from two anonymous referees, David Zilberman, Doug Parker,. Jerry Siebert, Adolfo Gallo, Steve Shaeffer, and participants in seminars at UC Riverside and UC ...
the
Measuring Nonuniform
Marginal
Cost
of
EnvironmentalRegulations
David L. Sunding A method is presented for measuring the marginal welfare cost of environmental regulations affecting agriculture. The method incorporates output market effects and recognizes diversity in production conditions among crops, regions, and seasons. An important advantage of the method is that only regional outputs and changes in regional production costs are needed to calculate deadweight loss, thus simplifying the measurement of welfare changes. This feature of the model is significant since the complexity and substantial data requirements of most existing impact models cause many environmental regulations to be enacted with inadequate analysis of their economic impacts. The method also disaggregates welfare impacts by crop, place, and time, thus encouraging the implementation of nonuniform interventions that achieve a given level of environmental quality more efficiently than uniform policies. Key words: environmental regulations, welfare analysis, microparameter models, distributional impacts.
Regulations intended to improve environmental quality often entail changes in agricultural production processes. Policies to improve water quality, ensure worker safety, maintain soil quality,enhance wetlands, and protectendangered species frequentlyrequireadjustmentsin farming practices. In this paper I develop a method for calculating the marginal welfare costs of environmental regulations affecting agriculture. To the extent that environmental regulations imply changes in farming practices, there is a natural tradeoff between environmental quality and production costs. Requiring growers to reduce pesticide applications, for example, typically reduces yields and may increase per acre production costs as growers adopt other pest control methods such as integrated pest management. The result of reducing pesticide application, then, is to increase the marginal cost of production. Agriculture differs from many other industries in that farming is highly dependent on the physical and biological environment. Because David L. Sunding is senior economist with the Council of Economic Advisers, Executive Office of the President. This research was supported by a grant from the Office of Pesticide Consultation and Analysis of the California Department of Food and Agriculture. The author acknowledges helpful comments from two anonymous referees, David Zilberman, Doug Parker, Jerry Siebert, Adolfo Gallo, Steve Shaeffer, and participants in seminars at UC Riverside and UC Santa Barbara.
environmental conditions vary widely among regions, agricultural production processes also differ among regions. Factors such as soil quality, drainage conditions, water availability, and pest control problems partially determine the methods and costs of production. Thus, the effects of environmental regulations on agriculture will most likely vary among regions, and, as emphasized by Zilberman et al. and Lichtenberg, Parker, and Zilberman, environmental regulations frequently have significant distributional impacts. Agriculture is also highly dependent on dynamic conditions such as temperature, humidity, rainfall, and pest population growth rates. When calculating the cost of environmental regulations of agriculture, it is also importantto temporally disaggregate the impact of the regulations as opposed to performing an annual analysis. There are several examples of timedependent environmental regulation in agriculture, which are discussed below, and many more cases where varying regulations by season would reduce the welfare costs of improving environmental quality. Disaggregating the marginal costs of environmental regulations encourages policy makers to design and implement more flexible, nonuniform policies that tailor regulations to individual crops, regions, and seasons. This point is emphasized in the Lichtenberg, Spear, and
Amer. J. Agr. Econ. 78 (November 1996): 1098-1107 Copyright 1996 American Agricultural Economics Association
Sunding
Marginal Cost of Environmental Regulations
1099
Zilberman analysis of re-entry intervals follow- Act, growers of a particular crop in a particular ing pesticide applications that calculates first- region can request a Section 18 exemption from best, region- and crop-specific regulations. The registration requirements in case of extreme Lichtenberg, Zilberman, and Bogen study of need, usually defined as large profit losses redrinking water contamination also suggests that sulting from the absence of alternative controls. different water quality standards should be de- Chemical bans and the taking of arable land for veloped for urban and rural areas as a result of critical habitat protection for endangered spethe significant economies of scale in urban wa- cies are also done on a regional basis. ter treatment. Physical information is being used to develop A method is presented for measuring the mar- localized environmental policies affecting agriginal costs of nonuniform environmental regu- culture. For example, the state of California is lations that recognizes differences in produc- currently banning the use of pesticides likely to tion conditions among crops, regions, and sea- leach into groundwater in certain Pesticide sons. There have been several attempts to de- Management Zones (PMZ). Growers operating velop