Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
SBenefit Assessment INTRODUCTION If EPA determines that a pesticide meets or exceeds the risk criteria defined in 40 CFR 162.11 (see Chapters 2 and 4), the Agency is obligated to issue an RPAR. If the presumption of risk is not successfully rebutted, some type of regulatory response is likely to be forthcoming. In making a determination as to the specific form the response should take, the Agency is allowed by 40 CFR 162.11 to take into consideration the findings of an analysis of the benefits arising from the use of a pesticide. Moreover, analyses of the economic and social impacts of pesticide regulatory actions are required by the FEPCA of 1972 and its subsequent amendments. The purpose of this chapter is to examine critically the current USDA/EPA approach to assessing the benefits of pesticide usage and, where appropriate, to recommend certain changes in that approach. This chapter argues that oPP's benefit-risk methodology could be improved by better use of standard analytical procedures. At the outset, we wish to make it clear that federal pesticide law does not require the use of a formal benefit-cost (or risk) analysis. Except where a statute expressly so provides, the courts have not interpreted a statutory duty to balance benefits against costs to require a formal benefit-cost (or risk) analysis. Instead, agencies like EPA have taken it upon themselves to adapt formal benefit-risk methodology to the regulatory problem at hand. The difference between a duty to carry out a formal benefit-risk 99
100 REGULATING PESTICIDES analysis and to balance benefits against risks is significant, for the degree of rigor required by the latter is much less than that required by the former. If an agency undertakes a formal benefit-risk analysis, it should use the best available state of the art; we offer some suggestions for improving the procedures that oPP has voluntarily agreed to follow. Before turning to a critical evaluation, two introductory and clarifying comments are in order. The first is definitional. In connection with the evaluation of government regulations, the term "benefit" is usually employed to describe the advantages of a regulation such as the improvement in human health that will result from prohibiting the use of a hazardous pesticide (see, for example, NRC 1977~. Consequently, the reader is cautioned that USDA/EPA defines benefits differently. Spe- cifically, "benefits" are defined as desirable erects resulting from continued use of a pesticide. In this context, a regulatory action limiting the use of a pesticide would result in a reduction in risks and a loss of benefits. This report adheres to the definitions employed by the USDA/EPA analysts. The second point is concerned with the presumed objectives of a benefit assessment. In keeping with conventional principles of benefit- cost analysis (which are discussed in more detail later in this chapter), a benefit assessment should strive to meet two separate but related objectives. First, a benefit assessment should attempt to measure the "real" (or economic efficiency) benefits of a pesticide that is, the extent to which use of a pesticide contributes to the available quantity of desirable goods and services, thereby enhancing society's standard of living. Second, a benefit assessment should attempt to identify and quantify the distributional erects (or "economic impacts") associated with use of a pesticide. In principle, a distributional (or economic impact) analysis involves more than a determination of how the real benefits are spread among various groups. It also encompasses an assessment of the distribution of "pecuniary" erects (that is, transfers of purchasing power from one group to another arising out of price changes). The distinction between economic efficiency erects and distributional impacts is important and will be emphasized in our . · . olscusslon. This chapter is devoted to procedures and methods for assessing the benefits of using particular pesticides; since most pesticides (almost 75 percent by volume, U.S. EPA 1979b) are used in the production of food and fiber, the discussion is framed primarily in terms of the agricultural uses of pesticides. This is not to say that the Committee believes nonagricultural uses do not pose significant human health and environ- mental risks. Rather, the Committee had to limit the scope of its review
Benefit Assessment 101 and chose agricultural uses since oPP devotes a majority of its effort to this area. The discussion is divided into the following major parts. The first section considers the problems associated with measurement of pesticide productivity, that is, the extent to which use of a chemical increases production of things people value. The second section focuses on the problem of estimating how much the use of a specific pesticide reduces pest control costs and other production costs. These two components pesticide productivity and cost savings-are brought together in the third section, which presents appropriate methods for measuring the real benefits of pesticide usage and for distributing both real and pecuniary erects among certain key groups. Our critical evaluation of the methods currently employed by USDA/EPA economists to assess benefits of pesticide use is based upon a review of several actual benefit assessments, namely, those for chloro- benzilate (Luttner 1977a, by, DBCP (U.S. EPA/USDA 1978a), endrin (Luttner 1977c, Mattson et al. 1977), lindane (U.S. EPA/USDA 1978c), toxaphene (U.S. EPA/USDA 1978b), and trifluralin (USDA et al. 1977~. These six assessments consist of fifty-eight separate use-pattern assess- ments or analyses, with each analysis focusing on a specific use of a pesticide. For instance, the trifluralin assessment consists of six separate analyses for uses on cotton, soybeans, other field crops, fruits and vegetables, miscellaneous crops, and noncrop sites. ANALYSIS OF PESTICIDE PRODUCTIVITY Pesticides are toxic chemicals used to kill a variety of organisms (e.g., rodents, insects, pathogens, and weeds) that people consider objection- able for any one of several reasons. Many organisms are viewed as pests because they interfere with agricultural and forestry production. Others transmit diseases to humans and thus pose a hazard to public health. Finally, some organisms, such as household lawn and garden pests, create nuisances or aesthetic problems. For whatever reason, control of pest organisms offers certain advantages, such as improvements in food and fiber yields and the quality of products or reductions in the incidence of human disease. A regulatory decision to prohibit or otherwise restrict the use of a pesticide may consequently force society to forgo substantial benefits. A crucial first step in analyzing benefits offered by a pesticide is an assessment of that pesticide's productivity that is, its electiveness at providing something that people value, such as higher crop yields. Of course, if a pesticide has a detrimental impact on nontarget organisms
102 REGULATING PESTICIDES resulting in negative productivity, such an impact must also be included in an assessment of the pesticide's ejects. Pesticide productivity is not the same as pesticide efficacy. "Efficacy" refers to the pesticide's effectiveness at reducing a pest population, whereas, in this context, "productivity" refers to the pesticide's electiveness at providing more or better food and fiber, improvements in public health, or more attractive surroundings. Thus, a highly efficacious pesticide would be relatively unproductive if its target organism actually had little eject on, say, crop yield or quality. Pesticide efficacy and productivity are related but quite clearly different concepts. This part of the chapter focuses on the issue of pesticide productivity. CURRENT APPROACH: METHODOLOGY AND DATA The USDA/EPA procedure for evaluating the productivity of an agricultur- al pesticide concentrates on quantifying the pesticide's ejects on crop yield or output. The possibility that quality ejects may also be important is, of course, recognized in the USDA/EPA benefit assessments; however, data are generally inadequate to quantify such ejects. Post-harvest losses resulting from pre-harvest infections by pests are also often omitted because of lack of data. In estimating yield ejects, an attempt is made to measure the change in output (e.~.. bushels of corn reaching the marketplace) that would rid- ~ occur if a particular chemical were withdrawn from the market and replaced by alternative chemical or nonchemical methods of pest control. The estimation requires two basic types of data: (1) an indication of the extent to which each of the various alternatives would be employed as a substitute for the suspect chemical, and (2) an estimate of differences in yield or output between the suspect chemical and the alternatives. Unfortunately, these data are not always available. Conse- quently, the benefit assessments are sometimes forced to consider these important aspects of productivity in qualitative terms only. For instance, 18 of the 58 use-pattern assessments reviewed for this chapter contain no quantitative evaluation of the impacts of cancellation. Primary responsibility for assembling evidence on the productivity of a pesticide and its alternatives falls to the USDA/EPA benefit assessment team (see Chapter 2~. More specifically, the biologists (e.g., entomolo- gists, plant pathologists) on the assessment team are responsible for providing estimates of yield and, if possible, quality ejects. Often these data are developed from published sources, but the team biologists may also rely upon their own judgment or upon other unpublished sources such as personal communications with other pest control specialists.
