Trade-Off Analysis of HerbicideWithdrawals on AgriculturalProduction and Groundwater Quality

Shiping Liu, Gerald A. Carlson and Dana L. Hoag*

Abstrac(

This study examines the trade-off between agricultural production and groundwatercontamination potential for ten potential herbicide cancellations. Theoretical and empirical modelsare developed for estimating losses in consumer and producer benefits in the agricultural commoditymarket and changes in groundwater quality. Using com and soybean production in the southeasternCoastal Plain as a study area, the analysis concludes that (1) effects of herbicide cancellations ongroundwater quality can be very significant; (2) a cancellation does not guarantee groundwaterquality improvement; (3) effects of a multiple cancellation are different from the summation of theeffects of independent cancellations; and (4) weed density has a very strong effect on losses tofarmers and consumers from cancellations, but output demand and supply elasticities do not.

Key Words: herbicide cancellations, corn/soybean supply shifts, and groundwater quality.

Introduction

Groundwater is an important resource in theUnited States that is sometimes threatened bycontamination from agricultural pesticides (Nielsonand Lee 1987; USEPA 1990). More than 70pesticides have been found in groundwater in 38states (NGA 1989). One regulatory response topesticide residuals in groundwater is to suspend orcancel use registrations for those pesticides that maylead to groundwater contamination. Herbicides arean important part of crop production, andcancellation can increase production costs if highercost herbicides are substituted or if the substitutesare not as effective in controlling weed damage, Acancellation, however, does not always guarantee areduction in risk of contamination by pesticides.

Cancellations sometimes lead to increased humanrisk through replacement pesticides (NationalResearch Council 1987), The United StatesEnvironmental Protection Agency (USEPA) has theresponsibiht y under the Federal Insecticide,Fungicide, and Rodenticide Act (FIFRA) toformulate policies for the use of agriculturalpesticides to balance the benefits of use againstenvironmental risks (Osteen and Kuchler 1987;Gianessi et al. 1989).

The social impact of banning, canceling, orvoluntary withdrawal of a pesticide can bedetermined through a detailed analysis of changes incosts and benefits. The costs to society are theconsumer and producers’ losses (reduced pest

control benefits) in the related agricultural

*Shiping Liu is a visiting scientist at the Center for Agricultural and Rural Development, Iowa State University, Ames,Iowa. Gerald A. Carlson is a professor of Agricultural Economics at North Carolina State University, Raleigh, NorthCarolina. Dana L. Hoag is an associate professor of Agricultural Economics, Colorado State University, Fort Collins,Colorado. The authors thank Rebecca Olson for typing and formatting. The authors also thank David L. Debertin andtwo anonymous reviewers for their helpful comments. Remaining errors are our own. This is Journal Paper No. J-16128 of Iowa Agricultural and Home Economics Experiment Station, Ames, Iowa. Project No. 2872.

J. Agr. and Applied Econ. 27 (1), July, 1995:283-300Copyright 1993 Southern Agricultural Economics Association

284 1.lu.C’at’l.wmand Hoag Trade-of] AIIUIV.TI.Yo/ Ilerhtclde Withdrawals on .4grtcullural Production

commodity market. i The 10SSCSoccur because ofa downward shift in the supply curve of pesticidc-using crops~ (Just et al. 1987). The benefit (riskreduction) for the society of a pwticidc cancellationis an improvement in environmental quality. Thesecould include to protect groundwdter, to protectsurface water, to protect applicators (for applicator’ssafety), or to protect wildlife. The focus of thisstudy is groundwater, which is frequently cited as arationale for cancellation (Batle et al., 1989).

The purpose of this study ]s to cwduate theimpact of hypothetical corn and soybean herbicidebans (cancellations) on crop output, herbicide USC,economic returns and potential groundwatercontamination. Seven single or multiplecancellations of corn herbicides are considered:Atrazme (atrazmc), J3anvel (dicamba), Dual(metolachlor), Atrazinc and Banvel, Atrazme andDual, Banvel and Dual, and Atrazinc, Banvcl andDual. The three soybean herbicide cancellationsconsidered arc: Lasso (aPdchlor),Dual (metolachlor),and Lasso and Dual, These hypotheticalcancellations were chosen because each of theseherbicides has been found in groundwdtcr and theupesticide leaching potentials (~LP) are high relativeto those of other corn and soybean herbicides(Danielson ct al. 1993).

