increased (in
about $380,000 per employee. The increases in sales per employee nominal dollars) the dividends paid in each subsequent year except
were statistically significant in both the 5-year post-privatization 2002 when they dropped very slightly. In 2011, it paid dividends of
period and the 16-year post-privatization period relative to the pre- $585 million. The difference in dividends between the long-run
privatization period. Productivity, measured by NI per employee, ex- post-privatization period and the pre-privatization period was statisti-
perienced similar statistically significant increases. These results are cally significant (at the 0.10 level), but not between the short-run
consistent with Laurin and Bozec (2001) who found significant im- post-privatization period and the pre-privatization period.
provements in CN's total factor productivity following privatization. Similarly, in the pre-privatization period, CN paid virtually no
taxes. In fact, in some of these years, and the year of privatization,
4.3. Capital structure it received a tax rebate. In contrast, in the first 3 years following
privatization, CN paid taxes of $151 million per year on average (in
Following privatization, we would expect that debt relative to as- real dollars). The taxes paid have trended higher over time, although
sets would decrease for two reasons. First, CN now had direct access they have varied considerably from year-to-year. In 2011, for example,
to capital markets and should have found it easier to raise equity. CN paid taxes of $889 million, but less than half this amount in 2009.
Second, the removal of government ownership would, in theory, in- The difference in taxes paid between the long-run post-privatization
crease the cost of debt. This assumes, however, that CN actually paid period and the pre-privatization period was statistically significant
interest on its government debt, which it may not have done. In the (at the 0.01 level), but not between the short-run post-privatization
longer run, the debt-to-assets ratio might increase as profitability in- period and the pre-privatization period.
creases, risk is reduced and lenders become more willing to lend to CN.
In fact, however, the most noticeable change in CN's capital structure 4.5. Conclusion on operational performance changes resulting from
occurred prior to privatization when the debt-to-assets ratio increased privatization
from about 50% (in 1990) to 66% (in 1993 and 1994). As expected, the
debt-to-assets ratio did decrease following privatization, although the Our analysis shows that privatization has resulted in improved per-
change from the 5-year pre-privatization period to the 5-year post- formance by CN, using a broad range of operational performance mea-
privatization period was not statistically significant. Thereafter the sures. The statistically significant growth in sales, NI, profitability as
24 A.E. Boardman et al. / Research in Transportation Business & Management 6 (2013) 19–30