1 Benefit Assessment RECOMMENDED APPROACH 103 The measures of pesticide productivity that the USDA/EPA benefit assessments attempt to implement are basically sound; that is, the assessments define pesticide productivity correctly. Of course, it does not follow that the actual estimates of pesticide productivity will be accurate. As is noted at various points in this chapter, incomplete data often prevent the analyst from making more than a very crude estimate of the productivity of a pesticide. There are some problems with the current method of predicting the use pattern that would arise among the alternatives were a suspect chemical to be withdrawn; however, discussion of the problems are deferred until later in this chapter. Another methodological problem arises not only in connection with estimating pesticide productivity, but also throughout most of the benefit analyses. Specifically, USDA/EPA estimates of yields (and of other benefits) are usually expressed as clearly determined magnitudes even though there may be considerable uncertainty about their accuracy. The practice lends an unjustifiable aura of precision to the benefit estimates. In general, an uncertain measure should be reported as an interval a probable estimate with upper- and lower-bound estimates rather than as a single number. The central recommendation specific to the measurement of pesticide productivity relates to the current procedure for assembling the yield and quality data. Benefit assessments currently place excessive reliance on "data," often unpublished, and sometimes contained in controversial reports, that have not been subjected to conventional scientific tests of validity. The credibility of a benefit analysis depends ultimately upon the credibility of the data that support it. Of course, the regulatory process cannot await the generation of a complete set of sound scientific data. However, the regulatory process can, and should, demand that benefit (and risk) analyses be performed with the best available data and with data that have withstood some scrutiny. The problem with the current procedure for assembling benefit data is illustrated by the USDA/EPA analyses for the pesticides chlorobenzilate and dimethoate. Data for these analyses were obtained mainly from unpublished reports and personal communications with pesticide spe- cialists. A few relevant published studies were identified by oPP with the aid of several computerized bibliographic indexes. However, literature searches commissioned by this Committee identified numerous impor- tant published studies that were either missed or intentionally omitted by the USDA/EPA procedure (see Chapter 7 and Appendix D). Moreover, the
104 REGULATING PESTICIDES independent literature searches revealed that the data obtained through personal communications on balance differed significantly from the published information. For several pests and crops, the chlorobenzilate and dimethoate data derived from unpublished sources tended to overstate the benefits of the suspect chemical. The recommended procedure for assembling an acceptable data base for a benefit analysis involves the following steps. First, a thorough search of the published literature must be undertaken. The literature search should use all major indexing bibliographies, and recent issues of relevant, but unindexed, scientific journals should also be examined. Computerized indexing services may be of some supplemental use in this literature search, but they are too far from being complete to be relied upon as the sole guide to relevant literature. Whenever feasible, data on use of the suspect chemical and its alternatives outside the United States should also be assembled. Although it may not be appropriate to use such data directly, they may provide helpful support to the data based on U.S. experience. After the published literature has been searched thoroughly, it might prove useful to consult some unpublished sources in the form of unpublished reports or perhaps the firsthand knowledge of pest control specialists. Each specialist contacted should be informed of the relevant published information concerning the pest or pesticide and encouraged to think in terms of a total accounting of all major losses of the commodity in question due to pests. Presumably, the broader perspective will help reduce the likelihood that the yield or quality elects attributed to any one pesticide or to any one pest will be either exaggerated or understated. The problem of overestimating yield erects is especially troublesome. The pest control literature offers examples in which estimates of aggregate yield losses from the combined erect of insects and other pests exceed 100 percent (Pimentel et al. 19784. Once data on productivity effects of both chemical and nonchemical alternatives have been assembled, such data should be further validated for thoroughness and accuracy through critical internal and external reviews by knowledgeable scientists. At present, it is the legal responsi- bility of the SAP to provide an external scientific review of oPP's health and environmental hazard assessments. In practice, the SAP has also provided some review of the benefits analysis (see SAP comments published with EPA'S notice of final determination on chlorobenzilate, U.S. EPA 1979a), but the arrangement results in inadequate reviews of the benefits data and analyses. Since the involvement of the SAP iS limited largely to the final stages of the RPAR process (see Chapter 2), the SAP iS confronted with the task of reviewing analyses that are virtually
Benefit Assessment 105 complete and thus unlikely to be changed in any significant way. The usefulness of such reviews could be enhanced greatly through earlier SAP involvement in the analyses. An additional problem with the current scientific review procedure is that the SAP was created, and its membership chosen, primarily to provide a review of risk-related issues. Consequently, benefits data and analyses appear to receive relatively little attention from individuals chosen for their expertise in the benefits area. For these reasons, the Committee makes the following recommenda- tion: · oPP should establish an external Benefits Review Panel, similar in organization to the SAP, consisting of entomologists, plant pathologists, weed specialists, economists, and others with expertise in the assessment of benefits. The panel would have the responsibility of providing external scientific reviews of the benefits data and analyses. For each REAR compound, a review team (consisting for example of one entomologist and one economists should be selectedirom the panel. This team, in contrast to current SAP procedures, should be involved from the earliest stages of the benefits assessment, and should have primary responsibility for presenting an evaluation of the assessment to the entire Benefits Review Panel. ESTIMATING CHANGES IN PEST CONTROL COSTS CURRENT APPROACH The second key component of a benefit analysis is an assessment of the erect that withdrawal of an RPAR chemical would have on pest control costs. (Other costs of cultivation or production may also be important and are considered later in this chapter.) The operational definition that oPP attempts to implement for each specific use of the suspect chemical is (assuming the treated item is a crop): /` P Ci = id, 2\ [A jTj ~ MCj + A Cj) ~ ~ 1 K ~[A kTk( MCk + A Ck) ], k I where /`PCi is the change in pest control costs for the ith use of the RPAR chemical (e.g., a particular crop in a certain region); j denotes one of the J pest control methods not involving the suspect chemical; k indicates one of the K control methods employing the suspect chemical; Aj and Ak are the number of acres treated per year by methodsj and k, respectively (sometimes referred to as "base access; Tj and Tk are the average
106 REGULATING PESTICIDES number of treatments per year by methods j and k, respectively; MC and MCk are material costs per acre-treatment; and ACj and ACk are application or treatment costs per acre-treatment. The data for implementing this measure come from a variety of sources. Much of the information (e.g., the number of acres treated) is supposed to be developed by the USDA/EPA benefit assessment team. Although it is difficult to generalize about the data sources employed by the assessment teams, there appears to be fairly extensive reliance upon unpublished sources and the judgments of individual team members or other specialists. In certain instances some of the data are obtained from surveys, especially those provided by Doane Agricultural Service, Inc. Finally, some of the data used are simply plausible assumptions. For instance, benefit analyses commonly suppose that the number of acres treated would not be affected by withdrawal of the RPAR chemical (e.g., see Luttner 1977a). RECOMMENDED APPROACH For the most part, the current approach to estimating the change in pest control costs is sound, but there are some aspects that should be altered. First, the imprecision of many of the estimates is obscured by the practice of reporting point estimates rather than interval estimates. As a general rule, oPP's benefit assessments should be more forthright about the uncertainty surrounding the estimates by reporting plausible mini- mum and maximum values along with the most-probable estimates for key variables. A second recommendation relates to the current practice of estimating material and application costs with data obtained from a variety of sources. For example, oPP's chlorobenzilate benefit assessment (Luttner 1977a, b) employs information from the Doane Speciality Crops Survey, dealer price lists, and pest control specialists (especially those serving on the assessment team). Estimation of changes in pest control costs requires information about differences in material and application costs. Whenever feasible, these cost differentials should be estimated from a single, consistent set of data (e.g., dealer price lists) rather than from data generated from several sources. Estimating cost differentials with data obtained from several different sources or in a variety of ways inevitably heightens the inaccuracy associated with the estimates. A third recommendation pertains to the current method of predicting the extent to which the various alternatives would actually substitute for a withdrawn RPAR chemical. In some instances the use-pattern forecast for the alternatives is based on unrealistic assumptions. For instance, the