methods for measuring marginal costs of within these areas are denied access to these environmental regulations affecting agriculture, chemicals through the registration process, most notably Lichtenberg, Parker, and wherein growers must file for permission to use Zilberman. This paper extends the existing lit- certain agricultural chemicals at the time of erature by explicitly considering temporal as purchase. The state has developed a Geographic well as spatial diversity, thus facilitating the de- Information System (GIS) that enables the persign of environmental regulations that are sea- mit issuer to tell whether the grower's field is son- and region-specific. This modification is within a PMZ and act accordingly.A similar proespecially important in markets for perishable gram is being developed for the Corn Belt by the U.S. Environmental Protection Agency to commodities that have widely fluctuating control nutrient contamination of groundwater. over time. shares and market prices, quantities, The formal analysis in the next section reFinally, there are a small number of current sults in an equation characterizing marginal environmental regulations that are seasonal. welfare impacts for each crop and season com- For example, some re-entry and preharvest inbination as a weighted average of the changes tervals after pesticide applications vary by seain regional marginal production costs. This son as foliar residue decay rates depend on theoretical result is appealing on practical temperature, humidity, and rainfall. Making more environmental regulations season depengrounds because the method requires only the dent can have significant welfare benefits. In to assess information obtainable readily of environmental costs fact, in the empirical example of a pesticide regulations. marginal Further, the welfare impacts can be calculated cancellation presented below, seasonal differwith a spreadsheet, thus making the method ences in the marginal welfare impacts of canlow-cost and accessible to noneconomist policy cellation within a region are as large as the makers. Econometric measures of demand and variation between regions in a given season. The formal analysis, presented in the next supply elasticities, which are difficult to estimate and interpret, are only needed to partition section, culminates in an expression for the the total welfare losses into consumer and pro- marginal welfare costs of environmental reguladucer surplus changes; the aggregate welfare tion. The method is then used to calculate the loss does not depend on these elasticities. The marginal costs of banning the pesticide, welfare loss expression developed in the next mevinphos. State and federal agencies are insection also is shown to be a close approxima- vestigating whether this organophosphate intion to true welfare loss in an important class of secticide poses unreasonable risks to farm workers and consumers, particularly infants production models. The practical value of the model developed in (State of California), and the state of California this paper depends on the ability of regulatory has recently taken steps to ban its use in the agencies to enforce crop, region, and time-spe- state entirely. Measures of the marginal welfare cific environmental regulations. Nonuniform costs of banning mevinphos in California, regulations are feasible, as the following para- based on information from four vegetable crops graphs demonstrate. Nationally, under the Fed- grown in various parts of the state and elseeral Insecticide, Fungicide and Rodenticide where, are presented.
1100
November 1996
Amer. J. Agr. Econ.
Marginal Welfare Costs of Environmental Regulation
P
(5) iel
Economic welfare is defined here as the unweighted sum of producer and consumer surplus; the impacts described here are gross welfare changes from regulation since the analysis does not quantify the benefits of regulation such as increased levels of human and environmental health. Denote the level of production of some crop in region i in period t by qit and the market price in period t by the inverse demand function p,(q,), where q, is total production of I regions at time t. The cost of production in region i in period t is denoted by the continuous and differentiable function ci,(qi,, git), where git indexes environmental regulation in region i and period t. Suppose that the production technology is characterized by , it)
acit (qit,
aqit
where Eitis the elasticity of supply in region i in time t, and rt,is the elasticity of demand in time t. The system of I + 1 equations can be solved to obtain marginal changes in regional production and market price, dq,, and dp,. Note that equations (4) and (5) can accommodate any type of shift in regional supply curves and that we have made no assumption about the functional form of regional marginal cost curves other than eliminating increasing returns to size and assuming continuity and differentiability. Expressions for the change in producer and consumer surplus follow naturally from the calculated marginal changes in quantities and prices. Consumer surplus in period t is given as a,
a2Cit(qit, git )
>0; >0. _
dqit - dpt = 0 Vt
tqit
(6)
aqit2
CS, = Jq(p,)dp, Pt
Thus, technology in each region exhibits constant or decreasing returns to size. Consider a perfectly competitive output market. The period t profit of region i is given as (1)
where p, is a variable of integration and a, is the vertical intercept of the demand curve. The marginal change in consumer welfare is then dCS, = -qdp,.