measured by ROA and ROS, and productivity as measured by the 5.1. Change in consumer surplus
sales per employee ratio and the NI to employee ratio, convincingly
demonstrate that CN's operating performance improved in the Any change in CS is largely determined by the magnitude of
post-privatization period. changes in prices and quantities following privatization. As mentioned
Some of these results were affected by CN's acquisitions in the U.S. and above, the 1987 National Transportation Act marked a transition
Canada. Certainly, assets, sales and employees were affected by the acqui- from an era of tacit collusion or parallelism to one of more direct
sitions. Improvements in profitability and productivity might also be intra-modal competition (Bonsor, 1995; Timur & Ponack, 2002). At
partially due to these acquisitions if there were economies of scale or the same time, both CN and CP faced increasing inter-modal competi-
economies of density. However, we would argue that these improve- tion from trucking and shipping. They responded to these pressures by
ments should be treated as benefits from privatization because, without reducing the track they owned and operated. In 1981, CN owned about
privatization, it is very unlikely that the acquisitions would have occurred. twice as much track as CP and, as shown in Fig. 1, it operated about 32%
Our results also show that CN's privatization has provided significant more track than CP. Between 1982 and 1997, both CN and CP reduced
benefits for the major stakeholders. The increase in profits generated the amount of track they operated in Canada by 27–28%. In 1998,
liquidity that CN distributed to shareholders as increases in dividends CN reduced the amount of track it operated by more than CP, thereby
and to the Canadian government as increases in taxes paid. The next reducing the gap between the two firms; this gap remained fairly con-
section directly examines the social welfare impacts of CN's privatization. stant until 2005. In 2005 and subsequent years, however, CN increased
the amount of operational track, while CP continued to reduce track in
5. A social welfare (CBA) assessment of the CN privatization each of these years, thereby increasing the gap again.
These competitive pressures resulted in falling real prices in
This section presents a CBA of CN (Domah & Pollitt, 2001; Florio, Canada (measured by dividing real Canadian sales revenue by the
2004; Galal, Jones, Tandon, & Vogelsang, 1994; Jones, Tandon, & number of revenue tonne-kilometres of Canadian freight) from
Vogelsang, 1990; Pollitt & Smith, 2002). The methodology is similar 1986 to 2004, as shown in Fig. 2. 1 During this period, real prices at
to that employed in Boardman, Laurin, Moore, and Vining (2009), CN and CP fell by almost 50%. The downward trend was slightly faster
although some assumptions differ slightly. Also, our data extend five in the 1986–1991 period, prior to Paul Tellier's appointment, than
more years to 2008. It is important to note that these data pertain subsequently. Due in part to these falling prices, both CN and CP
to Canadian freight operations, not to CN (or CP) as a whole. For were markedly less profitable than their U.S. counterparts (Waters,
some years, we applied a smoothing procedure for unusually high 1997). Fig. 2 also shows that, with the exception of 1982, CN and CP
restructuring and miscellaneous costs (Boardman et al., 2009, 64). Our charged very similar prices in Canada.
financial estimates are converted to present values (PV) as of 1995, As real prices fell, output (in revenue tonne-kilometres of Canadian
the year of privatization, but are expressed in 2011 Canadian dollars. freight) increased at CN and CP, despite the reduction in tracks
An important advantage of this CBA vis-à-vis most other privatiza- operated. As shown in Fig. 3, both firms experienced growth in
tion CBAs is that we use data from CP, a private sector company freight volumes from 1981 until 2008 that was in excess of 50%.
operating in the same industry and in the same country, to control Fig. 3 also shows that throughout this period, CN's and CP's Canadian
for contemporaneous industry-wide changes and other external freight volumes grew at similar rates and these rates did
factors. The use of such a closely-related external benchmark is, as far not change significantly following privatization. Since the amount
as we can ascertain, unique. It leads to a more accurate counterfactual of track operated fell at both firms, the track usage of both improved
and considerably strengthens the credibility of the estimated benefits. considerably.
From a normative perspective, the most appropriate measure of Given that CN and CP charged similar prices, and the output at
the social “value” of privatization is the net change in social welfare, both companies increased at similar rates, the market shares of both
which can be written as: firms (measured in terms of total Canadian freight revenues) have
    remained very stable, as shown in Fig. 4.
ΔW ¼ Vp –Vg þ λg Z−Tg −λp Z þ Tp ð1Þ As shown in Fig. 2, real prices fell following privatization and,
although they were trending upwards during the 2005–2008 period,
where, W is the social welfare, Vp is the value to society of CN after they were still lower in 2008 than prior to privatization. Also, as
privatization, Vg is the value to society of CN under continued govern- shown in Fig. 3, output increased following privatization. Conse-
ment ownership and Z is the proceeds received by the government. Tg quently, railroad consumers were better off and enjoyed higher levels
and Tp denote the transaction costs that government and private sec- of CS following privatization than before privatization. However, we
tor actors incur, respectively, and λg and λp are shadow multipliers do not attribute this increase in CS to the CN privatization per se.
(weights) on government revenue and private funds, respectively. Prices started to fall in 1986 and, in our view, would probably have
Later, we set these multipliers equal to unity. All variables in Eq. (1) continued to fall in the absence of privatization. Florio (2004) reaches
are measured in terms of their PVs. the same conclusion concerning the long-term trends in prices
Although aggregate social welfare is the most relevant criterion by following UK privatizations. Our evidence on pricing, output and mar-
which we assess privatization, from a public policy perspective it is ket share suggests that the gains in CS that actually materialized
also informative to consider the distribution of benefits and costs post-privatization would have occurred in the absence of privatiza-
among consumers, shareholders, government, and employees. For tion. Although there may have been some improvements in CN's
this purpose, we use the fact that the change in social welfare can service quality and safety, we cannot measure them and therefore,
be written (Boardman, Greenberg, Vining, & Weimer, 2011): we do not attribute any change in CS to privatization.

ΔW ¼ λc ΔCS þ λp ΔPS þ λg ΔGS þ λe ΔES ð2Þ

where, ΔCS is the change in consumer surplus (CS), ΔPS is the change in 1
A revenue tonne-kilometres means that 1 tonne of revenue-generating freight is
producer surplus (PS, economic profits or rents, after tax), ΔGS is the transported 1 km. One tonne, also called a metric tonne, equals 1000 kilograms, which
change in the government surplus (GS, government revenues minus ex- is about 2204.5 pounds. On average over the most recent five years, freight generated
about 94% of the total Canadian revenues of both firms; passengers generated less than
penditures), ΔES is the change in the employee surplus (ES), and λc and 1%. Other Canadian revenue sources included government subsidies (1.5%), compensa-
λe are shadow multipliers for CS and ES, respectively. Later, these mul- tion for services rendered to Via Rail (1%) and other miscellaneous revenues (3%)
tipliers are set equal to unity. (Statistics Canada, 2003–2008).
 A.E. Boardman et al. / Research in Transportation Business & Management 6 (2013) 19–30 25

Fig. 1. CN and CP Canadian track operated (kilometres).