The primary contributions of this paper are (1)to provide new methods for compuhng potentialeconomic costs in agricultural commodity marketsof pesticide cancellations and (2) to analyze theeffects of both single and multiple herbicidecancellations in corrv’soybean markets andgroundwater quality for the study region, usingdetailed data about different weeds by crop. Wc

proceed with a discussion about the calculation ofthe costs and benefits of a cancellation, followed bya detailed discussion of estimated costs and benefitsfor the hypothetical cancellations in North Carolina,and three nearby states (Georgia, South Carolina,and Virginia). We conclude by presenting tradeoffsor comparisons of the costs and benefits of variouscancellations and a sensitivity analysis to examinemajor fttctors influencing the results,

Calculating Costs and Benefits

To conduct a cost-benefit analysis for pesticidecancellation decisions, costs and benefits must bc

both defined and measured. on the cost side, it isdifficult to predict the management response thatFarmerswill undertake to compensate for a pesticidecancellation, The most common assumption is thatother competitive pesticides will take a share of thesales of the canceled pesticide proportional to theircurrent share of the market (Grube 1992; Liu,1993), This method, however, fails to account forthe umqucness of the substitutes for controlling theprimary pests targeted by the canceled product orfor cropping or output adjustments that would occur.After a dcpailed comparison and discussion ofdifferent methods, Liu (1993) developed an indexthat will be used here to estimate potential substituteherbicides and their final use levels for the tenherbicide cancellations,

The shifts in agricultural product supplyassociated with a withdrawal of a herbicide can bedecomposed into three parts: shifts due to yield loss,those resulting from an input cost increase, andthose attributable to cropping adjustments. Themost commonly used method cited in the literaturefor cstlmatmg yield losses and changes in controlcosts is expert opimon (Kennedy et al. 1975; Tayloret al, 1979; Smith ct al. 1990; LAPIAP 1993). Onemajor problem with expert opinion is the widevariation among different experts (Osteen 1992), Inaddition to this technical problcm, the expertopinion method has been challenged in court cases(Jennings 1992; Housenger 1992). However, expertopinion is still the most widely used method in mostcase studies for estimating yield effects and changesin production costs of herbicide cancellations(Ferguson et al. 1992; Stemeroff et al. 1993). Themajor reason for the wide usage of expert opinionis lirmted options (Ostecn 1992),

Analytical models, such as the weedcompetition model called HERB (Modena et al.1991; Wilkcrson ct al. 1991; Coble and Mortenscn1992), can be used rather than expert opinion toestimate yield losses. These models account forspecific weed species and the differential impacts ofalternative pesticides on each weed type. However,they are rareiy used because they are designed forthe field level, and most FIFRA cancellation studiesare regional or national in scope. In this study wedo have farm survey information about weeds andpesticide use for four states from a private firm,Maritz Marketing Research, Inc,3 Therefore, we

J Agr and Applled EcotI JuI.v, 199S 285

are able to utilize the HERB weed competitionmodels together with weed specie reformation toestimate shifts in supply due to yield changes for aregion,

On the benefit side, it is difficult to obrdin aprecise estimate of physical environmental qualityimprovements with a pcstlcide cancellation becauseso many factors vary across croplandsite~hemical-physical and biological properties ofherbicides, properties of soils, agricultural practices,and climatic and hydrogeologic condmon%--andaffect groundwater contamination potential (Chcngand Koskinen 1986; IIelling and Gish 1986; Jury etal. 1987; EPA 1987; Nielsen and Lcc 1987; Wcber1990; Weber and Warren 1993). Lack ofknowledge about tlctors affecting pestlcldc leachingpotential and the complexity of the process make itdifficult to make precise estimates of the content ofgroundwater pollution. Several indices and modelshave been developed, such as GUS, DRASTIC,PLP/SLP matrix, screening models, simulationmodels, metamodels and Ground WaterContamination Potential (GWCP) (Aller et al. 1986;Gustafson 1989; Jury et al. 1987; Weber 1990;Hoag and Hornsby 1992; Bowzaher et al. 1993;Danielson et al. 1993; Wcber and Warren 1993), butnone of these has emerged as clearly superior toothers, To be suitable for measuring potentialgroundwater quality improvement by a pesticidecancellation for most economic analyses, a modelthat is structurally simple and physically meaningfulis desired. After a careful comparison, the pesticideleaching potential model PLP (Webcr and Warren,1993) was used to estimate groundwater pollutionpotential from pesticide uses in large scale (fourstates in the Southeastern Coastal Plain---eorgia,North Carolina, South Carolina and Virginia).

