I. HISTORY OF BENEFIT-COST ANALYSIS
Benefit-cost analysis attempts to solve collective action problems, which arise
when individuals or groups pursuing narrow self-interest without coordination arrive at
outcomes inferior to those that could be achieved by coordination. CBA was developed
as a coordinating decision rule that could solve collective action problems and lead to
outcomes that were both fair and efficient. (Porter)
The Army Corps of Engineers and CBA in Practice
The motivating force behind CBA in the United States was a desire to allay
conflict and reach agreement. Its development and increasing quantification was not the
product of technical elites but of disagreement, suspicion, and conflict, particularly
bureaucratic conflict.1 Benefit-cost methods were introduced in the U.S. by the U.S.
Army Corp of Engineers.2 Before the creation of the Corps, evaluations of pubic
investments were almost completely ad hoc.3 The prestige of the Corps, along with its
increasing quantification of project costs and benefits, was used by Congress to simplify
their decision-making and avoid the wildly uneconomic projects produced by logrolling.4
After 1902, water projects needed to be certified as beneficial by a Board of Engineers
for Rivers and Harbors that was established within the Corps. The Corps was far from a
rubber stamp: they rejected more than half of the proposed projects, usually on the basis
that their economics were unfeasible. By the 1920s the Corps required its recommended
projects to promise benefits in excess of costs. Through the 1930s the numbers put
forward by the Corps were generally accepted without question.5 The Corps was
recognized as a relatively neutral and respected arbiter in Congressional fights over water
projects.6 The creation of the Corps, then, represented not only the creation of an agency
to build projects, but an agency to increase Congressional and public efficiency.
1 PORTER, supra note 3.
2 Id, p. 149. According to Hammond, the use of formal benefit-cost ratios goes back at least as far
as the Rivers and Harbor Act of 1902, and were explicitly mandated in the amendment to the
Act in 1920. Richard J. Hammond, Convention and Limitation in Benefit-Cost Analysis, 6 NAT.
RESOURCES J. 195-222 (1966).
3 PORTER, supra note 3, at 150.
4 Id. at 148 (The Corps was fashioned after the famous Corps des Ponts de Chausse’es)
5 Id. [Porter, Trust]
6 Id. at 153.
The Flood Control Act of 1936 mandated what was already Corps practice: the
requirement that benefits of a project exceed its costs. More importantly, the Act allowed
Congressional authorization only for projects that had been approved by the Corps. The
Corps economic analyses limited debate in Congress over water projects, and the number
of really disgraceful projects became rarer.7 Political influence and special interest
legislation were not, of course, eliminated—the Corps tended to transgress its standards
when political forces were overwhelming—but they were curtailed.
After 1940, Corps decisions became the subject of bitter controversy as the Corp
was challenged by powerful electric and railroad utilities, by shipping interests, and by
rival Federal agencies, especially the Bureau of Reclamation and the Department of
Agriculture.8 Rival techniques or standards for CBA became the norm until an attempt
was made to resolve differences by relying on first principles of economics. The closest
result to reaching agreement was the “Green Book.” Although agreement was
significantly incomplete, the grounds for decision making were pretty well established as
rooted in economic theory. CBA was thereby transformed by conflict into a set of
rationalized economic principles.
The redefinition of CBA according to economic standards began “in earnest” in
the mid-1950’s.9 Porter notes that economists agreed with budget officials that the
standards for passing a benefit-cost test had not been set strict enough. They
recommended using uniform, higher discount rates than the rate on government bonds.10
Economists, however, were open to placing monetary values on what had been regarded
previously as intangibles, such as the value of landscapes, so that the scope of values
potentially subject to CBA expanded. The result of the new economic respectability of
CBA was its spread to all kinds of government expenditures and to regulatory activity.
