PERSPECTIVES ON BENEFIT-COST ANALYSIS: EFFICIENCY AND NON-EFFICIENCY CONSIDERATIONS
The wide and varied history of benefit-cost analysis motivates any attempt to develop a framework for project evaluation to consider the development and limitations of the benefit-cost analysis. In developing such a perspective, we can begin the process by addressing the first question raised in the introduction. That is, what is the appropriateness of benefit-cost analysis and further are there limitations that might require a different Wyoming perspective from that which is traditionally accepted?
As such, Section 2.2 presents the philosophical underpinnings of our analytical approach. Section 2.3 reviews the historical uses of benefit-cost analysis. In developing the outlines of our approach, we consider the notion of efficiency per se in Section 2.4. Finally, in Section 2.5, non-efficiency considerations are discussed and set in perspective regarding non-market considerations.
2.2 PHILOSOPHICAL UNDERPINNINGS OF THE STUDY'S ANALYTICAL APPROACH: BENEFIT-COST ANALYSIS VERSUS BENEFIT-COST MEASURE
Our analytical approach is somewhat different from what has seemingly become the "traditional" approach in benefit- cost studies. We stress the difference between procedures followed in the estimation of a benefit-cost measure, and those relevant for benefit-cost analysis.
The benefit-cost measure is an indicator of the economic efficiency of a project, and includes those beneficial and adverse effects of a project which can be expressed via a dollar metric. Benefit-cost analysis involves the more comprehensive consideration of all effects on welfare which are attributable to the project. Benefit-cost analysis reflects a broadly defined set of multiple objectives of the state, many of which are not quantifiable in dollar terms. Examples include project effects on the distribution of income and economic opportunities, environmental considerations, and the preservation and conservation of the state's rights to unused water resources.
During the early development of benefit-cost analysis as a sub-discipline in welfare economics, the limits of a benefit-cost measure in terms of providing decision makers with a comprehensive analysis of the social and economic impacts of a water project were recognized (see Section 2.4). A comprehensive framework required analyses of socioeconomic impacts beyond those relevant for assessing the economic efficiency of a project. For instance, Krutilla and Eckstein  stressed the need to look beyond incomes in the efficiency measure to project effects on the distribution of income and the growth in economic opportunities.
Krutilla and Eckstein's argument is that economic analysis cannot "solve" all of the decision makers problems. The efficiency criterion may well involve a conflict with other criteria, and only the decision maker can make the value judgments required to resolve these conflicts (Krutilla and Eckstein , pp. 49-50).
This philosophical position is echoed in three seminal works which appeared in the early 1960s: the 1962 Harvard water study (Maass et al. ); the 1966 report of the Western Resources Conference (Kneese and Smith ); and Bain et al.'s 1966 study of the "anatomy" of a water industry (Bain, Caves, and Margolis ). Maass argued that:
. . . the objective functions of most government programs are complex; yet benefit-cost analysis has been adapted to only a single objective-economic efficiency. Thus, benefit-cost analysis may be largely irrelevant, or relevant to only a small part of the problem of evaluating public projects and programs (Kneese and Smith , p. 312).4
Many economists retreated from the view that project efficiency is but one component in an array of descriptions of project impacts.5 This retreat may be attributable to economist's concern with the potential for nefarious uses of non-efficiency criteria for the simple purpose of justifying inefficient projects:
. . . we are frankly concerned about the potential effect that the move to multiple objective planning . . . will have on the federal water resources budget. The incentives to cast about for new . . . objectives . . . for a public works program . . . are obvious (Freeman and Haveman , p. 1537).
Further, Kneese, a proponent of multiple objective planning argued in 1968 that:
What I really fear is that many projects which are both inefficient . . . and poor vehicles for income distribution will be found justifiable on income distribution grounds . . .. Indeed, poverty can become every special-mission-oriented agency's hobbyhorse (Kneese , p. 66).
Others might argue that this retreat reflected the profession's "discovery" of mathematics, and their frustration with their inability to "cram" non-efficiency impacts into the benefit-cost measure (Bromley ). Regardless, benefit- cost analysis in the 1970s and 1980s, as conducted by a large part of the economics profession during this period, involved primary if not sole focus on the development of a benefit-cost measure, with little more than lip service being paid to the non-efficiency aspects of a water development project.
Notwithstanding underlying concerns with the misuse of secondary benefits, the apex of this trend is seen in its most developed form in the Water Resources Council's 1973 "Principles and Guidelines." Federal policy became effectively an argument for the supremacy of the benefit-cost measure ("National Economic Development").