5.2. Change in employee surplus and compute the profits for 1996 and subsequent years), minus the
profits that would have been earned under continued government
Conceptually, any change in ES should include changes in producer ownership for the same years, denoted Πg96 +, minus private sector
surplus (rents) gained (or lost) by either employees or by others, such and government transaction costs. Note that because of the assump-
as consultants and investment bankers. Unions often argue that in- tion that λp = λg = 1, the amount paid for the company at privatiza-
creased profits following privatization mostly come at the expense of tion, Z, is a transfer from shareholders to the government.
lower employee salaries. If so, such effects would raise PS and lower Now consider the first two terms in Eq. (3). Economic profits
ES. Furthermore, laid-off employees might not find jobs that pay as equal revenues minus costs. Earlier we argued that prices and output
well, they may experience periods of unemployment before finding an- were not affected by privatization and, therefore, sales revenues
other job or they may incur transaction costs in finding a new job. Each were not affected either. Thus, the difference in profits can be
of these factors would lower ES, especially during periods of negative or measured by the difference in costs. More specifically, suppose CpCN
low economic growth. denotes the PV of CN's costs under privatization, that is, from 1996
Fig. 5 shows Canadian employment at both CN and CP over time. and onwards:
In 1982, employment at CN was 66% higher than at CP, while in
2008 it was only 22% higher. This might suggest a loss of ES at CN. CN
CN ATCt CN
However, CN actually reduced employment at a slower rate following CP ¼ ∑ Q ð4Þ
t ð1 þ sÞt t
privatization (5.4% per annum during the 1995–2002 period) than
before it (6.6% per annum during the 1984–1994 period). Further-
more, CP increased the rate at which it reduced employment from where, ATCtCN and QtCN are CN's actual real average total unit costs
4.2% per annum during the 1984–1994 period to 5.3% per annum (ATC) and actual output levels in year t, respectively, and s is the
during the 1995–2002 period. Consequently, we conclude that the social discount rate. Further suppose CgCN denotes the PV of what
reduction in employment at CN would have occurred even in the these costs would have been with continued government ownership.
absence of privatization. Now, the change in welfare due to privatization is given by:
We also examined real wages. In 1982, (real) wages at CN were CN CN
slightly lower than (real) wages at CP, but the difference increased ΔW ¼ Cg −Cp −Tp −Tg : ð5Þ
during the pre-privatization period. By 1995, wages at CN were
slightly higher than at CP and they remained so until 2003. Thus, We use CP's actual costs to estimate the counterfactual; specifi-
there is no evidence that CN employees experienced pay reductions cally, the PV of what CN's costs would have been if it had not been
due to the privatization. Overall, therefore, we conclude that there privatized, CgCN. Fig. 6 shows that CN's and CP's real ATCs fell more
was no change in ES due to privatization. or less continuously from 1981 until 2004. From 1981 to 1987,
CN's real ATC was higher than CP's ATC, but was falling slightly
5.3. Change in welfare, assuming zero CS and ES faster. Thereafter, for the 7 years prior to privatization, 1988–1994
inclusive, CN's ATCs were on average 6.1% higher than CP's ATCs,
Under the assumptions that CS and ES equal zero and λp = λg = 1 and this difference remained fairly constant from year to year.
then, from Eq. (2), the change in welfare equals the sum of the Based on this evidence, we conclude that if CN had remained gov-
changes in PS and GS. This change in welfare can also be written: ernment owned, its real ATCs would have been 6.1% higher than
CP's ATCs.
96þ 96þ
ΔW ¼ Πp −Πg −Tp −Tg : ð3Þ Consequently, we construct a series of annual, counterfactual real
ATCs for CN by inflating CP's actual real ATCs by 6.1%. We then multi-
That is, the change in welfare equals the PV of profits following ply these ATCs by CN's actual annual outputs to arrive at CN's coun-
privatization, denoted Πp96+ (i.e., we treat 1995 as a transition year terfactual total annual costs. Finally, we discounted these amounts
26 A.E. Boardman et al. / Research in Transportation Business & Management 6 (2013) 19–30

Fig. 2. CN and CP freight prices per revenue tonne-kilometres in Canada (2011$).

at the social discount rate to obtain our estimate of the PV of CN's the PV of the cost savings attributable to privatization from 2009
^ CN :
total costs if it had not been privatized, denoted as C onwards. By using a real SDR of 3.5% (Boardman, Moore, & Vining,
g
2010; Moore, Boardman, Vining, Weimer, & Greenberg, 2004), we
CP estimate that the PV (in 1995) of the cost savings due to privatization
^ CN ¼ ∑ 1:061ATCt
C
CN
Qt : ð6Þ for the 1996–2008 period amounted to just under $8 billion (in 2011
g
t ð1 þ sÞt
dollars). From 2009 onwards, we assume that the real annual cost
savings will equal the average of the estimated real annual cost
From Eq. (5), our estimate of the PV of the change in welfare due
savings for the 2004–2008 period, which were just under $1 billion
to privatization (ignoring transaction costs) is given by:
per year (in 2011 dollars). Dividing this amount by the SDR implies
  that the value of this perpetuity as of 2008 is about $26.5 billion.
1:061ATCCP CN
t − ATCt
^ CN − CCN ¼ ∑ CN Finally, we discount this amount back to 1995, resulting in a PV of
Cg p Qt : ð7Þ
t ð1 þ sÞt just under $17 billion for the savings from 2009 onwards. Adding
these two amounts provides an estimate of the cost savings due to
It is important to note that this estimate of the change in welfare privatization of $24.7 billion (in 2011 dollars).
relies on the differences between CN's and CP's actual ATCs. By taking To calculate the change in welfare, we must subtract transaction
these differences, any industry-wide or economy-wide factor that costs. Private sector transaction costs, Tp, were relatively small and
might affect or bias CN's ATC, in effect, cancels out. Thus, using data can be ignored. It is reasonable to assume that the only non-trivial
from CP to compute our counterfactual provides a superior estimate transaction costs that government incurred, Tg, were the costs of
of the welfare gains from privatization. organizing the sale. These costs amounted to about $116 million in
Our data extend to only 2008. Therefore, we estimate Eq. (7) by 2011 Canadian dollars, which are calculated by multiplying $1.0125
using actual annual data for 1996–2008 and then estimate separately (the per share amount paid to the underwriters for the shares they