The use of CBA at the federal level significantly increased with the issuance of an
Executive Order in 1981 by President Reagan declaring that Regulatory Impact Analyses
7 Id. at 155-57.
8 Id. at 161-75.
9 Id. at 188.
10 This recommendation appears suspect today. See generally RICHARDO. ZERBE, JR. & DWIGHT
DIVELY, BENEFIT-COSTANALYSIS IN THEORY AND PRACTICE (1994)..
be conducted for major initiatives.11 An additional Executive Order was issued by
President Clinton in 1994, confirming the government’s commitment to CBA and
highlighting the bipartisan support for CBA in federal regulatory decision-making. While
comprehensive legislation requiring the broad use of formal CBA has yet to be approved
by Congress, the presence of CBA is nonetheless apparent within various levels of
governmental decision-making. Moreover, public pressure for increased governmental
efficiency increases the need for CBA in practice.12
The Pareto Criterion and CBA in Theory
As benefit-cost analysis became more the province of economists than engineers, its
foundations became rooted in the Kaldor-Hicks criterion. Although this criterion has the
virtue of attempting to be definite, it is not wholly successful. Moreover, it retains logical
flaws. These defects are rooted in its history.
Although many economists are themselves unclear about the core assumptions of
CBA, we can derive what can reasonably be called the main view of it by considering its
origins. The first form of CBA, developed by Nicholas Kaldor, built on the foundation
provided by Vilfredo Pareto. Pareto (1896) introduced a welfare criterion, the Pareto
optimum, which became a foundational concept in welfare theory.13 A Pareto optimum is
a state of affairs such that no one can be made better off without making someone else
worse off.14 A change in the economy is said to represent a Pareto improvement over
what came before it, or to be Pareto superior to what came before it, if at least one person
is made better off as a result of the change and no person is made worse off.15 The Pareto
11ANTHONY BOARDMAN, ET AL. COST BENEFIT ANALYSIS: CONCEPTS AND PRACTICE, 5
(2001).
12 Id. at 5-6.
13 Vilfredo Pareto, COURS D’ECONOMIE POLITIQUE, vol. II (1896).
14 In its strong form, Pareto efficiency states that state A is preferred to state B when state A is
ranked higher than state B for one person and all other persons rank A at least as high as B. If
the utility (well-being) of each individual is higher in state A, then state A is preferred
according to the weak form of Pareto efficiency. See ROBIN W. BOADWAY & N. BRUCE,
WELFARE ECONOMICS (1984).
15 The attraction of the Pareto notion of efficiency is that it seems to eliminate interpersonal
comparisons of welfare. Some economists feel that “the inescapable conclusion” is that if one
precludes interpersonal comparisons of welfare the only logically consistent foundation
analysis is the Pareto principle. Its obvious limitation is that it is not very policy relevant; few
policies have no losers. This limitation resulted in a search for a more applicable measure of
welfare that continues to this day and of which this article is a part.
criterion is not useful for most practical purposes for exactly the same reason that a
criterion of unanimity is not useful in most voting situations.16 The practical substitute for
the Pareto criterion is the potential Pareto criterion, also known also as the Kaldor-Hicks
criterion.
The Development of the Kaldor-Hicks Criterion
The Kaldor-Hicks (KH), or potential Pareto criterion, is the standard for CBA. It
arose during the late 1930s out of discussions among prominent British economists about
repealing the Corn Laws.17 Before that time it was generally assumed that each individual
had an "equal capacity for enjoyment," and that gains and losses among different
individuals could be directly compared.18 By 1939, however, leading British economists,
including the future Nobel Prize winner Sir John Hicks, were raising questions about
such policy prescriptions because they involved interpersonal comparisons of utility.19
Nicholas Kaldor provided a solution: he acknowledged the inability of economists to
establish a scientific basis for making interpersonal utility comparisons but suggested that
16 See Duncan Black, On Arrow’s Impossibility Theorem, 12 J.L. & ECON. 227 (1970). (who
shows that such a rule did not work very well for the Polish legislature).
17 These include: Robbins, Hicks, Kaldor, and Harrod, all writing in THE ECONOMIC JOURNAL.
See generally Lionel Robbins, Interpersonal Comparisons of Utility: A Comment, 48 ECON. J.
635 (1938); John R. Hicks, The Foundations of Welfare Economics, 49 ECON. J. 696 (1939);
Nicholas Kaldor, Welfare Propositions in Economics and Interpersonal Comparisons of Utility,
49 ECON. J. 549 (1939); Roy F. Harrod, Scope and Method of Economics, 48 ECON. J. 383
(1938).