Cancellation Costs

The major changes in private production andconsumption costs duc to regional or nationalpesticide cancellations are decreases in crop yieldand increases in pesticide material and apphcationcosts. These two effects can be expressed throughshifts in the supply curve due to the cancellation,For simplicity, several assumptions are made. First,there are n identical farmers involved in agriculturalproduction of com or soybeans. Second, both thepre- and post-regulation margmal cost fi.mctions

(short-run supply functions) are linear in the rangerelevant for the cancellations (Kopp and Krupnick1987; Gianessi et al. 1988; Danielson et al. 1993).Third, a regional level demand function is linear.With these assumptions, supply curves before andafter cancellations can bc obtained by using thefollowing quadratic profit-mammzation function

Max x = P*Q - C(Q)= P-FA - m“Y”A (1)

- 0.5+(A.Y)2 - R(A),

where

Yis ylcld per acre,A is acres used for the production,P is the output price,Q is the output (= Y*A),m and k are coefticicnts related to marginal

cost of production, andR(A) is a fried cost such as land rent.

Taking the derivative of Equation 1 withrespect to Q yields the first-order condmon;marginal cost (MC) is equal to marginal benefit(MB)

MB = P = A’lc(,=k.Q+~ (2)

= k.(A”Y) + m.

This MC, equation can be used to representthe initial short-run supply curve (S.) in Figure 1; kis the slope of initial supply curve SOand m is theintercept. The demand function represented by Do

in Figure 1 can be expressed as

Q=ct-~”P, (3)

whererz and [~ arc the intercept and slope of the

demand curve.

The effects of a herbicide cancellation areseparated into two parts: average control costchange and yield loss. Considering yield loss first,marginal cost would increase and the supply curveshift left (represented as S, in Figure 1) with theloss of a herbicide. The ratio of k, (slope of curveS,) to k is equal to Y. (the average yield per acrebefore a regulation) over 1’, (the average yield per

286 Liu, Car[son and Hoag. Trade-off Ana/vsts of Herbicide Withdrawals on Agricultural Production

Figure 1. A Simple Model for Measuring Welfare Loss in a Given Agricultural Product Market from aHerbicide Cancellation.

mca$rmL

P3 ----

P2

t)

a;,

%: Q2

o Ql Qo @w(mlllimk.)

acre after regulation). The ncw equilibrium Ac can be positive or negative. Material costs willcorresponding to this special case will be (Pj, QJ) inFigure 1.

If weed control cost for herbicide materialsand application costs also increase, the supply curvewould shift even further to the left, to .’ij. Thiscurve can bc determined by S, and the herbicidecost change. The vertical distance between S,, andSz will equal the change in herbicide cost (ab inFigure 1); given linearity, it can be represented witha change in the intercept. If the change in averagematerial cost per output unit (bushel) is Ac, then thenew supply curve Sz can be expressed as

Sz =MCZ =k,.Q + m + Ac

(1.’) (4)

increase if a producer substitutes more expensiveherbicides to maintain weed control, however, costswill fall if the grower reduces weed controlobjectives in response to the ban.

The demand and supply equations can be setequal and solved for two unknowns—newequilibrium price (Pj) and quantity (Qj). ComparingQ, with Q,(Y1 “Ao), it is possible to determinewhether or not more land would be used forproducing the agricultural crop under study. If Qjis greater than Q!, then more land should be usedfor production because the price increase wassufficient to offset the yield loss and control costincrease. Otherwise, some land used for theproduction of, say, corn, would be shifted toproduction of another commodity, say, soybeans, ifland and other input prices do not change because

J Agr and App[ied Econ, July, 1995

of a herbicide cancellation (Taylor and Hewitt inCarlson, Zilberman, Miranowski, p, 149-150, 1993).

Yield Loss

Average percentagecan be estimated with

5‘;(s,;-s;)“[1

reduction in yield (“AAY)

-D,, /100] ~(5)

1=1 ,=1.

.1.1-<

m S’;”[1-D, /100]1=]]=1

where

AY is the average change in yieldfollowing a herbicide cancellation,

Yhis the average yield per acre beforethe herbicide cancellation,4

YO is the expected weed-free yield(Taylor et al, 1979),

Y. is the average yield per acre after theherbicide cancellations

SY”is the share of post-regulation acretreatments of pesticide i for weed j,

SYbis the share of pre-regulation acretreatments of pesticide i for weed j, and

D,, is the percent damage (yieldreduction) after the treatment.

Syb,SV”,and D, can be estimated from studyarea data. The values of SV~were obtained for allmajor herbicides and all major weeds from theMaritz survey information. If a herbicide is to becanceled, its post-regulation share is zero. If aherbicide is a potential substitute for the herbicideto be canceled, its post-regulation share is unknown.The shares of potential substitute herbicides for agiven cancellation can be estimated by

S;= S;+ SH, +;,

where

SY[’is the share of post-regulation acretreatments of pesticide i for weed .j,

SVhis the share of pre-regulation acretreatments of pesticide i for weed j,

SH, is the substitute percentages ofherbicide i for the one to be canceled,which was estimated with the indexmethod (Liu 1993),Cand

S.,b N the share of pre-regulation acretreatments of pesticide A (to becanceled) for weed ,j.