Fig. 3. CN and CP freight output in Canada (millions of revenue tonne-kilometres).

 A.E. Boardman et al. / Research in Transportation Business & Management 6 (2013) 19–30 27

Fig. 4. CN and CP market shares in freight in Canada.

sold) by the 83.8 million shares sold (Bruce, 1997, 149) and then less the foregone profits if CN had not been privatized, Πg96 +, and
converting to 2011 dollars. Including these transaction costs, gives government transaction costs, Tg:
an estimate of the welfare gain from privatization of $24.6 billion
(2011 dollars). ΔGS ¼ Z þ tp U þ tc Πp
95þ 95þ
−Πg −Tg : ð8Þ
As discussed above, Paul Tellier started to make efficiency im-
provements at CN immediately after his appointment. Thus, there
is some merit to the notion that the PV of the cost savings during Z equals the proceeds actually received by government, based on
the commercialization period (1993–1995) prior to actual privatiza- the issue price. While underpricing reduces this amount relative to
tion should also be included as a component benefit of privatization. what it could have been, this reduction is a transfer: it increases PS
In fact, however, the PV of these cost savings amounts to only and reduces GS.
$52 million (in 2011 dollars) and can, therefore, be ignored. For convenience we rewrite Eq. (8) as:

   
96þ 96þ 96þ
ΔGS ¼ Z þ tp U þ tc Πp −Πg –ð1−tc Þ Πg −Tg : ð9Þ
5.4. Change in government surplus

The change in government surplus, ΔGS, equals the proceeds from Here, tc(Πp96+ −Πg96+) represents the change in corporate taxes due
privatization, Z, plus personal taxes paid at the personal rate of tax, tp, to privatization, assuming that the government would pay corporate
on the underpricing, U, that is tpU, corporate taxes paid at the corpo- taxes at the same rate as the private sector, and (1−tc)(Πg96+) repre-
rate rate of tax, tc, on CN's subsequent profits Πp96+, that is tcΠp96 +, sents the after-tax profits under continued government ownership,

Fig. 5. CN and CP rail employment in Canada.

28 A.E. Boardman et al. / Research in Transportation Business & Management 6 (2013) 19–30

Fig. 6. CN and CP average total costs per revenue tonne-kilometre (2011$).

again assuming that the government would pay corporate taxes at the profits following privatization, denoted (1 − tc)Πp96 +, less the amount
same rate as the private sector. paid for the shares, private-sector transaction costs, and capital gains
Using the procedure described in Boardman et al. (2009), we taxes paid at the personal rate of tax tp on the underpricing, U:
estimate that the sale price, Z, was just over $3 billion (in 2011
dollars) and that the government tax receipts on the underpricing, 96þ
ΔPS ¼ ð1−tc ÞΠp −Z−Tp −tp U: ð11Þ
tpU, amounted to $110 million (in 2011 dollars). To obtain the corpo-
rate tax revenues on the incremental profits following privatization,
If we subtract the sale price of $3038 million and the personal
tc(Πp96+ − Πg96+), we multiply each year's real discounted, estimated
taxes on underpricing of $110 million from our estimate of the gain
cost savings due to privatization by the prevailing, combined federal–
in producer surplus, we obtain an estimate of the increase in
provincial corporate tax rate reported by CN and sum them. This yields
after-tax profits of $16,367 million, as shown in Table 2. As we
a PV of $10,077 million (in 2011 dollars).
discussed above, the expected after-tax increase in profits equals
We do not have a direct estimate of the after-tax profit if CN had not
$3497 million. Consequently, we estimate that the unexpected in-
been privatized, (1− tc)(Πg96+). However, the before-tax profits under
crease in after-tax profits was $12,870 million, which is also shown
continued government ownership, Πg96+, are likely to have ranged
in Table 2. Clearly, it paid shareholders to buy and hold their shares
between zero and the PV of the expected future profits under private
until after these increases were capitalized into the share price.
ownership at the time of privatization, denoted Πpe. We further note
that the sale price, Z, should equal the PV of the expected after-tax
5.6. Distribution of the welfare effects
future profits under private ownership, less any underpricing:
e
Z ¼ ð1−tc ÞΠp −U: ð10Þ Table 2 shows how the benefits were distributed among Canadians
and non-Canadians. Canadians, of course, received all of the ΔGS. The
Consequently, under the assumption that Π96+g = Πpe, Eq. (10) im-
Table 2
plies (1− tc)Πg96+ equals Z+ U, the actual sale price of $3038 million Summary of the welfare gains from the privatization of CN and their distributiona.
plus the underpricing, $459 million, which amounts to $3497 million
Total Canadians Rest of the world
(in 2011 dollars). The average of this extreme amount and $0 yields
an estimate of the after-tax profit under continued government owner- Government surplus
Proceeds from sale 3038 3038
ship of $1748 million.
Taxes on underpricing 110 110
Consequently, the estimated change in GS is about $11.4 billion (in Increased corporate taxes 10,077 10,077
2011dollars) as shown in Table 2. The receipt from the sale is not the
Total revenues 13,225 13,225
largest component of ΔGS; by far the largest benefit to government Less
comes about as a result of increased tax receipts on increased profits. Foregone govt profits after tax 1748 1748
Transaction costs 116 116
5.5. Change in producer surplus Net government surplus 11,361 11,361
Producer surplus
Assuming, as we have argued above, that privatization had no Expected after-tax profits 3497 2098 1399
effect on the surplus of shippers (consumers) or employees (ΔCS = Unexpected after-tax profits 12,870 5792 7079