Producer surplus in region i and period t is given as
it = P,qit - cit(qit, Jit).
The first-order condition for profit maximization is
qa7
(7)
PS,, = p,q, - JMCi,(Oi,,
i)di,
0
(2)
acit (qi, [tit )
-
, aqi a,qit
t = O Vi,t
where 0, is a variable of integration. The marginal change is found using Leibniz's Rule as
which simply requires that price equal regional marginal cost in each time period and region. Market equilibrium is characterized by
q'aMCi(Oit.git) aOit
o
(3)
p,(qt) - pt = 0 V t
-
where qt is total production in period t. Totally differentiating equations (2) and (3), and using the fact that competition implies marginal cost is equal to price, we have the system 4 (4)
P - dP ,itqit
dq,
+
aMCi,,(q,,, d)di t _ 0Vit
it)
t
- dpt = 0 Vi, t
i
MCi, (i,
[it)dqi,
which simplifies to
(8)
aMC,, (qit, dPSit = qi,dp, -
it )dqit
agit
In regions that are not directly affected by environmental regulation, this expression reduces to q,dp,. These growers should gain from the
Marginal Cost of Environmental Regulations
Sunding
regulation to the extent that it raises marginal costs in competing regions and increases output price as a result. Expression (8) can also be interpreted in terms of changes in quasi-rent: the first term is simply the change in farm revenue resulting from the regulation, while the second term is the change in production costs. The gross marginal welfare cost of a change in environmental regulations is calculated by summing the expressions for the change in consumer and regional producer surplus to obtain (9) (9)
(qit, tit)dqit q q i,.
_jI aMCit ddW, = = -,"-, iEl
agit
The marginal change in social welfare in each time period is thus equal to a weighted average of changes in regional marginal costs of production, with weights given by preregulation output levels. Expression (9) is highly intuitive: the marginal cost of environmental regulation is equal to the increase in the value of inputs to agricultural production. This expression also shows that marginal welfare loss is separable in the change in regional marginal production costs and separable over time, a feature that greatly simplifies the computation of impacts. Expression (9) is significant for policy analysis since it implies that decision makers can assess the total marginal welfare costs (i.e., the marginal effects on consumer and producer surplus) of environmental regulation with information that is simple to obtain. Lichtenberg, Parker,and Zilberman show that when marginal production costs for each region are equal to average costs, as in case of the step-function supply curve, the change in regional marginal cost at time t is equal to ( MC,it(qit,git) ait
Ptit
+ (dCit/it) 1i,t
where Vritis the percent change in regional yield, Yitis regional yield, and dCitis the change in per acre production costs, all at time t. While there may be a large number of regions and time periods considered [in fact this is preferred as it increases the accuracy of equation (9), as discussed later], each individual datum in equation (10) is readily available. Prices, regional outputs, and yields are published by state and federal sources. Changes in per acre yield and production cost can be obtained from noneconomists via survey methods, which we discuss in more detail below. Thus, demand and
1101
supply elasticities are not required to assess the marginal welfare impacts of environmental regulation; elasticities are needed only to decompose aggregate deadweight loss into producer and consumer surplus components. The simplicity of the method developed here is an important practical advantage over theoretically more exact, but also more complicated, procedures. Currently, many environmental regulations affecting growers are enacted with no economic analysis of their impacts. This lack of analysis is due in part to the complexity of many existing agricultural impact models, and to the fact that the basic data used in these models, especially supply and demand elasticities, are difficult to obtain. Equation (9) is a first-order approximation to the change in deadweight loss and thus is accurate only for small changes in marginal production costs. For large changes, it may be necessary to solve a system of supply and demand equations directly and compute new equilibrium prices following imposition of the environmental regulation.While more elaborate methods are required to assess the impacts of quantumchanges in environmental regulations, such as massive pesticide cancellations (e.g., Chambers and Lichtenberg), such regulations are the exception rather than the rule. Indeed, environmental regulations affecting agriculture are increasingly promulgated at the state or local level. There is an interesting connection between the impact framework in equation (9) and the "microparameter"or "putty-clay" class of production models of Hochman and Zilberman; Berck and Helfand; Paris; and Moffit, Zilberman, and Just. These authors have shown that local von Leibig production functions are consistent with Cobb-Douglas or other continuous aggregate production functions if there is continuous variation among individual atomistic productionunits. In the microparameterframework with a finite number of individual production units, each with positive measure, aggregate marginalcost is a step function, with the flat portion of each step given by constant marginalcost in a particularproductionregion. Each region also faces outputcapacity constraints due to a limited natural resource base or the type of production technology employed. In the finite microparameter case, the marginal change in welfare resulting from environmental regulation is approximated by equation (9) since variation in production conditions is completely captured by inter-regional differences in marginal cost (provided, of course, that the underlying yield and cost change estimates are accurately measured).
