cost–benefit analysis _ The New Palgrave Dictionary of Economics

Bookmark Print David L. Weimer From The New Palgrave Dictionary of Economics, Second Edition, 2008 Edited by Steven N. Durlauf and Lawrence E. Blume Abstract Cost–benefit analysis (CBA) is a collection of methods and rules for assessing the social costs and benefits of alternative public policies. It promotes efficiency by identifying the set of feasible projects that would yield the largest positive net benefits to society. The willingness of people to pay to gain or avoid policy impacts is the guiding principle for measuring benefits. Opportunity cost is the guiding principle for measuring costs. CBA requires that appropriate shadow prices be derived when policies have effects beyond those that can be taken into account as changes of prices or quantities in undistorted markets. Keywords consumer surplus; contingent valuation; cost–benefit analysis; distortions; donor value; equivalent variation; Hicks compensation; Hicks, John R.; Kaldor, N.; Marshallian demand curves; opportunity cost; option price; present value; pure time preference; revealed preference; shadow prices; social choice; social surplus; substitutes and complements; travel-cost method; value of statistical life; willingness to pay Article Public policies, such as infrastructure projects, social welfare programmes, tax laws and regulations, typically have diverse effects in the sense that people would be willing to pay something to obtain effects they view as desirable and would require compensation to accept voluntarily effects they view as undesirable. If, across all members of society, the total amount willing to be paid by those who enjoy desirable effects (benefits) exceeds the total amount needed to compensate those who suffer undesirable effects (costs), then adopting the policy would make it potentially possible to achieve a Pareto improvement on the status quo. If the benefits do not exceed the costs, then adopting the policy does not offer a potential Pareto improvement. How should such costs and benefits be determined? Cost–benefit analysis (CBA) is the collection of generally accepted methods and rules for assessing the social costs and benefits of alternative public policies. The US Flood Control Act of 1936 appears to be the first call for CBA to be systematically used to inform public policy (Steiner, 1974); it became embedded within modern welfare economics with articles by John R. Hicks (1939) and Nicholas Kaldor (1940) that set out the efficiency rationale for requiring policies to have positive net benefits. Two forces have contributed to the increased use of CBA since the 1960s. First, budget pressures and the desire to avoid inefficient regulations have led many governments cost–benefit analysis : The New Palgrave Dictionary of Economics http://www.dictionaryofeconomics.com/article?print=true&id=pde20... 1 von 8 07.09.2012 10:13 to promote, or even require, the subjection of certain types of policies to CBA. Its use in the United States, particularly in the area of economic regulation, has been mandated by a series of Executive Orders (Hahn and Sunstein, 2002). Her Majesty's Treasury in the United Kingdom publishes the Green Book to help public sector organizations apply CBA to ensure that ‘public funds are spent on activities that provide the greatest benefits to society, and that they are spent in the most efficient way’ (HM Treasury, 2002: v). Second, economists have shown ingenuity in finding ways to value goods not traded in efficient markets, thereby expanding the range of policies to which CBA can be reasonably applied. For example, the travel-cost method provides a way to value recreational facilities that charge an administratively determined entry fee (Clawson and Knetsch, 1966); hedonic pricing models facilitate valuation of spatially varying local public goods (Smith and Huang, 1995); and the development of the contingent valuation survey method, propelled by environmental damage assessment suits in US courts, permits the valuation of public goods, such as existence value, that lack readily observable behavioural traces needed for revealed preference estimation (David, 1963; Bateman and Willis, 2000). CBA promotes efficiency by identifying the set of feasible projects that would yield the largest positive net benefits. Three conceptual criticisms can be made against this proposition. First, because those who suffer costs from a policy are almost never fully compensated, CBA in any particular application generally will not guarantee a Pareto improvement. The counter-argument is that, if CBA is consistently used to select policies offering the largest net benefits and there are no consistent losers, then it is likely that overall everyone will actually be made better off. Second, the CBA techniques for measuring net benefits cannot guarantee a coherent social ordering of policy alternatives. For example, it is possible to identify situations in which moving from one policy to another offers positive net benefits as does moving back to the original policy (Scitovsky, 1941; Blackorby and Donaldson, 1990). As no fair social choice rule can guarantee a transitive social ordering (Arrow, 1963), this result is not surprising and is of minor consequence compared with the practical difficulties encountered in applying CBA. Third, and most important, only a few economists argue that public policies should be selected solely to promote the goal of efficiency. Other goals, such as equity and preservation of human dignity, are often legitimately viewed as relevant to policy choice, so that CBA is inappropriate as a decision rule. Nonetheless, as efficiency is almost always one of the relevant goals of public policy, CBA remains useful as a method for assessing efficiency in the context of a broader multi-goal analysis. Social perspective CBA assesses social costs and benefits, which distinguishes it from the self-regarding calculus of individual economic actors. The meaning of ‘social’ in this context is twofold. First, it involves the definition of the relevant society; that is, it requires a determination of whose costs and benefits have standing (Whittington and MacRae, 1986). Economists generally argue for national standing, recognizing that those in a particular country live under the same political contract, or constitution, and share a common economy with its own fiscal and monetary policy. In practice, however, sub-national governments often base their decisions only on their own costs and benefits and therefore demand CBA with standing restricted to those under their jurisdictions. Even when geographic standing is resolved, issues remain as to whether the costs and benefits of all residents – citizens, legal aliens, illegal aliens, those with legally proscribed preferences – should count (Zerbe, 1998). Second, it requires comprehensive assessment of the valued effects of policies on those with standing. cost–benefit analysis : The New Palgrave Dictionary of Economics http://www.dictionaryofeconomics.com/article?print=true&id=pde20... 