18 See Ezra J. Mishan, INTRODUCTION TO NORMATIVE ECONOMICS 120–21 (1981); Peter
Hammond, Welfare Economics, in ISSUES IN CONTEMPORARY MICROECONOMICS AND
WELFARE 406 (George Feiwel ed., 1985). For example, Harrod argued that the net social
benefit from a policy could be established on the assumption that the individuals affected were
equal in their capacity to enjoy income. That is, an improvement can be assumed by looking at
changes in income as long as, in modern terminology, the marginal utility of income with
respect to income changes are the same for all individuals. See Harrod, supra note 34, at 387.
Harrod used this reasoning to justify the 1846 repeal of the English Corn Laws, a classic test
case for British economists. In response, Lionel Robbins pointed out that interpersonal
comparisons of utility couldn't rest on a scientific foundation since utility cannot be measured,
and that the justification for such comparisons is more ethical than scientific.18 Harrod
complained that in the absence of comparability of utility of different individuals, “the
economist as an advisor is completely stultified.” See Harrod, supra note 34, at 396–97.
19 See Hicks, supra note 34, at 670. This debate about whether or not prescriptions of economics
were scientific is paralleled by the 1980s debate, mostly in the legal literature, about the
normative foundations of wealth maximization. For example, see 8 HOFSTRA L. REV., volumes
3 and 4 (1980). The 1980s debate was haunted, and confounded, by the issues that we consider
in this article.
this difficulty could be made irrelevant.20 His argument was that policies that led to an
increase in aggregate real income were always desirable because the potential existed to
make everyone better off:
[T]he economist’s case for the policy is quite unaffected by the question
of the comparability of individual satisfaction, since in all such cases it is
possible to make everybody better off than before, or at any rate to make
some people better off without making anybody worse off.21
According to Kaldor, a project is desirable if the money measure of gains exceeds the
money measure of losses. With regard to the potential compensation that could turn
losers into winners in such situations, Kaldor goes on to note that whether actual
compensation should take place “is a political question on which the economist, qua
economist, could hardly pronounce an opinion.”22 Hicks, perhaps the most prominent
economist of the time, accepted the Kaldor approach, which eventually became known as
the Kaldor-Hicks (KH) criterion.23 The KH criterion is the usual CBA criterion.24
20 See Kaldor, supra note 34,, at 549–550; Robbins, supra note 34, at 640.
21 See Kaldor, supra note 34, at 549–550.
22 See Kaldor, supra note 34, at 550. It was thought that politicians or non-economists should
make judgments and decisions about income distribution effects.
23 See Hicks, supra note 34, at 671. A few years after the creation of KH, Scitovsky (1941)
introduced a parallel, but slightly different, criterion that states that a project is desirable if the
losers are unable in the original state of the world to bribe the potential winners not to
undertake the project. Both of these criteria are referred to as potential compensation tests.
Shortly after this Sir John Hicks showed that the Kaldor and Scitovsky criteria are related to
measures of willingness top pay for a good and willingness to accept payment for a good.
24 As envisioned by Kaldor, non-pecuniary effects were to be included in benefit-cost analysis.
Kaldor, supra note 34 at ____?
"An increase in the money value of the national income (given
prices) is not, however, necessarily a sufficient indication of this condition [the
potential compensation test or Kaldor criterion] being fulfilled: for individuals
might, as a result of a certain political action, sustain loses of a non-pecuniary
kind, e.g., if workers derive satisfaction from their particular kind of work, and
are obliged to change their employment, something more than their previous
level of money income will be necessary to secure their previous level of
enjoyment; and the same applies in cases where individuals feel that the carrying
out of the policy involves an interference with their individual freedom. Only if
the increase in total income is sufficient to compensate for such losses and still
leave something over to the rest of the community, can the project be said to be
'justified' without resort to interpersonal comparisons."
Clearly it is sentiments that are to be valued and not just objects.