1102
Amer. J. Agr. Econ.
November 1996
$
p+dp p
MCI +dMC1
qI
MC1
ql
q
Figure 1. Welfare loss in the microparameter case
Figure 1 further develops the welfare loss calculations for the microparameter case. The level of each step of the base supply curve is constant marginal cost in a particular growing region, and the width of the step is given by the regional output capacity constraint. Changes in marginal cost resulting from environmental regulations alter the height of each step differently depending on the nonuniformity of the regulation and regional production and environmental conditions; indeed, the regulation may not change marginal cost at all in some regions. The shaded area in figure 1 is the gross marginal welfare loss as given by equation (9), which is the sum of the changes in regional marginal production costs. Note that this analysis can incorporate any type of supply shift (i.e., parallel, proportional, or more complicated types) as determined by the regional data. There is some ex ante uncertainty about the yield impacts of adopting alternative production technologies. Zilberman et al. suggest that this uncertainty should be explicitly incorporated into impact analyses of environmental regulations. Formally, they suggest treating the percentage yield change in a particular region and season, jit, as drawn from some density. Given this density, it is straightforward to calculate the mean impacts of the regulation, E(dW,), as well as the associated distribution of marginal impacts. It is convenient to describe the distributionof impacts in terms of confidence
levels, where a denotes the probability that the impacts exceed some level. Formally, define dWtaas the solution to prob(dWt> dW,a) = a. In the empirical example below, dWt is calculated for several different confidence levels. There are several possible sources for the density of yield impacts of the regulation in a particular crop, region, and season. The error on statistical assessments of field trials is one possible source, but this error may not represent actual farm conditions. Instead, it is preferable to conduct interviews with industry experts, including growers, pesticide dealers, university researchers, extension agents, and chemical company representatives. The survey responses are then used to calculate the distribution of impacts by a Monte Carlo method. Yield change estimates for each crop, region, and season are selected randomly from the set of survey responses, and marginal welfare impacts are then calculated according to equation (9). This procedure is repeated many times to create a set of marginal welfare impacts. Finally, the set of impact estimates is used to obtain dW,a for various confidence levels. Application to Mevinphos Regulation In this section the method developed above is applied to a specific problem:measuringthe marginal welfare costs of banning the mevinphos
Sunding
pesticide for vegetable production in California. Mevinphos (2-carbomethyoxy-l-methylvinyl dimethyl phosphate) is an insecticide-acaricide with contact and systemic activity that is most commonly used by California growers to control aphids on broccoli, cauliflower, head lettuce, and leaf lettuce. Mevinphos is used to eradicate aphids just prior to harvest so that growers can meet the stringent U.S. Department of Agriculture quality standards. The four crops considered here are produced in four California regions: Imperial Valley, Monterey, South Coast, and San Joaquin Valley. Output from other domestic areas, especially Arizona, Texas, and Florida, are aggregated into an "other domestic" category. Monthly crop prices and regional outputs are taken from the Federal-State Market News Service; 1990 is taken as the base year to avoid complications caused by the silverleaf whitefly infestation that began in 1991. Yields are taken from various University of California crop budgets from the counties comprising the four California production regions, and mevinphos use percentages for the California regions are found in the 1991 State of California Pesticide Use Report. The basis of the marginal welfare analysis, equation (9), indicates that it is necessary to compute the effects of the cancellation on marginal production costs separately for each crop, region, and season. Equation(10) relates changes in marginal cost to changes in per acre production costs and yields when growers adopt the next-highest profit alternative to mevinphos. Several alternative controls were considered: dimethoate, diazinon, thiodan, imidacloprid (available only to Imperial Valley growers under a Section 18 exemption), and pyrellin. Telephone interviews were conducted with fifty-six growers, pesticide dealers and applicators, extension advisers, commodity group representatives, produce packers and distributors, and university researchers to assess the yield effects from switching to each of the alternative aphid controls. Interview subjects were asked to give the countywide yield change resulting from replacing mevinphos with alternative aphid controls and specifically were asked to assess actual changes rather than report the results of experiments on highly managed plots. It is important to remember that the alternative chemical controls listed above have existed for many years, and survey respondents were generally familiar with the actual field performance of the alternatives. These survey re-