ΔES = 0) and assuming, as we have above, that λp = λg, then the Total after-tax profits 16,367 7890 8477
gains to shareholders, ΔPS, can be derived from Eq. (2) by subtracting Less
Taxes on underpricing 110 83 28
the gain in government surplus from the total welfare gain. Given our
Sale price 3038 1823 1215
estimate of ΔW equal to $24,580 million and our estimate of ΔGS Transaction costs 0 0 0
equal to $11,361 million, we estimate ΔPS equals $13,219 million
Net producer surplus 13,219 5984 7235
(in 2011 dollars).
The change in producer surplus, ΔPS, is the PV of the net benefits Total increase in welfare 24,580 17,345 7235
a
to shareholders discounted at the SDR. It equals the PV of the after-tax All figures are in millions of 2011 real Canadian dollars.
 A.E. Boardman et al. / Research in Transportation Business & Management 6 (2013) 19–30 29

distribution of ΔPS is based on the assumption that Canadians held gains amounted to approximately $24.5 billion 2011 Canadian dollars.
about 60% of the shares when they were first listed (Bruce, 1997). The increase in government surplus was about $11 billion, and the
Therefore, Canadians received 60% of the expected after-tax profits, increase in producer surplus was about $13 billion with just under
amounting to just over $2098 million, but paid tax of $83 million on half of this amount going to Canadian citizens.
that gain and, of course, paid for their initial shares at the offering The considerable size of these estimates demonstrates the poten-
which was $1823 million. This amounted to $192 million. Soon after tial value of a well-conducted privatization. Factors that contributed
the stock was listed some Canadians sold their shares. About a decade to the success in CN's case include the following. First, CN was well
after privatization, Canadians owned about 55% of CN (Boardman et prepared for privatization and the privatization was not rushed.
al., 2009). Those who held on received their share of the unexpected Non-core businesses were divested over a number of years prior
after-tax profits, which amounted to $5792 million, resulting in a to privatization. A few years before privatization, a new CEO was
total benefit to Canadian shareholders of $5984 million. appointed with a mandate to improve efficiency and profitability in
preparation for an eventual sale. Second, CN focused almost exclu-
5.7. Concluding comments on the welfare effects sively on freight, and provided no passenger services (although VIA
Rail Canada and the Rocky Mountaineer provide some passenger ser-
One can never be certain that a state-owned operator would not have vices over parts of the network). Third, it was privatized as a vertically
achieved the same cost reductions as the privatized CN. Nonetheless, we integrated company without separation of ‘wheels and rail’, thereby
think this is unlikely. Our analysis makes conservative assumptions reducing transaction costs. Fourth, it was privatized through a
concerning the counterfactual. First, while CN had closed the cost gap share-issue privatization, rather than as a direct sale privatization.
with CP prior to privatization, the difference had not been eliminated This ensured that CN was listed on the stock exchange and that
and had remained stable (at about 6%) over the 7 years immediately ownership was not concentrated in a few hands. (In theory, this aspect
prior to privatization (for an even more conservative counterfactual, could have been either advantageous or disadvantageous.) Fifth, there
see Boardman et al., 2009). Second, by using data from CP we are able was a direct competitor. Sixth, there was some inter-modal competi-
to control for Canadian industry-wide effects. As we mention above, tion from trucking and shipping, although these alternative modes
the use of a very close competitor for such purposes is unique. Most im- were not as close substitutes as in many other countries, given the
portantly, if there were economies of scale effects, as Pollitt and Smith long-haul distances in Canada.
(2002) argued in the UK, they would likely affect both CN and CP simi- This privatization offers some lessons for private sector and
larly. Because we take differences, these effects would cancel out. government management. This case study shows that company
We know of no other factor apart from privatization that would management can matter a great deal. While CN's average total costs
explain why CN's unit costs decreased relative to CP's unit costs were about 6% higher than CP's ATC prior to privatization, they
after privatization. The effect of the U.S. acquisitions (such as Illinois were about 13% lower than CP's 5 years after privatization and