KH attempts to avoid interpersonal utility comparisons by separating equity from
efficiency. Kaldor proposed that decision makers address sentiments regarding equity
outside the purview of CBA.25 The change in aggregate gains was to be the measure of
efficiency, so that there was a separation of effects into those of efficiency and
distribution.26 Kaldor endorsed the procedure adopted by Pigou, which Kaldor describes
as “dividing welfare effects into two parts: the first relating to production, and the second
to distribution.”27 (Of course, the KH approach produces outcomes that are equivalent to
those produced by the assumption that the marginal utility of income is the same across
all individuals, i.e., that each dollar of benefit or cost is treated the same regardless of
who received it.28) Hicks agreed with this separation and noted that “if measures making
for efficiency are to have a fair chance, it is extremely desirable that they should be freed
from distributive complication as much as possible.”29 To Hicks it would be “‘rather a
dreadful thing’ to have to accept the view that welfare analysis was unscientific. If it
were, its conclusions would “… depend on the scale of social values held by a particular
investigator. Such conclusions can possess no validity…; one’s welfare economics will
inevitably be different according as one is a liberal, or a socialist, a nationalist or an
internationalist, a christian [sic] or a pagan.”30
25 It cannot be said that this second assumption of equal marginal utility of income avoids
interpersonal comparisons; indeed it embraces them in a very particular way: all people are
treated equally in terms of the value they place on changes in income.
26 Kaldor, supra note 34, at 551.
27 The eagerness of economists to separate considerations of efficiency from those of distribution
arose from a desire to put economics on a firm base as a policy instrument. Kaldor suggests,
“the economist should not be concerned with prescriptions at all . . . For, it is quite impossible
to decide on economic grounds what particular pattern of income-distribution maximizes social
welfare.” Kaldor, supra note 34, at 551; see also A.C. Pigou, THE ECONOMICS OF WELFARE
(4th ed. 1932).
28 See John Chipman & James C. Moore, The New Welfare Economics 1939–1974,
INTERNATIONAL ECONOMIC REVIEW 19(3), at 578 (1978); Kaldor, supra note 34, at 551.
Certainly Mishan was aware that questions of distribution belonged to welfare economics and
recognized that the separation was useful, since there was less agreement about the income
distribution issues. See Ezra J. Mishan, The Principle of Compensation Reconsidered, 60 J.
POLITICALECON. 312 (1952).
29 Hicks, supra note 34, at 712.
30 Hicks, supra note 34, at 696.
This separation of efficiency and equity has remained the common, though not
universal, basis of normative analysis to this day.31 The more modern justifications for
the separation are that changes in the income distribution are usually better effected
through macroeconomic policy rather than through individual projects.32 This defense,
however, leaves unaddressed matters of equities for identifiable peoples or groups, or
sentiments attached to particular projects that cannot be handled by macro policy; equity
and justice are particular as well as general.33
Measurement of Benefits and Costs
31 Posner’s term wealth maximization appears identical to KH except that he would allow
altruistic concerns where there is a WTP for them. A strong theory of wealth maximization is
said to have three crucial features. First, wealth maximization is an aggregate concept. Id. at
251. That is, it is more concerned with societal well-being than with individual welfare. For
example, a wealth maximizer is not concerned with the distribution of wealth among citizens,
and any coerced payment to affect distribution is presumed unproductive. See, e.g., Richard A.
Posner, Wealth Maximization Revisited, 2 NOTRE DAME J.L. ETHICS & PUB. POL'Y 85, 103
(1985) ("To the wealth maximizer, altruism is neither good nor bad; but given that it exists,
there is a legitimate if limited role for public wealth redistribution."). Second, to the extent that
altruism exists within society, public efforts to reduce poverty may be justified because poverty
reduction will benefit both non-donors and donors. Id. Finally, questions of need and desires
are irrelevant. Posner, supra note 6, at 61 ("The individual who would like very much to have
some good but is unwilling or unable to pay anything for it – perhaps because he is destitute –
does not value the good in the sense in which we are using the term 'value.'"). Note that this is
not the same as including altruism in a benefit-cost analysis as its inclusion may change the
sign of a benefit-cost analysis yet the altruistic component might not benefit both parties as
when the transfer from altruists is greater than they would prefer.
32 See generally Mitchell A. Polinsky, AN INTRODUCTION TO LAW AND ECONOMICS 5 (2d ed.
1989).