Marginal Cost of Environmental Regulations
1103
sponses are the basis for the Monte Carlo analysis of welfare impacts described in the previous section. Changes in per acre production costs are determined by calculating per acre chemical expenditures for each of the alternatives (at standard application rates and market prices) and subtracting this number from the per acre cost of mevinphos application. Generally, changes in per acre cost are empirically insignificant for the set of crops considered here, since chemical cost is only a small fraction of the crop budget. Table 1 shows the highest-profit alternatives to mevinphos for each of the crop, region, and season combinations, and also reports the associated yield and per acre cost changes. The highest-profit alternatives to mevinphos are imidacloprid (all crops in Imperial), diazinon (broccoli, cauliflower, and leaf lettuce in the remaining regions) and dimethoate (head lettuce in regions other than Imperial). Expected yield losses obtained from the industry survey vary by season and by region; generally, expected yield losses are higher in the Monterey and South Coast areas and higher in the summer months due to weather conditions favoring aphid growth. Table 1 also gives sample standard deviations for the assessed yield changes. Table 2 presents the marginal welfare costs of banning mevinphos use in California, calculated using equations (9) and (10) and the data in table 1. The expected marginalwelfare costs of banning mevinphos are $53.3 million annually,as compared to total annual revenues of $924 million for these four crops. Mean monthly deadweight losses, where the expectation is taken over the sample distribution, vary widely over the year according to market share among growing regions and yield and cost changes, thus underscoring the importance of disaggregating impacts over time as well as region. The marginal welfare costs of a mevinphos ban are highest for head lettuce since this is the largest market considered and a significant share of the nation's output comes from California. Mean losses are largest during MayJune and November-December, during which time most head lettuce is produced in California's Monterey and San Joaquin Valley regions. It is also interesting to note that mean losses are virtually zero during December-February, during which time nearly all output is produced in the Imperial Valley and other domestic regions that do not rely on mevinphos. Since disaggregating environmental regula-
1104
November 1996
Amer. J. Agr. Econ.
Table 1. Changes in Yield and Per Acre Cost Region
Highest Profit Alternative
Mean Percent Change in Yielda
Per Acre Cost Change ($)
Broccoli and Cauliflower Imperial
Imidacloprid
Monterey
Diazinon
San Joaquin
Diazinon
South Coast
Diazinon
0 (1.73) -10 (4.08) -4 (1.41) -8 (2.89)
50 -10 -10 -10
Head Lettuce Imperial
Imidacloprid
0 (0.00)
50
Monterey Winter
Dimethoate
-11 (4.91) -20 (7.12) -5 (5.00)
-10
Summer
Dimethoate
San Joaquin
Dimethoate
South Coast Winter
Dimethoate
Summer
-9 (4.04) -12
Dimethoate
-10 -10 -10 -10
(5.13)
Leaf Lettuce Imperial
Imidacloprid
0 (0.00)
50
Monterey Winter
Diazinon
-17 (7.53) -28 (9.83) -8
-10
Summer San Joaquin
Diazinon Diazinon
-10 -10
(3.54)
South Coast Winter Summer a
Diazinon Diazinon
-14 (4.78) -18 (2.89)
-10 -10
Standard deviations are in parentheses.
tions along the lines suggested in this paper has some cost implications for the administrative agency designing and enforcing the rule, it is importantto assess the social welfare value of a disaggregated as opposed to a uniform regulation. The Lorenz Curve, which is often used to represent inequality in the distribution of income, is a useful way of summarizing the ben-
efit from implementing nonuniform regulations. Figure 2 is constructed by first calculating the number of pounds of mevinphos applied and the marginal welfare impact per pound applied in each crop, region, and month. Next, the per pound marginal welfare impacts are ranked in ascending order, cumulated and plotted against cumulative pounds applied.