Central) would be minor because our welfare analysis is restricted thereafter. Tellier and Harrison took CN from underperforming rela-
to Canadian freight operations, both before and after the takeover. tive to CP to outperforming it. According to Thompson (2003, 331)
Furthermore, even if these acquisitions did have a beneficial effect, CN “is currently considered perhaps the best managed railroad in
it would be appropriate to include this effect because these acquisi- North America.” Management focused clearly on the core business
tions were unlikely to have occurred without privatization. and eliminated non-core assets, such as Wisconsin Central's holdings
There are many other reasons to presume that our estimates of in a number of smaller railroads.
the welfare benefits are conservative. Most importantly, we assumed Government managers carefully prepared CN for privatization, as
that price decreases and quantity increases were not attributable to we discussed above. Most importantly, they did not rush the privati-
privatization. Thus, we do not include any CS benefit to shippers or zation process. Tellier had 3 years to prepare CN for privatization.
other customers. Furthermore, ΔPS should also include any changes Another important lesson is that vertical disintegration (unbundling)
in the profits of CN's competitors and suppliers as well as the change may not be necessary for successful privatization when there is at least
in CN's profits in its other (non-Canadian freight) businesses. We very one other vertically integrated competitor. Indeed, Pittman (2005)
conservatively assume these additional effects are zero even though argues vertically separated freight railroads tend towards bilateral
CP achieved some efficiency gains following CN's privatization and monopoly, which has significant costs and only limited benefits. The
there may well have been spillover benefits to other parts of CN. Canadian experience suggests that having a vertically integrated
Similarly, we have not included any longer-term, dynamic benefits of duopoly is often a superior arrangement. Of course, these lessons pertain
the privatization. Since privatization, CN has initiated or participated to freight privatization and may not apply to passenger privatization.
in numerous railroad supply chain initiatives (Miller, 2010) that have A single, detailed study such as this one, in conjunction with
transformed the North American transportation network and have other research findings, should provide some guidance to govern-
benefitted shippers as reduced time-costs (Boyd, 2011; Rodrigue, ment policy makers in helping improve railroad privatization
2008; Starling, 2011). These benefits are probably substantial, but outcomes. There are some areas where additional research would
they are very difficult to measure and cannot be attributed entirely to provide further guidance. First, the performance and welfare effects
CN's privatization. of existing privatizations should be tracked over longer time periods
It is particularly important to point out that we have assumed that (a large percentage of our estimated welfare gains are due to the
λg = 1. Boardman et al. (2011) and other cost–benefit analysts sug- assumed perpetual benefits that arise beyond the end of our data
gest that λg = 1 plus the marginal excess tax burden (METB). Based period). Second, it would be helpful if the welfare impacts of rail
on a survey of relevant research, Boardman et al. (2011) suggest privatization could be further disaggregated by markets. This kind
that the METB is about 0.23, that is, it costs society about $1.23 to of analysis would shed light on the extent to which any measured
raise $1 in taxes. Multiplying our estimated increase in GS by 1.23 im- gains from privatization flow from the presence of inter-modal or
plies that ΔGS equals $16,259 million and our estimate of the increase intra-modal competition. It may be that those routes on which CN
in social welfare would increase concomitantly to $27,620 million. faced no effective competition did not produce the same cost savings
as on those where it had to compete directly against CP or where
6. Conclusion: the aggregate impact of privatization trucking or shipping were close substitutes. Third, more systematic
evidence is needed to determine the conditions under which share
Our main findings are that the privatization of CN improved sales, issue privatization is superior to direct private sale. Along the same
efficiency, profitability and social welfare. The total social welfare lines, more research is needed to substantiate our tentative conclusion
30 A.E. Boardman et al. / Research in Transportation Business & Management 6 (2013) 19–30