The analytical approach we adopted takes up the calls by such authors as Hanke and Walker , Ng , and most recently (and perhaps most eloquently) Bromley , to push the analytical pendulum from benefit-cost measures back in the direction of comprehensive benefit-cost analysis. Concern with the limitations of benefit-cost measures is, of course, not limited to economists. At the 1986 Conference of the Universities Council on Water Resources, the university community exhorted the federal government to reexamine its role in national water and related resources programs to the end of:
Our central theme is that legislative and public mandates for water development in Wyoming require considerations and assessments which extend beyond the criterion of economic efficiency. Benefit-cost analysis, as a comprehensive analyses of diverse and potentially conflicting components, is not problem free.
The result from our approach is an array of information of the form:
Developing methods for "trading off" these various considerations is quite difficult, however. For instance, trade-offs may involve situations where the benefit-cost measure is "high," but so are adverse environmental effects and/or adverse effects on the distribution of income; or the project substantively contributes to the security of Wyoming's water rights and has desirable effects on the distribution of income, but the benefit-cost measure is "low." (See Appendix F for an illustration of this problem.) Noting the desire for a single integer that might serve as a yes or no indicator, we know of no way that this can be accomplished:
. . . the selection of appropriate water projects is a political process, no matter how deeply hidden the political choices are beneath the complex analytics of benefit-cost analysis. Benefit-cost analysis may well be a useful administrative tool for organizing and utilizing technical and economic information . . . but it should not be asked to do more than it can reasonably do (Hanke and Walker , p. 907).
2.3 HISTORICAL USE OF BENEFIT-COST MEASURES
Benefit-cost analysis as a new branch of welfare economics was developed in response to the mandates of the 1936 Flood Control Act.6 The provision of the 1936 Act required analyses of water projects to encompass benefits "to whomsoever they may accrue." This left a great deal of latitude for the consideration of project impacts. During the 1950s and 1960s, benefit-cost measures became increasingly inflated by the dominance of "secondary benefits" (see Burness, et al. ). The inflation of benefit-cost measures via secondary benefits is reasonably attributed to concerns for extra-economic aspects of water projects (see Mann et al. ; North ; and Stoevener and Kraynick ). The result was a growing disenchantment with benefit-cost measures resulting in an impression that benefit- cost measures were little more than a "game" played by the Bureau of Reclamation (Burness et al. ). Martin et al.  describes the role of benefit-cost measures as providing measures of Western water users' "willingness to play," as opposed to their "willingness to pay" for water projects.
The view that benefit-cost measures were being used to justify projects is reinforced by the manner in which the Bureau of Reclamation has treated "reimbursable" and "nonreimbursable" costs. Reimbursable costs are project costs which are to be repaid by beneficiaries of reclamation projects. Nonreimbursable costs are attributable to project features which result in "public benefits." Since such costs cannot be assigned to specific, identifiable beneficiaries they accrue to the public at large and are absorbed by the federal government and are therefore "nonreimbursable."7
The "game" was not limited to reassignment of capital costs. The Reclamation Act of 1939 provides for subsidies of capital costs to agriculture, based upon agriculture's "ability to pay" (Burness et al. ), and at a minimum water users are expected to pay all operating and management (O&M) costs. Thus, if benefits "promised" in ex-ante benefit- cost measures are related to actual benefits then benefits should cover O&M costs of the project. This simply did not occur.8
However, Franklin and Hageman in noting the divergence between expected and actual project "performance" find justification in the role of water projects:
water resources have and continue to play a major role in the provision of economic opportunity in western states; the value of this role is seen as extending well beyond economic measures and the ability to pay of direct water users (Franklin and Hageman , pp. 1050-1051).
Thus, water planners, lacking objective measures for broad community objectives related to water reclamation, may view the use in benefit-cost analyses of values higher than those which will likely result- from the project as the only means by which these extra-economic values can be reflected in the benefit-cost measure.
The argument developed above relates directly to our central theme: one should not expect so much from the benefit-cost measure. If water planners recognize and accept the limitations of benefit-cost measures, the benefit-cost measure can be objectively derived. Other project effects, rather than being "crammed" into the benefit-cost measure, are presented as data of equal importance with the benefit-cost measure for the purpose of project evaluation.