Marginal Cost of Environmental Regulations
Sunding
1105
Table 2. Expected Marginal Welfare Impact of Banning Mevinphos ($ Thousands) Broccoli
Cauliflower
Jan
253
100
31
144
528
Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
257 420 791 785 549 566 641 737 836 901 727
155 384 653 637 621 329 378 405 500 567 351
2 194 2,952 4,552 3,910 628 5,327 7,178 4,325 5,522 225
94 125 328 622 408 580 766 1,052 629 1,065 72
508 1,123 4,724 6,596 5,488 2,103 7,112 9,372 6,290 8,055 1,375
7,463
5,080
34,846
5,885
53,274
Month
Total
100
HeadLettuce
LorenzCurve
--
Leaf Lettuce
Total
- - - 45 DegreeLie
80 _ _
%Expected Deadweight Loss
60
-O.., '
.0 ,-- "
40
' _
20
0
,_0.
~.,-
..0
'__
80
100
%Mevinphos Use Figure 2. Lorenz curve for expected marginal welfare loss The heterogeneity in the gross marginal welfare impacts of banning mevinphos is depicted by the size of the area below the 45? line and above the Lorenz Curve in figure 2. Of course, if all uses of the chemical generated the same level of producer and consumer surplus, then the Lorenz Curve would lie on the 45? line. The size of the area above the Lorenz Curve and below the 45? line shows the welfare improvement from using nonuniform regulations as compared to proportional use reductions that treat each crop, region, and season similarly. For example, suppose that regulators wish to reduce mevinphos use by half. Nonuniform regulations that target specific crop, region, and season combinations can reduce welfare losses by over 50% as compared to a uniform proportional reduction that ignores differences in marginal productivity.
This type of Lorenz Curve analysis can be used to assess the benefits of many types of nonuniform environmental regulations. For example, many economists have observed that permit trading to reduce point and nonpoint source emissions can lower the welfare costs of improving air and water quality, as opposed to programs that proportionally reduce emissions. The method described here can measure the benefits of decentralized, market-based regulations, such as permit trading, that efficiently allocate the burden of emissions reductions vis-avis proportional reduction. Other potential applications of this technique include measuring the benefits of agricultural water trading and conservation reserve programs. The Monte Carlo simulation of gross welfare impacts using the expert survey also shows the importance of explicitly accounting for yield
1106
November 1996
change uncertainty when measuring marginal welfare impacts. Table 3 gives the marginal welfare impacts of canceling mevinphos at various confidence levels, represented as different choices of a. Ex ante uncertainty about the yield impacts of switching to alternative pest controls clearly affects the calculation of marginal costs. Mean marginal welfare impacts of banning mevinphos in California are $53.3 million annually;due to ex ante uncertainty about yield changes, impacts exceed $64.8 million annually with a 25% level of confidence and exceed $85.9 million with a 5% level of confidence. The magnitude of the confidence intervals on welfare impacts is significant. If producer and consumer surplus impacts are large in relation to total industry revenues, environmental regulations may result in bankruptcy and extreme consumer effects. Thus, firms and consumers may oppose environmental regulations based on worst-case impacts even if mean impacts are small in relation to the size of the industry. Implications for Policy Design Standard arguments in the economic theory of environmental policy suggest that regulations be set at their first- or second-best levels. In the notation of this paper, the first-best regulation satisfies max{B(git) - C(Qit)} Vi, t, where B((i,t) is the benefit from the regulation derived from higher environmental quality or public health, and C(it,) is the cost of the regulation in terms of lost producer and consumer surplus. Alternatively, the regulation may be set to min{C(Qi,)} s.t. B(iit) > , Vi, t, where P is some predetermined level of benefit. It is also possible to find efficient regulations incorporating the uncertainty inherent in the impact analysis by using dW,, in conjunction with "safety-fixed" rules (Kataoka) for environmental quality in the manner suggested by Lichtenberg and Zilberman. The large degree of variation in the confidence interval estimates in table 3 implies that there is value in obtaining better scientific information about the yield impacts of environmental regulations. A risk-averse regulator will change regulations significantly in response to uncertainty about marginal welfare impacts, and, thus, reducing ex ante uncertainty about yield changes will result in regulations that more accurately balance marginal costs and benefits.