that freight rail privatizations have been more successful than passen- Galal, A., Jones, L., Tandon, P., & Vogelsang, I. (1994). Welfare consequences of selling
public enterprises: An empirical analysis. Oxford: Oxford University Press.
ger rail privatizations. Finally, in order to control as much as possible Gomez-Ibanez, J. (2010). The simple analytics of open access with illustrations from
for confounding effects, performance and welfare analyses of privat- railroads. Annual meeting of the Australian Competition and Consumer Commission,
izations should compare actual post-privatization results to those Surfer's Paradise, Queensland, July 29.
Grimm, C., & Winston, C. (2000). Competition in the deregulated railroad industry:
that would have occurred in an explicit counterfactual situation of Sources, effects and policy issues. In S. Peltzman, & C. Winston (Eds.), Deregulation
continued government ownership. Simple before-and-after compari- of network industries: What's next?. Washington, DC: Brookings.
sons cannot truly confirm that a privatization produced better results Heaver, T., & Waters, W. (2004). Canadian transport policy. In K. Button, & D. Hensher
(Eds.), Transport strategy, policy and institutions. Oxford: Pergammon Press.
than state ownership. Jones, L., Tandon, P., & Vogelsang, I. (1990). Selling public enterprises: A cost–benefit
It is now widely accepted that privatization improves perfor- methodology. Cambridge, MA: MIT Press.
mance in competitive markets in developed countries. But there East Japan Railway Company Management Planning Department (2008). 20 years of JR
East — Making a fresh start to shape the future. Japan Railway and Transport Review,
have been a number of reservations and caveats around this broadly
49, 18–24.
positive conclusion. Do these improvements persist over time? Can Karikari, J., Brown, J., & Nadji, M. (2006). The Union Pacific/Southern Pacific Railroads
these improvements occur in more oligopolistic markets? Although merger: Effect of trackage rights on rates. Journal of Regulatory Economics, 22(3),
based on a single case study, this analysis suggests a positive answer 271–285.
Laurin, C., Boardman, A. E., & Vining, A. R. (2004). Government underpricing of share-
to both questions. issue privatizations. Annals of Public and Cooperative Economics, 75(3), 399–429.
Laurin, C., & Bozec, Y. (2001). Privatization and productivity improvements: The case
References of Canadian National (CN). Transportation Research (Part E), 37(5), 355–374.
Mathieu, G. (2003). The reform of UK railways — Privatization and its results. Japan
Boardman, A. E., Greenberg, G. H., Vining, A. R., & Weimer, D. L. (Eds.). (2011). Cost– Railway and Transport Review, 34, 16–31.
benefit analysis: Concepts and practice (4th edition). Upper Saddle River, NJ: McCartney, S., & Stittle, J. (2008). ‘Taken for a ride’: The privatization of the UK rolling
Prentice Hall. stock industry. Public Money and Management, 28(2), 93–100.
Boardman, A. E., Laurin, C., Moore, M. A., & Vining, A. R. (2009). A cost–benefit analysis of McNish, J., Jang, B., & Silcoff, S. (2012). Collision course. Report on business, globe and
the privatization of Canadian National Railway. Canadian Public Policy, 35(1), 59–83. mail, Saturday, January 14, B1. (pp. B6–B7).
Boardman, A. E., Laurin, C., & Vining, A. R. (2002). Privatization in Canada: Operating Megginson, W. L., & Netter, J. M. (2001). From state to market: A survey of empirical
and stock market performance with international comparisons. Canadian Journal studies on privatization. Journal of Economic Literature, 39(2), 403–452.
of Administrative Sciences, 19(2), 137–154. Miller, L. (2010). CN's productivity pipeline. Railway Age, 211(11), 12–16.
Boardman, A. E., Laurin, C., & Vining, A. R. (2003). Privatization in North America. In D. Mizutani, F. (1999). An assessment of the Japan railway companies since privatization:
Parker, & D. Saal (Eds.), International handbook on privatization (pp. 129–160). Performance, local rail service and debts. Transport Reviews, 19(2), 117–139.
Cheltenham, UK: Edward Elgar (Chapter 7). Mizutani, F., & Nakamura, K. (1996). Effects of Japan National Railways' privatization
Boardman, A. E., Moore, M. A., & Vining, A. R. (2010). The social discount rate for on labour productivity. Papers in Regional Science, 75(2), 177–199.
Canada based on future consumption. Canadian Public Policy, 36(3), 325–343. Mizutani, F., & Nakamura, K. (1997). Privatization of the Japan National Railway:
Boardman, A. E., & Vining, A. R. (1989). Ownership and performance in competitive Overview of performance changes. International Journal of Transport Economics,
environments: A comparison of the performance of private, mixed, and state-owned 24(1), 75–99.
enterprises. Journal of Law and Economics, 32(1), 1–33. Moore, M. A., Boardman, A. E., Vining, A. R., Weimer, D. L., & Greenberg, D. H. (2004).
Boardman, A. E., & Vining, A. R. (2012). A review and assessment of privatization in ‘Just give me a number!’ Practical values for the social discount rate. Journal of Pol-
Canada. School of Public Policy. University of Calgary: SPP Research Papers, 5,(4). icy Analysis and Management, 23(4), 789–812.
(pp. 1–28). Pittman, R. (2005). Structural separation to create competition? The case of freight
Bonsor, N. (1995). Competition, regulation and efficiency in the Canadian railway railways. Review of Network Economics, 4(3), 181–196.