Extra-economic impacts can be presented as part of an analysis if one follows the Water Resources Council's "Principles and Standards" (38 Fed. Reg.) published in 1973, but are displayed in sections distinct from the derivation of benefit-cost measures. While a number of inconsistencies exist in the rules of the Principles and Standards (see, e.g., Burness et al. ), the comprehensive, multi-objective display of all impacts from a project, wherein the benefit- cost measure was but one datum, is precisely the approach which we argue here is appropriate for analyses of water reclamation projects. Regrettably, however, and reflecting (we posit) a general lack of appreciation of the limits of benefit-cost measures, the perception remains that the benefit-cost measure is somehow more important than the information displayed in other "accounts" of the Principles and Standards analytical framework.
2.4 PROBLEMS WITH EFFICIENCY
This section is focused on how one might define a framework for assessing water projects in Wyoming. We argue that a wide range of social, economic, and political considerations are relevant for the decision making process. The "traditional" benefit-cost measure can provide, and should be expected to provide no more than, a limited amount of information relevant for this process. In arguing this viewpoint, we consider directly the notion of efficiency.
Relevant for the emerging field of benefit-cost analysis, was the focus of welfare theorists on what might be referred to as a "truth rule" or "efficiency rule" which could be used to distinguish between "good" and "bad" social projects(Hicks , Kaldor , and Scitovsky ). 9
The stage was then set for the implementation of benefit- cost analysis. Benefits consisted of gains in income or income-equivalents to beneficiaries of a project. Costs would include losses in income or income-equivalents. A project could be judged as "good," and thus was efficient, when benefits were at least as great as costs.
The efficiency objective became viewed as value-free. Moreover, many economists argued that they do not "advocate" any particular policy, but indicate (with the benefit-cost measure) to the decision maker(s) what would be "efficient." Thus, the following dilemma. If the economist's advice is ignored, is the decision maker "inefficient"? Some economists argue that if a choice is then made which ignores the efficient choice, the choice is then obviously a political choice in which distributional issues dominate and the economist remains an objective scientist. Bromley argues that this subtle illusion will not wash:
To suggest to a decision maker the course of action that would be "efficient" is to load the debate in an unsavory way. No one, not even the much maligned public decision maker, knowingly wishes to be "inefficient"; the problems arise in defining efficiency .... Decision makers . . . understand that most public policy is about either reallocating economic opportunities, or redistributing economic advantage . . . . When the economist suggests that something would be "efficient," it is likely understood by decision makers to be that policy change which will effect a certain reallocation . . . . The economist, of course, means something quite different . . . (Bromley , p. 26).
Clearly, the measure is not value-free. The measure totally abstracts from considerations related to the distribution of income such as who are the gainers and who are the losers, and how are they affected by such gains or losses. Thus, the efficiency test ". . . is optimal (only) with reference to those value judgments that are consistent with the Pareto principle" (Ng , p.30). Furthermore, acceptance of the given distribution of income must be acknowledged.
However, economists have long been concerned with the potential disservice to the public of socioeconomic analyses which are based solely on the efficiency criterion studies which seemingly shroud themselves in the cloak of "scientific objectivity." Such analyses are characterized by Bromley as "Bogus Science, Bad Advice":
Still uncomprehended, apparently, is that it is a value judgment to claim that economic efficiency ought to be the decision rule for collective action . . . . Economists who have persevered in this tradition seem content to overlook the logical inconsistencies in welfare economics; this obduracy apparently being justified on the grounds that a little economic analysis-even if indefensible on theoretical grounds, and therefore bogus-is better than a political process left to its own devices (Bromley , p. 20).
The limitations of benefit-cost measures, and its appropriate role in the process of project evaluation is argued by Hanke and Walker in the following way.
Economists have erroneously placed the onus of political bias on government agencies and their application of benefit-cost and have felt that if only benefit-cost could be perfected theoretically and applied impartially, decisions would necessarily improve. This 'incremental' strategy has not borne fruit. Furthermore, it may actually do real harm. . . . The public and its representatives . . . who innocently believe that the single-number ratio represents economic truth are thereby excluded from the actual decision-making process (Hanke and Walker , p. 907).