Amer. J. Agr. Econ.
Table 3. Distribution of Marginal Welfare Impact a
MarginalWelfareImpact ($ Thousands)
0.05 0.10
85,872 74,275
0.25 0.50 0.75 0.90
64,838 53,274 41,870 32,336
0.95
21,214
Regardless of whether environmental regulations affecting agriculture are set at their firstor second-best levels, it is necessary for policy makers to assess the marginal cost of the intervention. The method developed in this paper is a simple algorithm for performing such an analysis. Expression (9) is an improvement over previous impact assessment methods in that it gives a theoretically appealing and easily computable formula for measuring region- and time-specific welfare loss.' The method developed here is valid for marginal changes in environmental regulations, including regulations promulgated at the state or local level. Global methods are needed to evaluate quantum changes in regulations. Finally, the framework developed here can be used to integrate economic information with existing earth science data in a single regulatory approach. Geographic Information Systems are rapidly gaining acceptance among environmental professionals. These data bases contain highly detailed information on spatial characteristics, such as land use patterns, and environmental conditions, such as groundwater depth and quality, soil characteristics, and microclimate. These data bases may also contain dynamic information for particular locations, such as lateral groundwater flow. GIS data is increasingly used to identify environmentally sensitive areas, for example, agricultural areas where pesticides have a high probabilityof infil-
It is interesting to compare the regional distribution of marginal costs and benefits. Since most citizens live in urban areas, the benefits from improving environmental quality will often be concentrated in these regions, particularly if environmental regulation results in increases in wildlife populations. The benefits from environmental regulations of agriculture can be local as well. For example, regulations to ensure farm worker safety and improve drinking water quality primarily will benefit rural residents. Producer surplus losses that are the bulk of the welfare costs of environmental regulations primarily are felt in rural areas.
Sunding
trating groundwater,or regions with high densities of an endangered species. The model developed in this paper can be used to measure the marginal costs of regulating agriculturalproduction at a disaggregated level, and it thus can be paired with detailed GIS data to give regulators a full picture of the marginal costs and benefits of localized environmental regulation. [Received July 1994; final revision received September 1996.] References Berck, P., and G. Helfand. "Reconciling the von Leibig and Differentiable Crop Production Functions." Amer. J. Agr. Econ. 72(November
1990):985-96. Chambers,R., and E. Lichtenberg."SimpleEconometricsof PesticideProductivity." Amer.J. Agr. Econ. 76(August1994):407-17. Hochman,E., and D. Zilberman."Examinationof EnvironmentalPolicies Using Productionand Pollution Microparameter Distributions." Econometrica (July 1978): 1-21.
Kataoka, S. "A Stochastic ProgrammingModel." Econometrica(February1963):181-96. Lichtenberg,E., D. Parker,and D. Zilberman."Marginal WelfareAnalysis of WelfareCosts of En-
Marginal Cost of Environmental Regulations
1107
vironmental Policies: The Case of Pesticide Regulation." Amer. J. Agr. Econ. 70(November
1988):867-74. Lichtenberg,E., R. Spear, and D. Zilberman."The Economics of Reentry Regulation of Pesticides." Amer. J. Agr. Econ. 75(November
1993):946-58. Lichtenberg,E., and D. Zilberman."EfficientRegulation of EnvironmentalHealth Risks." Quart. J. Econ. 102(February1988):167-78. Lichtenberg, E., D. Zilberman, and K. Bogen. "RegulatingEnvironmentalHealthRisks under Uncertainty: GroundwaterContamination in California." J. Environ. Econ. and Manage.
17(July1989):22-34. Moffitt, J., D. Zilberman,and R. Just. "A 'PuttyClay' Approachto Aggregationof Production/ PollutionPossibilities:An Applicationto Dairy Waste Control." Amer. J. Agr. Econ. 60(November 1978):452-59.
Paris,Q. "Thevon Leibig Hypothesis."Amer.J. Agr. Econ. 74(November1992):1019-28. State of California,CaliforniaEnvironmentalProtection Agency. Mevinphos Risk Characterization Document, Sacramento, 1994.
Zilberman, D., A. Schmitz, G. Casterline, E. Lichtenberg,and J. Siebert."TheEconomicsof PesticideUse andRegulation."Science254(December1991):518-22.