Benefit Assessment 107 chlorobenzilate analysis assumes that the base acres treated with chlorobenzilate would be evenly divided among the several alternatives, even though there are significant cost differences among those alterna- tives. A benefit assessment should estimate changes in pest control cost using a more plausible assumption. In the absence of information to the contrary, it would be reasonable to adopt, as a working hypothesis, the assumption that the relative distribution of base acres among alternatives is unaffected by withdrawal of the RPAR chemical (unless, of course, some of the alternatives were also likely to be withdrawn or otherwise restricted. As is noted in Chapter 3, in making these comparisons it is necessary to have a well- founded judgment as to which alternatives are likely to be permitted or cancelled.) That is, the increase in base acres treated with the jth alternative (AA,) following cancellation of a suspect pesticide should be presumed to equal AAj = Pj Ok, where Ak is total base acres treated with the RPAR chemical, end pj is the ratio of (1) base acres currently treated with thejth alternative to (2) base acres currently treated with all of the alternatives to the RPAR pesticide. Even this estimation procedure is highly arbitrary and, whenever possible, should be amended to reflect the best available information. ECONOMIC EVALUATION OF PRODUCTIVITY AND COST EFFECTS This section appraises current USDA/EPA procedures for evaluating the productivity and cost ejects of proposed pesticide regulations. Two related tasks are involved. The first is to assign values, generally monetary, to the real benefits (i.e., goods and services) that would be forgone were the compound to be withdrawn or its use otherwise restricted. The second task is to determine, to the extent feasible, the distribution of the benefits (which may be negative in some cases) among affected segments of the population. This section outlines the current USDA/EPA approach to these two tasks, compares it with standard methods of benefit-cost analysis, and recommends some changes in the USDA/EPA procedures. As noted at the outset of this chapter, the discussion is framed primarily in terms of agricultural uses of the suspect chemicals.
108 CURRENT APPROACH REGULATING PESTICIDES The USDA/EPA benefit analyses all adopt a partial equilibrium framework referred to as "partial budgeting." In measuring the benefits that would be forgone if a pesticide's uses were restricted or cancelled, the USDA/EPA economists usually assume that proposed restrictions would significantly affect only (1) the quantity of substitute pesticides that would be used in place of the one that is restricted, and (2) the output of the treated item. Of course, for agricultural uses and most nonagricultural uses, changes in these two variables imply changes in production costs and revenues. In a few instances, the ejects these changes would have on output prices is also taken into account (see, for instance, USDA et al. 1977~. (The quantitative assessments of public health uses of pesticides assume that tighter restrictions on use of a suspect compound will significantly influence only the quantities of alternative pesticides applied in these uses [A.L. Aspelin, oPP' EPA, Washington, D.C., personal communica- tion, October 1978~.) Typically, other variables in the quantitative analyses are assumed to remain constant, including the quantity of other inputs to the productive process, input prices, and output quality. The benefit assessments usually note that these variables may change as the result of a regulatory action, but inadequacy of available data commonly prevents the analysts from developing quantitative estimates of these changes. In a few instances, estimates of output price changes are employed in evaluating the distributional (but not the economic efficiency) consequences of a regulatory action (see, for instance, Luttner 1977c and U.S. EPA/USDA 1978a). The operational benefit measure currently used in the USDA/EPA assessments is the saving in pest control costs arising from continued use of the RPAR pesticide plus the value of the output that would be lost without the RPAR chemical. This concept can be expressed in equation form as: USDA/EPA annual benefit measure = (PCs - PCr) + (Pr Xr Pang), (5.1) where PC', is equal to the aggregate annual pest control costs with only the substitute controls; PC, is equal to the aggregate annual pest control costs with the RPAR chemical; Pr is the price per unit of output (e.g., of a crop), assumed constant (estimated, for example, by the average of the
Benefit Assessment previous 3 years); Xr is the annual output (e.g., total crop produced) if the RPAR pesticide is available; and X',, is the annual output that would be produced in the absence of the suspect chemical, assuming the price Pr remains constant. Of the 58 use-pattern assessments reviewed, 21 estimate both components of the above benefit definition. Another 19 of the assessments estimate only the expected change in pest control costs (PC`- PCr); they assume that the substitutes are as efficacious as the RPAR chemical. This assumption is often adopted because data on actual yield differences are unavailable. The remaining 18 assessments contain no quantitative evaluation of the economic impacts of a cancellation. It is often necessary to rely upon qualitative assessments because data are either nonexistent or inaccessible, or time and resources are inadequate to undertake quantitative analyses. In some instances the benefit assessments intermix estimates of the real benefits of a compound with estimates of the monetary gains and losses associated with use of the compound (see, for example, USDA et al. 1977~. This practice of combining estimates of real and pecuniary ejects will sometimes obscure the underlying definitions and methods used by the USDA/EPA analysts in measuring the separate effects. We will return to this issue in a later part of this chapter. 109 Treatment of Uncertainty The basic USDA/EPA approach to coping with uncertainty about the magnitudes of key variables in the quantitative benefit assessment is simply to omit highly uncertain variables. For instance, of the 40 quantitative use-pattern assessments reviewed (an additional 18 were nonquantitative), 19 assumed that cancellation would not reduce the yields of the crop in question. The assumption was adopted primarily because of lack of data concerning the comparative ejects on yields of different pest control measures. The USDA/EPA benefit assessments treat the uncertainties of long-run ejects in the same spirit. The assessments generally adopt a short-run perspective (3-5 years) to avoid uncertainties inherent in long-run forecasts of such key factors as technological changes in pest control or development of pest resistance to a compound (H. Gaede, oPP' EPA, Washington, D.C., personal communication, October 1978~. Some of the assessments use "sensitivity analysis" to generate alternative estimates for certain variables. Apparently this procedure is used only when there is some convict in, say, estimates of yield ejects reported in different studies. Sensitivity analysis is not routinely used to
110 REGULATING PESTICIDES estimate upper- and lower-bound values, that is, to convert the analyst's (or an "expert's") judgment about best and worst possible outcomes (e.g., yield losses) into upper- and lower-bound estimates of the overall economic impact of a regulatory action. Finally, the expected value approach to dealing with uncertainty about the magnitudes of key variables is employed in a number of instances. For instance, in the endrin/small-grains assessment, information about the historical frequency of "high" and "low" pest infestations and the accompanying yield losses in the absence of endrin are combined in the computation of an expected yield loss. In addition, estimates of pesticide prices are often based upon expected value calculations. Usually, however, too little is known about a variable's distribution to support formal expected value calculations (unless they are based on subjective judgments about the distribution). Discounting Only six of the assessments reviewed discounted future values to convert them to present-value equivalents. (The assessments used without explicit justification a 7 percent discount rate.) The practice of stating estimated erects in annual terms, combined with the short-run perspec- tive, provides the USDA/EPA rationale for not discounting future effects in the majority of assessments. In general, discounting seems to be used only when output erects following a regulatory action are partially delayed for some years, as, for example, in the case of lengthy harvest cycles. Distribution of Gains and Losses Once the various economic effects of a proposed pesticide regulation are quantified, there remains the problem of determining how those effects are distributed across population and economic groups. For the most part, the quantitative USDA/EPA assessments assume that the burden of a regulatory decision will be borne largely by the users of the RPAR pesticide. The impacts of a regulatory decision on consumers or nonusers of the RPAR pesticide are usually discussed in qualitative terms, although in a few instances, such as the DBCP citrus assessment, the erects of alternative regulations on consumers and nonusers are analyzed quanti- tatively.