2 von 8 07.09.2012 10:13 The effects are commonly divided into the categories of active and passive use. Policies affect active use by changing the observable quantities of goods consumed, such as day care or fishing. Passive use includes all those effects that cannot be readily identified with observable changes in behaviour: existence value, or the willingness to pay for some good, such as wilderness, that one never expects to consume actively (Krutilla, 1967); option value, or the willingness to pay for some good that one may wish to consume actively in the future (Weisbrod, 1964); donor value, or the willingness to pay for redistributions of goods to others (Hochman and Rogers, 1969). The absence of observable behaviour precludes valuation of passive use through the revealed preference methods most favoured by economists. Stated preference methods, such as contingent value surveys, are thus necessary for undertaking comprehensive assessments of policies with effects on passive use. Social benefits: willingness to pay A common metric for policy effects is required if these effects are to be aggregated across individuals within the relevant society. If more than one policy alternative is to be compared with the status quo, then this metric must have ordinal properties. Further, if it is to be compared with the resource costs of implementing the policy, then it must be measured in the monetary unit of the society. Equivalent variation (EV) satisfies these conditions (McKenzie, 1983). Consider the expenditure, or cost-utility, function C(U,P), where C is the amount of money needed to achieve utility U with price vector P. If U1 is the person's utility under the price vector P1 that would result from the policy change and P0 is the price vector that would result under the status quo, then the equivalent variation of the policy change is given by EV C U1, P0 C U1, P1 the difference between the expenditure needed to achieve U1 without the policy and with it. The EV is the amount of money that one would have to give to the person instead of implementing the policy so that the person is as well off as he or she would have been had the policy been implemented. A negative EV indicates that the person finds the net effects of the policy undesirable. In its actual use, CBA almost always evaluates policy effects with willingness to pay, which differs conceptually from EV. Willingness to pay answers the question: how much money could be taken away from a person in conjunction with the policy so that he or she has the same utility with the policy as without it? Rather than corresponding to EV, which holds utility constant at a level with the policy, willingness to pay corresponds to compensating variation, which holds utility constant at the pre-policy level. Although compensating variation is more intuitively appealing, it does not provide a fully satisfactory money metric like EV. The equivalent or compensating variation of a price change in a single market can be calculated as the change in social surplus as measured under the appropriate Hicksian, or utility-compensated, demand schedule. In practice, however, analysts typically work with econometrically derived demand curves that do not hold utility constant. Changes in consumer surplus measured with these Marshallian demand cost–benefit analysis : The New Palgrave Dictionary of Economics http://www.dictionaryofeconomics.com/article?print=true&id=pde20... 3 von 8 07.09.2012 10:13 curves only approximate the compensating variation, with differences driven by income effects that can be large for either large income elasticities or large price changes. Some progress has been made to put bounds on the differences between the Marshallian and Hicksian measures (Willig, 1976; Seade, 1978), but these bounds are rarely applied in practice. The interpretation of Marshallian consumer surplus as willingness to pay becomes even more complicated when policies have secondary effects in the markets of complements and substitutes of the goods primarily affected by policies. Although a general equilibrium model would be most appropriate for taking account of these secondary market effects, common practice is to approximate the combined effect of the primary and secondary markets by measuring surplus changes with the use of an estimated demand schedule for the primary market that does not hold the prices of substitutes and complements constant (Sugden and Williams, 1978; Gramlich, 1990; Boardman et al., 2006). In such cases, analysts need not account for price changes in undistorted secondary markets. Indeed, doing so would likely result in double counting of benefits. Social costs: opportunity costs Public policies generally require the use of real resources to produce their effects. The guiding principle for monetizing the forgone value of these resources is opportunity cost: what is the value of the resources in their next-best use? That is, what is the value forgone by using the resources for the project? When factor markets are undistorted and the additional demand created by the project does not increase price, the opportunity cost of the resource just equals its market value, which, if the resource is obtained by purchase, just equals the expenditure on the resource. When factor markets are undistorted but the additional demand induced by the policy drives up price, then the opportunity cost of the resource equals the sum of expenditure and the change in social surplus, the algebraic sum of the change in consumer surplus and the change in rents usually measured as change in producer surplus based on the short-run supply schedule (Mishan, 1968). For example, if supply and demand curves are linear, then the opportunity cost equals the average of the pre- and post-purchase prices of the resource times the quantity purchased. If markets are distorted, then even if price does not change the opportunity cost does not necessarily equal the expenditure required to secure supply. For example, a common factor-market distortion is involuntary unemployment resulting from minimum wages imposed by either law or custom. The expenditures needed to hire workers from a market