Thus, benefit-cost measures have important limitations. A wide range of political, social, and economic values are relevant for the analyses of public projects such as water projects. The benefit-cost measure, which includes only those aspects of a proposed project which are amenable to income- like measures, is itself value-based. The value reflects only economic efficiency. Thus, the appropriate use for benefit- cost analysis is that of providing information as to one aspect of a project which can be useful for the political process of project selection. Finally, the appropriate measure of success for a comprehensive socioeconomic analysis of a proposed project is the extent to which the study identifies the relevant range of impacts from a project. These impacts may extend well beyond those included in the benefit-cost measure into the benefit-cost analysis arena. 10
2.5 PERSPECTIVES ON NON-EFFICIENCY AND NON-MARKET
Given our distinction between the benefit-cost measure and the benefit-cost analysis, a comprehensive framework for evaluating Wyoming water projects should be more than the benefit-cost measure. In this section, we want to briefly discuss what this implies and what it does not imply. The discussion is divided into two components: (1) non-efficiency considerations and (2) non-market considerations. Further, in discussing these elements of a benefit-cost analysis, we consider in Appendix F what are the implications for integrating the array of information that constitutes a benefit-cost analysis.
2.5.1 Distributional Considerations
A central aspect of the non-efficiency considerations of benefit-cost analysis pertain to distributional effects. Distributional considerations take the analysis far from the traditional notions of efficiency. In examining the potential distributive effects of a water project, the concern focuses on the following question: who are the recipients of benefits and who are the people who bear the costs associated with the project?
In general terms, the difference between a project whose benefits accrue primarily to wealthy households and a project whose benefits accrue primarily to low income households may be relevant. Thus, one aspect of the distributional issue is the distribution of benefits and costs relative to income levels. Another distributional consideration might be the spatial distribution of the projects throughout the state. That is, suppose that all of the projects were located in the southeast corner of the state. Clearly, this would potentially lead to a differential in economic activity in the near and long term. Yet, another consideration might be the many possible users of water projects. It might well be relevant to ask if the distribution of the water from an irrigation project goes to a very small number of ranchers or is in fact distributed across a larger group. Finally, the distribution of water uses across all of the water projects might be relevant. That is, will the projects only serve the agricultural sector or will, in fact, the projects provide additional water for a multitude of uses.
In considering non-efficiency notions (distributional impacts), one is raising the possibility that benefits and costs may not be distributed in an "equitable" manner. Further, one might well be creating a situation whereby efficiency considerations are in "conflict" with non- efficiency considerations. To the extent this issue is important is ultimately dependent upon the views of the citizens of Wyoming regarding the relative importance of distributional effects vis-a-vis other effects or considerations. Thus, our consideration of distributional effects and their relative importance motivates one aspect of the survey (Chapter 3). That is, if given the trade-off between efficiency considerations and non-efficiency distributional considerations, how would the citizens of Wyoming choose?
2.5.2 A Spectrum of Other Impacts
If one were to envision a spectrum of considerations that arguably should go into a benefit-cost analysis, efficiency measures would be at one end of the spectrum and non- efficiency measures such as distributional effects would be towards the other end of the spectrum. The primary differentiation between these two types of measures lies in the ability to place the effects in income or monetary terms.
Between these two ends of the spectrum lies a set of considerations that fall in-between in the sense that some of the considerations can be measured in national income terms while others may not. Typically these are referred to as non- market effects or considerations. For the state of Wyoming these type of effects or impacts are well represented by the array of natural resources that the state is endowed with. For instance, there is an abundance of wildlife and many rivers and streams that remain in their natural state. Water projects potentially can change the availability of and the nature of these resources.
A significant literature has developed over the years regarding the economics profession's ability to quantify changes in these resources. As this literature is well documented we will not detail the methods at this point. The extent that changes in these resources can be represented in income terms is discussed by Cummings et al. , Mitchell and Carson  and numerous other authors.
This is not to say, however, that the analyst is relieved from best efforts to provide monetary measures for project impacts which are amenable to such measures. For objectives of water projects which are non-market in nature such as the protection of water resources from claims of downstream states, one can take advantage of advances made over the last two decades (see, e.g., Cummings et al. ; Folmer and van Ierland ; and Peterson et al. ).
An example of an application of the contingent valuation method for valuing the protection of Wyoming's water resources from possible future claims of downstream states is seen in a recent assessment of the Sandstone Project in Wyoming (Watts, Brookshire and Cummings ). While, in many cases, the usefulness of values for non-market impacts of a reclamation project may be limited to that of providing order-of-magnitude insights of relevant values, and may therefore not be meaningfully summed in the benefit-cost measure, their derivation and presentation along with descriptive information of a project impact can provide decision makers with useful insights as to the income implications of the impact in question. If the impacts can not be represented in income terms then other methods will have to be used. For instance, one might represent wildlife impacts through a study of herd size changes. 11
90-12 Table of Contents
Water Resources Publications List
Water Resources Data System Library | Water Resources Data System Homepage