Benefit Assessment RECOMMENDED APPROACH 111 In this section the USDA/EPA approach to benefit assessment (described above) is compared with standard methods of benefit-cost analysis. (For textbook treatments of the general principles of benefit-cost analysis, see, for example, Anderson and Settle 1977, Mishan 1976, and Sugden and Williams 1978; for an application involving pesticides, see Headley and Lewis 1967 and Kennedy et al. 1975.) The discussion concentrates on those aspects of the current methods used by USDA/EPA that appear inconsistent with accepted principles of benefit-cost analysis. Some of these inconsistencies may be justifiable because of data problems, time constraints, and budgetary limitations. Others seem less defensible, however, and it is these shortcomings to which the Committee's recommendations are directed. It should be understood that the following discussion of conventional benefit-cost principles will not recommend that USDA/EPA discard their current approach in favor of the full-fledged ideal methodology. Rather, the purpose of developing benefit-cost principles is to provide a means for determining which feasible changes in the current USDA/EPA approach would actually represent improvements. Economic Theory of the Benefits of Pesticide Use Within the context of conventional benefit-cost analysis, the annual benefit of continued use of an RPAR chemical can be defined as the sum of (1) the value of productive resources saved due to use of the pesticide and (2) the value of any additional output arising from use of the RPAR chemical.) This definition does not allow for the cost of transferring resources among competing uses and for the capital losses from resources that cannot be transferred effectively. When these factors are important, allowance must be made for them or the benefits will be understated. The value of productive resources saved can be expressed as: TCs- TC,, where TO is the total economic cost of producing X, the output that would be produced in the absence of the suspect chemical; and TCr is the total economic cost of producing Xr, the output forthcoming with continued usage of the RPAR pesticide.
112 REGULATING PESTICIDES The relation involves total production costs, not simply pest control costs. Moreover, it allows for the possibility that output of the treated commodity will decline if use of the RPAR compound is restricted. It cannot be determined, a priori, whether TC8 is larger or smaller than TCr, since output is allowed to vary. The RPAR chemical may well lower unit production costs; however, it may also cause total production costs to rise because of the additional production it stimulates. To facilitate the following discussion, it will be useful to draw a distinction an arbitrary one, but one implicitly made by USDA/EPA between pest control costs (PC) and "other" production costs (denoted OC). Generally, the preliminary benefit analyses include as pest control costs only expenditures on pesticides and their application (see the preceding sections of this chapter). Of course, if nonchemical controls suck as cultivation practices are feasible, it may be difficult to determine whether certain expenditures constitute pest control costs or "other" production costs. These considerations imply that TC= PC + OC; consequently, TCs- TCr = (PCS - PC,) + (OCs- °Cr) That is, the value of productive resources saved (if any) can be expressed as the change in pest control costs plus the change in other production costs. The conventional measure for the value of increased output associated with use of an RPAR pesticide is the total willingness to pay for that additional output. Total willingness to pay for extra output consists of the sum of (1) the actual payments demanders would make (PrXr- Pr X8~; and (2) the consumers' surplus associated with that incremental output [1/2(P8 - Pr )(Xr - X8 )] (where the price and quantity that reign when the RPAR pesticide is available are indicated by Pa and Xr, respectively, and those that reign in the absence of that compound are denoted by P8 and X, respectively). The formulation assumes no externalities from the consumption of the agricultural commodity and, for convenience, a linear demand curve for the treated output. Even if the demand curve is nonlinear, the consumers' surplus measure (which reflects the availability of substitutes) will generally provide a reasonable approximation to the actual change in consumers' surplus.
Benefit Assessment 113 Combining the two components of the conventional annual benefit measure,2 we have: Conventional annual benefit measure = ~ TCs - TC,) + (`P,Xr- P'Xs) + 1 /2(Ps - P,)(`X,- Xs). (5.2) One further point to note in connection with the conventional benefit measure is that it includes a term for enforcement costs that would be avoided if the suspect chemical were simply (re~registered. In the preceding discussion we have neglected these costs, as do the USDA/EPA benefit analyses. In principle, such costs would have to be taken into account for the benefit measure to be fully consistent with conventional benefit-cost measures of economic efficiency ejects. Evaluation A comparison of Equations 5.1 (the USDA/EPA annual benefit measure) and 5.2 (the conventional annual benefit measure) reveals a number of conceptual flaws in the USDA/EPA benefit measure. First, their measure omits the consumers' surplus component of Equation 5.2: 1/2P- Pr) (Xr- X:. The omission, which arises because they generally assume that output prices remain unaffected by a regulatory decision, would lead to an understatement of true benefits. To approximate this component of the correct benefit measure, one would need estimates of the price elasticities of demand and supply for the commodity in question. The problem can be illustrated graphically with the aid of Figure 5.1, which depicts the competitive demand and supply conditions for a commodity both before and after cancellation of an REAR pesticide. The pre- and post-cancellation equilibrium combinations of market price and output are (Pr, Xr) and (P8, Hi), respectively. The consumers' surplus on the increment of output, Xr - X8, equals area D in Figure 5.1. Graphically, it is this area that the USDA/EPA benefit measure omits. It should be noted, however, that the omission of this component of consumers' surplus generally produces a minor understatement of true benefits. The chlorobenzilate analysis in Chapter 7 illustrates the point. The removal of chlorobenzilate is estimated to increase the costs of producing Florida oranges for processing by roughly $7.2 million annually. The maximum-plausible consumers' surplus loss in this instance is about $200,000. Consequently, failure to incorporate the $200,000 consumers' surplus component into the measure of benefits would understate true benefits by only about 2.8 percent. Moreover,
114 I* - Z PS tar: lL CL () Pr CC REGULATING PESTICIDES \ \ \ Supply without RPARed Pesticides N/ ~ S F mu. ~RPARed Pesticide Market Demand l xs xr OUTPUT OF X FIGURE 5.1 Competitive demand and supply conditions for commodity X before and after cancellation of an RPAR pesticide (see text for discussion). Omitting this small component from the benefit measure is probably justifiable in most instances, since the data are usually too crude to support such a refinement in the estimates. Returning to the comparison of Equations 5.1 and 5.2, we note that the USDA/EPA operational benefit measure proxies the change in total production costs, TCS- TCr, with an estimate of the change in total pesticide treatment costs, PCa- PCr. The approximation to the actual change in total production may be defensible on the grounds that the change in total pesticide treatment costs can be estimated more easily