and highway industries. In F. Paldi (Ed.), Essays in Canadian surface transportation. Pollitt, M. G., & Smith, A. S. G. (2002). The restructuring and privatisation of British Rail:
Vancouver: Fraser Institute. Was it really that bad? Fiscal Studies, 23(4), 463–502.
Boubakri, N., & Cosset, J. -C. (1998). The financial and operating performance of newly Rodrigue, J. -P. (2008). The thruport concept and transmodal rail freight distribution in
privatized firms: Evidence from developing countries. Journal of Finance, 53(3), North America. Journal of Transport Geography, 16(4), 233–246.
1081–1110. Secretariat, O. E. C. D. (2006). Structural reform in the rail industry. OECD Journal of
Boyd, J. D. (June 20). BNSF under Buffett. The Journal of Commerce, 10–13. Competition Law and Policy, 8(2), 69–79.
Bruce, H. (1997). The pig that flew: The battle to privatize Canadian National. Vancouver: Shirley, M. M., & Walsh, P. M. (2001). Public vs. private ownership: The current state of
Douglas and McIntyre. the debate. Washington, DC: The World Bank policy research working paper, No. 2420.
Canadian National Railway Company (1990–2011). Annual reports. Smith, A. (2006). Are British railways costing too much: Perspectives based on TFP
Caves, D., & Christensen, L. (1980). The relative efficiency of public and private firms in comparisons with British Rail 1963–2002. Journal of Transport Economics and
a competitive environment: The case of Canadian railroads. Journal of Political Policy, 40(1), 1–44.
Economy, 88(50), 958–976. Soto, R. (2010). End of the line: Railroads in Chile. Working paper 391. Instituto de
Chirwa, E. W. (2004). Industry and firm effects of privatization in Malawian oligopolistic Economia, Pontificia Universidad Catolica de Chile available at http://www.economia.
manufacturing. Journal of Industrial Economics, 52(2), 277–290. puc.cl/docs/dt_391.pdf, accessed 16 Oct 2012
Clark, R. (2010). Full circle: Rail industry privatisation in New Zealand, and a new Starling, D. L. (2011). Intermodal's explosive growth. Railway Age, 212(9), 57–58.
theory of its fundamental conceptual weakness. ETC. Conference Proceedings. Statistics Canada (1983–2010). Rail in Canada, 1981–2008. Catalogue # 52-216-X.
Cowie, J. (2009). The British passenger rail privatization. Journal of Transport Economics Ottawa: Minister of Industry.
and Policy, 43(1), 85–104. Thompson, L. (2003). Changing railway structure and ownership: Is anything working?
Cowie, J. (2010). Productivity and performance in the British rail freight industry since Transport Reviews, 23(3), 311–355.
privatisation. 12th WCTR, Lisbon, Portugal. Thompson, L., Budin, K., & Estache, A. (2001). Private investment in railways: Experience
Crompton, G., & Jupe, R. (2007). Network rail — Forward or backward? Not-for-profit in from South and North America, Africa and New Zealand. PTRC Conference, Cambridge.
British transport. Business History, 49(6), 908–928. Timur, A. T., & Ponack, A. (2002). Labor relations and technological change at Canadian
D'Souza, J., & Megginson, W. L. (1999). The financial and operating performance of Pacific Railway. Journal of Labor Research, 23(4), 535–557.
privatized firms during the 1990s. Journal of Finance, 54(4), 1397–1438. Waters, W. (1997). Changes in structure and performance in the North American
De Fraja, G. (1991). Efficiency and privatisation in imperfectly competitive industries. rail industry. In J. Jones, & A. Stewart (Eds.), The new environment in freight rail.
Journal of Industrial Economics, 39(3), 311–321. Kingston, ON: Queen's University School of Policy Studies.
Deveau, S. 2012. Corporate culture shift underway in Canada after CP Rail proxy fight. Williams, R., Greig, D., & Wallis, I. (2005). Results of railway privatization in Australia and
Financial Post, May 11. New Zealand. Washington, D.C.: The World Bank.
Domah, P., & Pollitt, M. G. (2001). The restructuring and privatisation of electricity dis- Willner, J., & Parker, D. (2007). The performance of public and private enterprise under
tribution and supply business in England and Wales: A social cost–benefit analysis. conditions of active and passive ownership and competition and monopoly. Journal
Fiscal Studies, 22(1), 107–146. of Economics, 90(3), 221–253.
Evans, A. (2007). Rail safety and rail privatisation in Britain. Accident Analysis and Wilson, J. (2010). Short history of post-privatization in New Zealand. New Zealand Treasury
Prevention, 39(3), 510–523. 1982752v1. Available at: http://www.comu.govt.nz/resources/pdfs/mixed-ownership-
Evans, A. W. (2010). Rail safety and rail privatisation in Japan. Accident Analysis and model/mom-shppnz-wilson-dec10.pdf (accessed 15 October 2011)
Prevention, 42(4), 1296–1301. Winston, C., Maheshri, V., & Dennis, S. (2011). Long-run effects of mergers: The case of
Florio, M. (2004). The great divestiture: Evaluating the welfare impact of the British U.S. western railroads. Journal of Law and Economics, 54(2), 275–304.
privatizations, 1979–1997. Cambridge, MA: MIT Press. Yvrande-Billon, A., & Ménard, C. (2005). Institutional constraints and organizational
Fowkes, A., & Nash, C. (2004). Rail privatisation in Britain — Lessons for the rail freight changes: The case of the British rail reform. Journal of Economic Behavior & Organi-
industry. ECMT Round Table 125, European integration of rail freight. zation, 56(4), 675–699.