with involuntary unemployment for a project clearly overestimate the opportunity cost of this labour. Nonetheless, the opportunity cost is almost certainly not zero, as sometimes argued by policy advocates, because the time of the workers hired by the project has an opportunity cost in terms of forgone leisure and household production. Accommodating uncertainty CBA requires prediction of the effects of adopting a policy. Predictions are inherently uncertain. In addition to uncertainty about such parameters as price elasticities required for predictions of changes in social surplus, CBA often requires analysts to confront fundamental uncertainty about future states of the world. For example, preparing a vaccine to guard against a potential pandemic is costly but offers large benefits in the event that a pandemic actually materializes. CBA requires analysts to convert these cost–benefit analysis : The New Palgrave Dictionary of Economics http://www.dictionaryofeconomics.com/article?print=true&id=pde20... 4 von 8 07.09.2012 10:13 uncertainties into risks by specifying representative states of the world and assigning probabilities to these states. Common practice is to model the policy choice as a decision analysis problem, or game against nature, and to choose the policy that maximizes the expected value of social surplus. A more conceptually valid measure of the benefits of a project with certain costs in the face of risk about the future state of the world is option price (Graham, 1981). Option price answers that question: what is the maximum certain payment that an individual would be willing to make to obtain the project? The sum of these certain payments for all those with standing can then be compared with the certain cost of implementing the policy. In general, however, option price does not equal the expected value of an individual's surplus over the possible states of the world; it differs from expected surplus by the option value of the policy for the individual. Although contingent valuation surveys seek to elicit individuals' option prices directly, more commonly analysts estimate benefits as expected surpluses, and consider option value as an excluded value. Some progress has been made in signing option value (Larson and Flacco, 1992), but analysts rarely have enough information for confidently including it as a monetized correction to expected surplus. Discounting for time Policies typically have effects that extend far into the future. Infrastructure projects in particular are usually characterized by large initial investments followed by beneficial use over years or even scores of years. CBA requires that costs and benefits accruing in the future be converted into their present value equivalents. On the assumption that future costs and benefits are predicted in real dollars, then a dollar of cost or benefit occurring t periods beyond the present equals in present value terms 1/ 1 d t where d is the real discount rate for the period length. In practice, discounting is usually done on an annual basis. As valid comparison of projects requires that they be assessed over the same time horizon, it is often necessary to convert present values to equivalent perpetual streams of constant values through the use of an annuity factor. The appropriate value for the real discount rate remains controversial. One approach is to set the discount rate equal to the marginal rate of pure time preference, the rate at which consumers are indifferent between exchanging current for future consumption. Another approach is to set the discount rate equal to the opportunity cost of capital, the marginal rate of return on private investment. In an ideal capital market these two rates would be equal. In the presence of transaction costs and taxes, however, these rates differ substantially. For example, an estimate of the marginal rate of pure time preference based on the after-tax real rate of return on US treasury bonds is 1.5 per cent, while an estimate of the opportunity cost of capital based on the expected real yield on AAA corporate bonds is 4.5 per cent (Moore et al., 2004). If all costs and benefits correspond to changes in consumption, then the marginal rate of pure time preference is the appropriate discount rate. Instead, if all costs and benefits correspond to changes in private investment, then the marginal rate of return on private investment is the appropriate discount rate. cost–benefit analysis : The New Palgrave Dictionary of Economics http://www.dictionaryofeconomics.com/article?print=true&id=pde20... 5 von 8 07.09.2012 10:13 However, most projects involve changes in both consumption and investment. The shadow price of capital approach involves expressing all costs and benefits in terms of consumption changes so that the marginal rate of pure time preference can be applied (Bradford, 1975). In application, this means applying a shadow price to changes in private investment so that they are converted to the present values of their associated streams of consumption changes. Shadow prices Much of the challenge of CBA lies in deriving appropriate shadow prices when policies have effects beyond those that can be taken into account as changes of prices or quantities in undistorted markets. In developing countries, for example, import and export controls and the presence of subsistence agriculture often distort virtually all prices, necessitating the determination of a complete set of shadow prices based on prices in international markets (Little and Mirrlees, 1974; Squire and van der Tak, 1975; Dinwiddy and Teal, 1996). Economic research provides a number of shadow price estimates that can be used in conducting CBA. Indeed, were these shadow prices not readily available, the plausible range of application of CBA would be much narrower. One of the most commonly needed shadow prices is the value of a statistical life. That is, what is the willingness of a representative member of a population to pay for reductions in mortality risk? Economists have used a variety of methods to estimate the value of a statistical life, most commonly taking advantage of differences in risks and wages across occupations or the purchases of safety devices. The number of studies is sufficiently large that a number of meta-analyses have been conducted to develop estimates of the value of a statistical life for the United States in the range of roughly $4 million to $6 million in 2002 dollars (Miller, 2000; Viscusi and Aldy, 2003). Tied to any estimate of the value of a statistical life is the value of a life year. Health economists have developed a number of methods for e