Benefit Assessment ~5 than the change in total production costs. Generally, the available data simply do not permit estimation of changes in "other" production costs. Nevertheless, it should be noted that the procedure ignores any changes in "other" production costs, OCa- (my, even though they will probably change if output changes or if failure to reregister induces changes in the methods of cultivation. The role played by "other" production costs can also be illustrated in Figure 5.1. Again, the supply curves 'SS and S'S' represent the competitive supply conditions with and without the suspect chemical, respectively. For purposes of this specific illustration, we adopt the simplifying assumption that PC`' = PCr; that is, cancellation of the RPAR pesticide is assumed to have no effect on total (and, presumably, marginal) pest control costs. The assumption allows us to focus on changes in other production costs. When the RPAR pesticide is used, total production costs are measured in Figure 5.1 by area H + I. In this example, cancellation of the chemical presumably reduces yield (unless the alternatives are equally efficacious), thereby shifting the supply curve to S'S' where total production costs are C + F + H. Since pest control costs are assumed to be constant in this example, the change in total production costs is identical to the change in "other" production costs (°Ca - °Cr ). Thus, OC8 ~ °Cr is equal to F + C - I in Figure 5~1. Depending upon market demand and supply elasticities, the change in "other" production costs can be either positive or negative. Moreover, the USDA/EPA approach to estimating the change in pest control costs following a regulatory action (PCa- PCr) contains a number of conceptual problems. First, the usual USDA/EPA practice of assuming the quantities of nonpesticide inputs to be unaffected by the regulatory decision suggests that their estimate of PCs may exceed its true value. Such an overstatement would occur, for example, if it were relatively cheaper, following cancellation of an RPAR pesticide, to simply plant more acres to offset yield losses than to employ substitute chemicals. The convenient assumption that nonpesticide input quantities remain constant rules out the realistic possibility that such inputs can serve as substitutes for pesticides. Another conceptual problem with the USDA/EPA proxy for PCa- PCr, the change in pest control costs resulting from a regulatory action, also arises from the commonplace assumption that the number of acres treated remains unaffected by the regulatory decision. In fact, if the regulatory action occasions higher per-unit production costs and, ultimately, higher output prices, quantity demanded will decline (other things being equal), thereby possibly leading to a reduction in the
116 REGULATING PESTICIDES number of acres treated with the substitute pesticides. If the acreage treated declines, the USDA/EPA estimates of the change in pest control costs will tend to be biased upward. In summary, the USDA/EPA omission of the change in "other" production costs (OC`,- OCr) from the benefit measure introduces a potential bias whose direction depends upon supply and demand elasticities. At the same time, their approach to measuring changes in pest control costs (xPCe - PCr ~ has the potential of overstating the true value for this eject and, with it, the true value of the benefits from the RPAR chemical. Another potential difficulty with the USDA/EPA operational benefit measure is that their estimates of Pr Xr- Pa X`, a major component of willingness to pay for the additional yield allowed by the suspect pesticide, also tend to be biased upward. The reason lies in the usual USDA/EPA assumption that yield changes do not affect output prices. The point is illustrated in Figure 5.2, which again depicts the competitive demand and supply conditions for a commodity both before and after cancellation of the RPAR pesticide. The initial equilibrium price and output (with the RPAR pesticide) occurs at (Pr, XT); the equilibrium following cancellation occurs at (P8, XB ). Since oPP assumes that yield losses will not affect output prices, their estimator of the actual payment that demanders will make for the incremental output (Pr XT-Pr X~) is pTxr - PrX,`~, or area F' + G + H' + I in Figure 5.2. Clearly, this estimator will always overstate the true value of P. Xr - Pr X8, which is G + I in Figure 5.2. All of the various estimation problems can be summarized with reference to Equation 5.2 (rewritten as Equation 5.3 for convenience), Conventional annual benefit measure = (OC8 - °Cr) + (PCS-PCr) + (Pr Xr Pr X8) + 1 /2(Ps-Pr)(Xr-X). (5.3) The benefit methodology employed by oPP generally omits consideration of the change in "other" production costs (OCa- OCr) and of the consumers' surplus gain [1/2(P8 - Pr)(Xr - X8~. Omission of the consumers' surplus gain introduces a downward bias into the benefit estimate, although the magnitude of this bias is generally not very large. Omission of the change in "other" production costs has an uncertain eject on the accuracy of the benefit estimation. With regard to the change in pest treatment costs (PCS - PCT ~ and the amount demanders
Benefit Assessment it, UJ IIJ P c) r cr 117 s / \\\ Z p ~ s ' / . S' l s rs \ D X's xs xr OUTPUT OF X FIGURE 5.2 A bias in the USDA/BPA benefit measure (see text for discussion). actually pay for the incremental output (Pr Xr - Pr X8 ), the USDA/EPA approach tends to overstate both of these benefit components. On balance, the combination of biases working in different directions and the uncertainty about the bias associated with (OC,'- °Cr ~ imply that the EPA/USDA benefit estimator has the potential for either understating or overstating the true benefits of the RPAR pesticide. Unfortunately, the quality of the data is generally so poor that there is little the USDA/EPA analysts can do at present about these potential biases. The kind of data currently available do not justify any more refined or elaborate methods for assessing benefits. Nevertheless, these sources of potential biases should be noted clearly in each benefit
118 REGULATING PESTICIDES assessment. Needless to say, if the quality of the data improves over time, the potential sources of inaccuracies should be reduced or eliminated by altering the partial budgeting methdology so that it conforms more closely to the conventional methodology of benefit-cost analysis. Treatment of Uncertainty The only important change recommended in the USDA/EPA approach to dealing with uncertainty is to recognize openly the uncertainties surrounding the estimates presented and to indicate ranges of values, surrounding the most-probable estimate, within which the analysts feel some confidence that the true values lie. An ideal to strive for is the 90 percent confidence interval, even though such statistical precision will rarely be attainable. Discounting The USDA/EPA position on discounting is inconsistent in that some of the assessments (e.g., chlorobenzilate) discount while others avoid it. The function of discounting is to give costs or benefits to be received in the future a value comparable with costs incurred or benefits received in the present. The discounting procedure recognizes that people prefer returns sooner rather than later, that they prefer to postpone costs, and that all resources have alternative social uses or opportunity costs. In the case of a chemical pesticide registration that is presumed against, the present value of the benefits or costs of cancelling that registration could diner significantly when discounted at a rate of interest of, for example, 8 percent from the present value rather than at an interest rate of zero. This would be especially true if the expected remaining life of the chemical in question were, for example, more than 4 or 5 years. The present value of a dollar to be received 5 years hence is $0.68 when discounted using an 8 percent interest rate. At the same rate of interest, a dollar to be received 9 years hence has a present value of only $0.50. If a regulated chemical has an expected remaining life of 5-10 years, discounting the estimated stream of benefits from its continued use would provide a more accurate estimate of the social costs of restricting its use. However, before recommending that oPP adopt discounting for all benefit analyses, two important parameters need to be determined: (1) the length of time over which to apply discounting, and (2) the appropriate rate of interest. The remaining useful or economic life of a chemical pesticide that is in