33 We note some rather random examples from reputable sources. Boardman, Greenberg, Vining,
and Weimer note that, “Strict use of the Kaldor-Hicks test means that information on how
benefits and costs are distributed among groups is ignored in decision making.” See David L.
Weimer & A.R. Vining, POLICY ANALYSIS: CONCEPTS AND PRACTICE 412 (2d ed. 1992).
Friedman also notes, “Some analysts would like to ignore equity altogether and use the
compensation test as the decisive analytic test… [A] second rationale for relying on the
compensation test is the belief that concern for equity is simply unfounded.” See Lee Friedman,
MICROECONOMIC POLICY ANALYSIS 170 (1984). Additionally, Posner notes that wealth
maximization is simply the Kaldor-Hicks tests and that wealth maximization ignores
distributional effects. See Richard A. Posner, ECONOMIC ANALYSIS OF LAW 13 (3d ed. 1986);
Richard A. Posner, The Justice of Economics, 15 J. PUB. FIN. & PUB. CHOICE 23, 132–33
(1987). McCloskey incorrectly contends that the consumer surplus measure of social happiness
is the same as the national income measure. See Donald McCloskey, THE APPLIED THEORY OF
PRICE 229 (1982). Of course, the national income measure contains no measure of income
distribution.
In 1943 Hicks defined the compensating and equivalent variations (CV and EV) and from
these the concepts of the willingness to pay (WTP) and willingness to accept (WTA).34
In the context of a simple world of market goods and narrowly self-interested
individuals, Hicks described the relationship between the CV and the WTP or WTA test,
which became the standard for CBA. The move from A to B passes the KH test if and
only if the sum of individual’s willingness to pay for the change exceeds the sum of the
willingness to accept payment for the change. 35].36 The WTP reflects the amount that
someone who does not have a good would be willing to pay to buy it; it is the maximum
amount of money one would give up to buy some good or service, or would pay to avoid
harm. The WTA reflects the amount that someone who has the good would accept to sell
it; it is the minimum amount of money one would accept to forgo some good, or to bear
some harm. The benefits from a project may be either gains (WTP) or losses restored
(WTA). The costs of a project may be either losses (WTA) or gains forgone (WTP). Both
the benefits and the costs are the sum of the appropriate WTP and WTA measures. Thus,
the relation of benefits and costs to the WTP and the WTA is:
Benefits: The sum of the WTPs for changes that are seen as gains and of the WTAs for
changes that are seen as restoration of losses.
34 This CV is not to be confused with Contingent Valuation. To understand the concepts of
compensating and equivalent variations, consider a individual who will be affected by a move
from state A to state B. Her compensating variation (CV) for the move from A to B is the income
adjustment necessary in state B in order to make her indifferent between A and the incomeadjusted
B. (If she prefers B to A, her CV is positive and reflects the maximum amount she would
be willing to pay to move from A to B; if she prefers A to B, her CV is negative and its absolute
value reflects the minimum amount she would be willing to accept to move from A to B.) Her
equivalent variation (EV) for the move from A to B is the income adjustment necessary in state A
in order to make her indifferent between B and the income-adjusted A. (If she prefers B to A, her
EV is positive and reflects the minimum amount she would be willing to accept to move from B
to A; if she prefers A to B, her EV is negative and its absolute value reflects the maximum
amount she would be willing to pay to move from B to A.)
It follows from these definitions that the compensating variation for a move from A to B gives
the same value, but with the opposite sign, for the equivalent variation for a move from B to A.
It is also true that either of these variations are exact utility indicators in the sense that for a
single individual they rank preferences correctly: an individual’s CV or EV for the move from
A to B is positive if and only if the individual prefers B to A.
35 The equivalent variation uses the WTA for gains and the WTP for losses and is a test proposed
by Scitovsky test. [Should we add details here about Scitovsky like in the last paper?]
36 See ZERBE& DIVELY, supra note 27.
Costs: The sum of the WTAs for changes that are seen as losses and of the WTPs for
changes that are seen as forgone gains.