Benefit Assessment 119 use and subjected to RPAR action is likely to be in the range of 5-10 years (see Chapter 4, the section on Exposure Analysis). The economic termination of the use of a pesticide may be due to a loss in absolute or relative efficacy or both. Therefore oPP, after reviewing the literature on a compound, can determine a defensible useful lifespan. In Chapter 4, the Committee made a rough estimate that the economic lives of pesticides has averaged about 10 years. Selection of an interest rate for discounting is the second important decision. Ideally, the rate chosen should reflect the real value that society places on future financial benefits and costs resulting from pesticide use. In past analyses (e.g., chlorobenzilate), oPP has used 7 percent as the interest rate for discounting. The Committee believes that this is a reasonable estimate of the real rate of return on investment in the economy after adjusting for inflation. If the streams of benefits and costs of pesticide regulations are discounted in the fashion described above, the result should be a present value of net benefits for the remaining expected useful life of regulated compounds subjected to RPAR. This measure would help to differentiate between the value of compounds with similar annual benefits, but with different expected useful lives. Distribution of Gains and Losses The USDA/EPA benefit assessments are directed more toward measuring the distributive erects of a regulatory action (especially as the elects relate to the pesticide users) than toward quantifying the economic efficiency erects (A.L. Aspelin, oPP, EPA, Washington, D.C., personal communication, November 1978~. Thus, it is important to compare the USDA/EPA approach to estimating the distributive erects with the methodology of conventional partial-equilibrium incidence analysis. The conventional approach is illustrated in Figure 5.3, which repre- sents the same short-run competitive supply and demand conditions as the preceding figures in this chapter. With the RPAR chemical available, total revenue from sale of the treated crop equals Pr Xr, or area E + F + G + H + I in Figure 5.3. Initial short-run production costs (including a normal profit to growers) are measured by area H + I. Thus the initial economic profit or producers' surplus accruing to growers equals E + F + G. The total willingness of consumers to pay for output Xr is the entire area under the demand curve up to Xr, namely, A + B + . . . + H + I. Since consumers actually pay E + F + G + H + I for quantity Xr,
120 - P s LL p UJ r Cal \ A \/ E / S' F ~ / S l REGULATING PESTICIDES S' 1 < ~ S \D xs xr OUTPUT OF X FIGURE 5.3 Distnbutional consequences of cancelling the registration of an RPAR pesticide (see text for discussion). their initial net benefit or consumers' surplus from having Or available at price Pr is equal to A + B + C + D. Suppose that a regulatory action against the RPAR pesticide raises production costs and results eventually in a new equilibrium price and quantity of Pa and I, respectively. Total revenue is now B + C + E + F + H; total cost is C + F + H; producers' surplus is B + E; and consumers' surplus is area A. The regulatory action has made consumers of X worse on by the amount B + C + D. Growers, however, may have been made either
Benefit Assessment ·21 better on or worse off, since the change in producers' surplus equals B - F- G. Whether this change represents a gain or a loss depends upon the elasticities of the market supply and demand curves. We might also note that total revenue and total cost might either increase or decrease, depending on the shapes of the demand and cost curves: total industry revenue changes by B + C- G - I, whereas total industry cost changes byF+ C-I. Suppose that only a fraction of the growers of crop X use the RPAR pesticide. Under these circumstances, conventional incidence analysis suggests that if the regulatory action against the suspect chemical results in cost increases or output losses substantial enough to occasion price increases, then (1) consumers will clearly suffer, (2) nonusing growers will clearly benefit, and (3) users may super either gains or losses, depending upon demand and supply elasticities. If the users of the RPAR pesticide are few relative to the total market, the regulatory action is unlikely to affect output prices. In this case, only the users suffer losses, which would be measured by their forgone economic profits (rather than revenue reductions). In general, the smaller the number of users relative to nonusers, the greater the likelihood that the entire burden of the regulatory action will be borne by the users. A technical qualification to the preceding discussion is in order. The conventional approach assumes that reductions in the demand for productive inputs result in a virtually instantaneous transfer of those inputs into other equally productive activities. In reality, there will generally be some transitional unemployment as resources move from one productive activity to another. Incorporation of this effect into the conventional analysis would result in an increase in the costs of a regulatory action. Unfortunately the data available to the USDA/EPA analysts are generally inadequate for considering these transitional unemployment erects (H. Gaede, oPP, EPA, Washington, D.C., personal communication, October 1978~. There is an important flaw in the partial-equilibrium approach to evaluating the distributive erects of pesticide regulation. While it offers a more reasonable approach to assessing distributional implications than the traditional partial-budgeting method, it provides only a limited, partial view of true distributional impacts. A complete distributional analysis would incorporate not only the direct erects of the regulation, but also all of the major indirect or spillover erects on related markets. If a detailed quantitative assessment of the major direct and indirect distributive erects of a pesticide restriction is desired, it would be necessary to conduct the distributive analysis in the context of multi- sector (or multi-crop) programming or econometric models. Of course, if
122 REGULATING PESTICIDES the proposed regulation is expected to have only marginal yield or cost erects, the resultant spillovers will generally be negligible, implying that the partial-equilibrium estimates would be reliable. The USDA/EPA benefit analysts currently have access to at least two complex mathematical models that can, in principle, reveal how regulatory decisions would impinge on the markets for certain major agricultural crops (namely, cotton, corn, barley, soybeans, oats, grain sorghum, and wheat). One of these models is a multi-crop linear programming model developed by EPA in the early 1970s to analyze some of the distributive erects of prohibiting the use of chlorinated hydrocar- bon pesticides on corn (see Epp et al. 1977 for a more detailed description, especially pages 30~2~. The second is a multi-crop econometric model recently developed for oPP by Lacewell and Taylor (described in Taylor et al. 1979~. However, neither of these models is presently used by the USDA/EPA analysts in their benefit assessment work (A.L. Aspelin, oPP, EPA, Washington, D.C., personal communication, November 1978~. Reluctance to rely on the mathematical modelling approach to evaluate distributive effects of a pesticide regulation is justifiable. Frequently, the spillover erects of a proposed regulation are negligible and thus too small for mathematical models to detect with any degree of confidence (G. O'Mara, oPP, EPA, Washington, D.C., personal communi- cation, June 19781. In addition, the generally poor quality of the cost and, especially, the yield data raises serious doubt as to the appropri- ateness of basing the benefit assessment on the results of sophisticated programming or econometric models. The detailed, seemingly precise results provided by the models may lend an unwarranted aura of credibility to the economic impact estimates. Consequently, if the mathematical modelling approach does become more important in the assessment of pesticide benefits, the Committee recommends that the analysts routinely report the results of sensitivity analyses in order to reveal the actual uncertainty surrounding the impact estimates. In the qualitative discussions of the economic impacts of a regulatory action, the USDA/EPA analysts recognize the implications of conventional economic incidence analysis. They note, for example, that consumer prices are likely to rise following a regulatory response. However, the quantitative assessments usually assume that the burden of a regulatory decision will be borne entirely by the growers-or, more specifically, the users of the IlPAR pesticide. Thus, the USDA/EPA quantitative benefit assessments tend to overstate the losses suffered by users of a cancelled (or otherwise regulated) pesticide. Similarly, they tend to understate the losses suffered by consumers and the gains enjoyed by nonusers. These