The justification for adopting these methods of measurement is that they correspond with
the psychological sense of gains and losses.37 [Here, and to some extent throughout this
section, I find myself wondering if we’re talking about KH, or KHM, or both… According
to the outline, this should be a discussion of the development of KH, but the outline can
be changed if necessary…] The measurements are summarized in table 1 below.38
Table 1. The Measurement of Benefits and Costs in Terms of Gains and Losses*
The Compensating Variation (KH Measure)
Benefits GAIN:WTP–the sum of CVs for a positive change–is finite.
LOSS RESTORED: WTA-the sum of CVs for a loss restoredcould
be infinite.
Costs LOSS: WTA–the sum of CVs for a negative change–could be
infinite.
GAIN FORGONE: the sum of CVs is finite.
The KH test is said to be satisfied when the gains are sufficient to hypothetically
compensate the losses. (There are circumstances, discussed later, in which gains exceed
losses but such compensation is not possible, even hypothetically.) The economic worth
of a good to an individual is determined by his or her desire for it, whether a gain or a
loss is involved, the income and wealth of the person, and the uniqueness of the good.
These features are all captured by the WTP and WTA measures. The above discussion
and set of characteristics outlined above define what can reasonably be called the
mainstream view of economic efficiency and of BCA, a view based on KH. A full
37 See Daniel Kahneman & Jack Knetsch, Anomalies: The Endowment Effect, Loss Aversion, and
Status Quo Bias,5 J. ECON. PERSPECTIVES 193–206 (1991).
38 The difference between benefits and costs is simply their sign: positive for benefits and
negative for costs. Thus, without loss of accuracy, costs can be counted as negative benefits and
benefits can be counted as negative costs.
description of the assumptions of the mainstream view of CBA is reasonably
characterized by: (1) the use of the willingness to pay (WTP) for gains and for losses;39
(2) a reliance on potential compensation tests so that a project is KH efficient only when
it passes a potential compensation test; (3) an emphasis on efficiency that is separated
from equity; (4) an assumption that a dollar is to be treated the same regardless who
receives it, so that a dollar is assumed to have the same value to each person (equal and
constant marginal utility of income); (5) a recognition and inclusion of non-pecuniary
effects; (6) the omission of values represented by moral sentiments; (7) a reliance on
externalities and market failure to determine where BCA might be useful in making
corrections; (8) an assumption that transactions costs are zero;40 (9) the treatment of BCA
as a mechanism to provide the answer rather than an approach proving information as
part of an ongoing discussion; and (10) the inclusion of wealth maximization as
congruent with mainstream BCA.
The history discussed above resulted in what can reasonably be called the
mainstream view of economic efficiency and of CBA, a view based on KH. The
mainstream view of BCA is well shown in practice by the BCA’s of the federal
developmental agencies, and well illustrated in theory by the critical analyses of Lothrup,
Ackerman and Heinzerling and others.41
39 Although it is recognized that the willingness to accept is the correct measure for losses,
traditional opinion has held that there is little difference between the two measures so that WTP
may be used in practice.
40 For an explanation of why this leads to difficulties See RONALDH. COASE, THE FIRM, THE
MARKET, AND THE LAW5 (1988). Coase’s view is that the major failing of welfare economics
lies in its assumption of zero transactions costs. By transactions costs I mean the costs
necessary to transfer, establish and maintain property rights (See Douglas Allen, What are
Transaction Costs?, in RESEARCH INLAW& ECONOMICS 14, at 4 (Zerbe & Goldberg eds.,
1991).)
41 Robert C. Lothrup, The Misplaced Role of Cost-Benefit Analysis in Columbia Basin Fishery, 16
ENVTL. LAW. 517 (1986). Lothrup assumes BCA is unrelated to law and to legal rights, that the
willingness to pay is the correct measure for both gains and losses and that moral and ethical
values are excluded. He apparently arrives at this view in part from looking at actual BCA
studies. Because even mainstream BCA is incompletely defined, my characterization of it is
necessarily inadequate as it makes definite what is in fact vague. However, it is a useful starting
point from which to consider other possibilities. Also, this characterization of the mainstream
view is convenient as it is fairly close to the view critics of BCA hold, although the critics' view
is narrower than actual practice suggests