Benefit Assessment 123 inaccurate estimates arise mainly out of a failure to account explicitly for the output price changes likely to result from a regulatory action. In connection with the estimation of the distributive ejects, the USDA/EPA approach suffers from at least one other problem. Some of the distributive ejects involve income redistribution. Crop price increases, for instance, will in large part redistribute purchasing power from consumers to growers. Thus, if increases in crop prices benefit growers by, say, $50 million annually, those same price increases cost consumers $50 million annually. (For simplicity, this statement ignores the likelihood that the higher prices will lead to greater production.) Unfortunately, several of the benefit assessments have failed to spell out clearly the fact that gains to growers may entail losses for consumers. For instance, in the endrin/apple assessment (endrin is used in apple orchards to control voles) the following conclusion was reached regarding the likely impacts on apple growers: The greatest impacts will occur if current endnn users substitute zinc phosphide in their control programs. Under this program, current endnn users would incur losses in net returns equal to $19,110,000 after three years. Nonusers of endrin would experience increased net returns equal to $51,323,000 after three years due to higher apple prices caused by losses in the endr~n use areas. Over the initial three-year period following the cancellation of endr~n, users would experience a drop in net returns from $675 to $429 per acre, while nonusers would experience an average increase in net returns from $675 to $788 per acre. Under a cPN-DPN-herbicides-cultural methods program, the aggregate impacts upon users and non-users would be approximately one-half the magnitude Ace; "Stan ~ ~ ~ ~ or ~ On APE dreary m _ _ _ rlVJ-~ ~ t~4 ~ ~-c. ~ I. ..... [C]u~ent ends users would experience a loss in net returns of $9,479,000 over the initial three-year period. Non-users of endrm would receive an aggregate increase In net revenues of $25,773,000 over the same period (Luttner 1977c: 76, 80~. In contrast, the analysis of the likely effect on consumers is essentially limited to the following: Although the cancellation of endr~n has the potential to cause economic hardship for growers in the affected areas, the aggregate impact does not appear to be significant on a macroeconomic level (Luttner 1977c:84~. The reader should note that according to these three extracts, the net eject on the incomes of apple growers of suspending the use of endrin would be an increase of $32 million a year if zinc phosphide were substituted, or an increase of $16 million if cPN-DPN-herbicides were used. These figures give the misleading impression that there are no net benefits of using endrin in apple orchards, by including the gain to
124 a Cal so 0 O :O U. ~.O ~ O 4~ cat 0 C ~ aUo' ~ U' ~ O ~ =0 ~ CL o 3 ~ no ~ o To ~So <; ·U, U' ~7 ._ . - Cat - o o .= - I oo ~ C~ o ~ _1 ~( ~.. . 00 _ No ret ~ ~0- o ~o o o ~ Ioo ~ ~o o ret ~ ~O Go ret ~Do I =, ~ o oo oo ~C ~ ~ oo O O oo r ~- , r~ _ I oo ~ ~O' o ~ _~- ~ ~1 ~- 1 t- ~r ~ ~I _ _ _ _ V~ ~ _ ~t_ 00 r ~ ~O r~ 0 t~ ~_ _ _ 8 8 OO ~v __ V' _` ~ a, ~ ~ .= .~ ~ ~C,) . _ l~ O .~ O = - .E ~ ° _ ~ .e ~ o 3 ~ ° C~ C,) o - D DD _ __ O OO _ _. _ _ __ ._ ._._ {:: ~ . _ _ ~ ~2 2 o ~ C ~ C ;~ r ~ - ~ O ~ ~ a., ~ ~ a, ~ 3 c~ ~ c: ~ ~ 0 0 000 ._ ._ ._._._ _ _ ___ _ _ ___ .g .g .g .g .g 6~3 6e 69 6. 6~ .c a, ° 0 ~ u~ c ~° c°~ 0 0 _ _ ·_ ~ 0 0 ~ .= "o .Q,) ~ ~ ~ ~ ~ 9
125 Io oo ~ ~ . . . .. oo ~ ~oo Y_ r ~ ~00 o ~°° ~1 ~ oo ~ oo1 _ ~r~ ~ 1 1 11~o 1 1 1 1 =- ~4 . . c: ~ o o o o o . . .g .p .d .§ .§ 6 ~69 e9 eg 6~ Ct ·~ cO D C~ O ~ ~ ~L o ^^ t- t- a~ cr~ O 4- V~ 8 o ~ o ~ _ ~ P~ ~ D C~ :3 .c :> D - S: o ._ - .f r~ V' .f D S: o - .Z o 6~ o o o 69 ~ 0~ _ P E E e e c ~ 0 Y ~ o> , = ~ a ~ y ~ y ~ U e S ~ 8 1! C ~ ~ tj iY: ~ ~- ~ ~ §~3~i Z ~ ~ ~, ~- ~ t0 ~ ~ ~ ._ ~L o 8 . - u) c~ 5 __ . g o 1C'1 ~ ao ._ g C .= ~ ~: ~ V) o ~ a, ~ .. c~ n~ .° ° o ~ Z C~ C~ V,
126 REGULATING PESTICIDES TABLE 5.2 An Example ofthe Recommended Format for the Presentation of Benefit Assessment Results Cost of Denying Reregistration to Trifluralin for Use on Cotton (assuming demand elasticity is 0.3, horizontal supply curve) Real Costs cuff Dental Decrease in lint production 574 million lb at (0.60 + 0.715)/2 Decrease in seed production 959 million lb at o.oSa Value of substitute cropa Added cost of weed control Reduced variable cost on acres shifted to substitute Total real costs Monetary G'sts cuff Dental To trifluralin users: Lint production (million lb) Price Gross sales ($ million) Decrease in lint sales Decrease in seed production a Value of substitutea Added cost of weed control Reduced variable cost on acres shifted Net monetary cost to users To other cotton producers: Increase in gross sales 1637 million lb at $0.115 To consumers (foreign and domestic): Increased cost of output purchased 4829 million Ibat $0.115 With Trifluralin 3766 $0.60 2260 Loss in value from output restriction 574 million lb at (0.60 +0.715)/2 - 0.60 = 0.0575 Total monetary costs ($ Million)b $377 48 - 39 6 16 $376 Without Trifluralin 3192 $0.715 2282 -$ 22 48 39 - 16 23 -$188 555 33 $588 $377 a Assumes change in output does not induce change in price. b Negative entries indicate negative losses or gains. Source: USDA et al. (1977, Table 1) .
Benefit Assessment ·27 growers from the higher apple prices that would exist if it were not used, while ignoring the offsetting loss to consumers. As a general rule, the USDA/EPA benefit assessments need to pay more attention to identifying implications of proposed regulations for the distribution of income. In many instances, the data are inadequate for measuring distributional ejects with much precision. However, reason- able upper- and lower-bound estimates can usually be developed, for instance, by adopting assumptions about the plausible upper- and lower- bound values for demand and supply elasticities (see the chlorobenzilate analysis in Chapter 7 for an illustration). A final recommendation concerns the presentation of results of benefit assessments. At present, no clear distinction is made in the benefit assessment documents between the real and pecuniary (or distributional) effects of not reregistering (or otherwise regulating) a pesticide. Not to distinguish carefully between them is confusing to trained readers and misleading to untrained readers. An example of the type of format that is commonly used in reporting results of benefit assessments is provided in Table 5.1. According to the trifluralin assessment document (USDA et al. 1977), this table is supposed to present "the estimated short-run economic impact of a trifluralin suspension on cotton.... " Although the figures may be correct, the impression given by the table is entirely misleading.3 According to the table, if trifluralin were banned, the net incomes of cotton farmers who use trifluralin would increase by $24 million a year (see the last column) and those of other cotton farmers would increase by $188 million for a total gain of $212 million in net farm income. As far as can be determined from the table, the American economy would be better on without trifluralin even apart from its effects on public health and the ecology. Of course, the facts are otherwise. If trifluralin is forbidden, more resources will be used to raise cotton and less cotton will be produced. As a result, purchasers of cotton will pay higher prices and receive less cotton. A proper, "double-entry" accounting for the eject of banning trifluralin is shown in Table 5.2. The real resource cost entailed is constructed in the upper panel; it amounts to $376 million a year. The incidence of those costs is displayed in the lower panel. The monetary gains to cotton raisers are outweighed by the loss of $588 million sustained by cotton users. The net monetary loss ($588 - $212 = $376 million) is equal to the real resource cost. In this world every component of real cost is paid for by somebody. The entries in the consumers category require a brief explanation. In connection with the first entry, the 4,829 million pounds of lint available
128 REGULATING PESTICIDES without tr~fluralin would cost consumers $555 million more than that same quantity would cost at the price that would reign if trifluralin were available. In connection with the second entry, the $33 million measures the difference beween (1) the value consumers place on the 574 million pounds of lint that would be lost if tnfluralin were suspended and (2) the value they place on the alternative commodities they can buy with the purchasing power released by the reduction in the quantity of lint demanded. The difference is nothing more than the consumers' surplus triangle discussed earlier in this chapter. The display of separate real and monetary ejects makes several things clear. First, the monetary effects are a distribution of the real ejects. Apart from a rounding error in Table 5.2, the total of the monetary ejects equals the total of the real ejects. Second, in the present Instance, the consumers sustain a loss that is more than 50 percent greater than the real loss. The reason is the assumed inelasticity of demand, which makes it possible for producers to more than shift their increased costs to consumers. The key figures, which are not found in the USDA/EPA presentation in Table 5.1, are that there would be a real loss measured by $376 million, which would result in a purchasing power loss of $588 million on the part of cotton users. (Note that in keeping with a preceding recommendation, these estimates should be reported as ranges, rather than as precise numbers.) The Committee recommends that tables similar to Table 5.2 be routinely included in the benefit assessment reports. NOTES 1. In economists' terms, this expression is the sum of the changes in producers' and consumers' surpluses. The measure provides an estimate of the unobservable sum of the compensating variations-the correct theoretical measure of changes in economic welfare. For a discussion of some of the technical problems associated with estimating welfare gains and losses with the producers' and consumers' surplus measures, see Chipman and Moore (1976, 1979), Mishan (1976), Mohring (1971), Silberberg (1972), and Willig (1976). As some of these authors note, there is considerable theoretical controversy as to whether consumer surplus has any meaning beyond the level of individual consumers. 2. In some instances, a restriction on a pesticide used in the production of commodity X (e.g., wheat) may indirectly affect the prices of other commodities (e.g., barley and oats) through demand or supply interdependencies. The benefit measure set forth as Equation 5.2 abstracts from the gains and losses that would be associated with such price changes. When the indirect price changes (and the related gains and losses) are important, allowance has to be made for them. Otherwise, the benefits will be incorrectly measured. However, in the Committee's opinion these indirect price effects will usually be small enough to neglect safely. The reasoning behind this belief is as follows. Most of these indirect gains and losses will be transfers of purchasing power between consumers and producers. While such
Benefit Assessment 129 transfers may be relevant to an evaluation of the distributional consequences of a regulatory action, they are not relevant to an eval~l~tion of the economic efficiency consequences of that action. In measuring the efficiency effects, it is necessary to account only for the net (or "deadweight") gains or losses in the other, related markets. The net gains or losses in these related markets appear likely to be small relative to the efficiency gains or losses in the primary market. In any event, the data will generally not permit quantification of either the gross or the net gains or losses associated with these market interdependencies. 3. The trifluralin estimates are used in this table merely for illustrative purposes; the Committee has not attempted to evaluate their accuracy. REFERENCES Anderson, L.G. and R.F. Settle (1977) Benefit-Cost Analysis: A Practical Guide. Lexington, Mass.: D.C. Heath. Chipman, J.S. and J.C. Moore (1976) The scope of consumers surplus arguments. Pages 69-123, Evolution, Welfare, and Time in Economics, edited by Tang, Westfield, and Worley. Lexington, Mass.: D.C. Heath. Chipman, J.S. and J.C. Moore (1979) Compensating Variation, Consumer's Surplus, and Welfare. Mimeograph. Accepted for publication by the American Economic Review. Epp, D.J., F.R. Tellefsen, G.A. Shute, R.M. Bear, and K.P. Wilkinson (1977) Identification and Specification of Inputs for Benefit-Cost Modeling of Pesticide Use. A Report to the Office of Research and Development, U.S. Environmental Protection Agency, Washing- ton, D.C., EPA-600/5-77-012. Springfield, Va.: National Technical Information Service. Headley, J.C. and J.N. Lewis (1967) The Pesticide Problem: An Economic Approach to Public Policy. Washington, D.C.: Resources for the Future. (Distributed by The Johns Hopkins Press, Baltimore, Md.) Kennedy, R., R. Lowrey, A. Bernstein, and F. Rueter (1975) A Benefit Cost System for Chemical Pesticides. A Report to the Office of Pesticide Programs, U.S. Environmental Protection Agency, Washington, D.C., EPA-540/9-76-001. Springfield, Va.: National Technical Information Service. Luttner, M.A. (1977a) Preliminary Benefit Analysis of Chlorobenzilate. Criteria and Evaluation Division, Office of Pesticide Programs, U.S. Environmental Protection Agency, Washington, D.C. (Unpublished) Luttner, M.A. (1977b) Supplement to the Preliminary Benefit Analysis of Chlorobenzilate. Criteria and Evaluation Division, Office of Pesticide Programs, U.S. Environmental Protection Agency, Washington, D.C. (Unpublished) Luttner, M.A. (1977c) Preliminary Benefit Analysis of Endrin Use on Apple Orchards. Criteria and Evaluation Division, Office of Pesticide Programs, U.S. Environmental Protection Agency, Washington, D.C. (Unpublished) Mattson, C.D., F.T. Arnold, G.L. Ballard, R.A. Freund, R.C. Holtorf, M.A. Luttner, and D.E. Tegelman (1977) Preliminary Benefit Analysis of Endrin. Criteria and Evaluation Division, Office of Pesticide Programs, U.S. Environmental Protection Agency, Washington, D.C. (Unpublished) Mishan, E.J. (1976) Cost-Benefit Analysis. New York: Praeger Publishers. Mohring, H. (1971) Alternative welfare gain and loss measures. Western Economic Journal 9:349-368.
130 REGULATING PESTICIDES National Research Council (1977) Decision Making in the Environmental Protection Agency. Volume II, Analytical Studies for the U.S. Environmental Protection Agency. Report of the Committee on Environmental Decision Making, Commission on Natural Resources. Washington, D.C.: National Academy of Sciences. Pimentel, D., J. Krummel, D. Callahan, J. Hough, A. Merrill, I. Schreiner, P. Vittum, F. Koziol, E. Back, D. Yen, and S. Fiance (1978) Benefits and costs of pesticide use in U.S. food production. BioScience 28:772, 778-784. Silberberg, E. (1972) Duality and the many consumer's surpluses. American Economic Review 62:942-952. Sugden, R. and A. Williams (1978) The Principles of Practical Cost-Benefit Analysis. Oxford: Oxford University Press. Taylor, C.R., R.D. Lacewell, and H. Talpaz (1979) Use of extraneous information with an econometric model to evaluate impacts of pesticide withdrawals. Western Journal of Agricultural Economics 4(1): 1-7. U.S. Department of Agriculture, State Land Grant Universities, and U.S. Environmental Protection Agency (1977) Short-Run Economic Analysis of Trifluralin and Trichloro- benzoic Acid. (Note: This unpublished Final Report is available from oPP' U.S. EPA, Washington, D.C.) U.S. Environmental Protection Agency (1979a) Chlorobenzilate. Notice of Intent to Cancel Registrations and Deny Applications for Registration of Pesticide Products Containing Chlorobenzilate. 44 Federal Register 9547-9567. U.S. Environmental Protection Agency (1979b) Pesticide Industry Sales and Usage 1979 Market Estimates. Office of Pesticide Programs, EPA, Washington, D.C. 2~. (Unpublished) U.S. Environmental Protection Agency and U.S. Department of Agriculture (1978a) Economic and Social Impacts of Cancelling Use of DBCP as a Pesticide for all Registered Use Sites with Known Current Usage. (Note: This unpublished Final Report is available from oPP' U.S. EPA, Washington, D.C.) U.S. Environmental Protection Agency and U.S. Department of Agriculture (1978b) Impact of the Loss of Toxaphene on Selected Agricultural Uses: A Partial Budgeting Analysis. (Note: This unpublished Final Report is available from oPP, U.S. EPA, Washington, D.C.) U.S. Environmental Protection Agency and U.S. Department of Agriculture (1978c) Preliminary Benefit Analysis of Lindane. (Note: This unpublished Final Report was prepared under the general direction of Fred Hageman, Project Manager for Lindane, and is available from oPP, U.S. EPA, Washington, D.C.) Willig, R. (1976) Consumer's surplus without apology. American Economic Review 66:589-597.