In equation 6 we assume that thesubstitutability of herbicide i for A is the sameacross all weed species. It is also assumed that thetotal area treated by the herbicide to be canceledwould remain as treated area with various substituteherbicides, Mechanical control and cultivation areconsidered in this study, but they are not standalone replacements. This is accomplished byallowing cultivation to be part of each herbicideoption i, and effecting efficacy and costs.

Yield loss due to weeds can be estimated bya competitive load index (Modena et al. 1991;Coble and Mortensen 1992). The competitive loadof weed specie ,j after treatment with pesticide i(TCLY) can be estimated by

TCLU = [1.0 - K, (X)] ON,(.CI,, (7)

where

K,,(X) indicates the proportion of weedspecies j killed by herbicide i withdosage X, referred to as efficacy,

N,, is the amount of weed ,j present perunit of area at time t, and

287

(6)

288 Liu, Carlsan and Hoag. Trade- Ofi’A nal~sls ojIIerbicide Withdrawals an Agrmdtural Production

C~ is the competitive index for eachweed species, it is estimated based onexperimental data by weed scientistsCoble and Mortenson, 1992),

The herbicide efficacy of specific weed speciestreated at different times in the season was obtainedfrom weed control manuals prepared by weedscientists in each state. The target weed species inNorth Carolina (about 40 species) were obtainedfrom the Maritz Farmer survey. Frequencies forlow, medium and high weed densities in NorthCarolina were estimated for each weed species byweed scientists at North Carolina State University.The scientists were asked to think of low denslticsas a good year (one in 20 years) and high densitiesas the worse case (one in 20 years),

Using TCLq in Equation 7, the percent of yieldloss afier use of herbicide i against weeds) for corncan be calculated as (Modenta el al., 1991):

D,, s I 0.14 (K’L,, -1)014+

[1 + 0.002333 (TCL,, - 1)]if TCL,, > I

(8)

The percent of yield loss aiter use of herbicide i forsoybeans can be calculated as :

I

O 5 .TCL,, tf TCL,I<50

D,, = 55~TCL,, -50)(9)

25 +(TCL,,+60)

// TCL,,>50

Yield losses for each cancellation and eachweed density (high, medium and low) were basedon the 12-year (1979-1990) avemgc corn yield(76.25 bushels/acre) and the weed competitionindex. In addition to weed pressure, average yieldloss depends on the efficacy of the herbicide to bcbanned relative to its substitutes and the percentageof fields treated with the herbicide to bc canceled.Average yield 10SSCSacross North Carohna rangedfrom 0.00 (0,00 percent) bushels pcr acre for aBanvel ban with low weed pressure to 2.16 (2.84percent) bushels per acre for an Atrdzinc, Banveland Dual ban with high weed pressure (table 1).7The Iargcst average yield loss for a single herbicide

ban was for Atrazine (1.56 bushels/acre or 2.05percent), The index method accounts for the weedsbeing treated by the banned herbicide and theefficacy of the substitute herbicides on those weeds.Multiple year or weed dynamic effects are notconsidered because of frequent crop rotation in thestudy area.

The same procedure is used for the threesoybean bans, Using 12-year (1979-1990) soybeanaverage ylcld (23.92 bushels/acre) as the yieldbefore the cancellations, average yield losses acrossNorth Carolina ranged from 0.01 bushel per acre(0.04 percent) for the Dual ban with low weedpressure to 1.22 bushels per acre (5.11 percent) forboth Lasso plus Dual ban with high weed pressure(table 2). The largest average yield losses for asingle herbicide ban are for Lasso. The averageyield loss per acre is 0.78 bushel with high weedpressure. This is about 3.26 percent of the pre-rcgulation average yield.

Control Cost Changes

Another component in estimating supply shiftsfor proposed pesticide bans is cost changes inpesticide materials. Average material cost changeper acre (AC = AC * average yield per acre) can becalculated by

where

CYis the average cost per acre treatmentwith herbicide i for weed ,j,

(10)

S,)” is the post-cancellation acreageshdres for pesticldc i with weed ,j, and

.$,: is the pre-cancellation acreage sharesfor pesticide i with weed ,j.

The CVmaterial costs were obtained from theMaritz farm survey; it measured expenditure peracre treatment for each herbicide applied at a givenhme used to control each specific weed species,

Estimated cost changes corresponding to theseven corn herbicide cancellations are given in table

J Agr und Applied Econ Jul\, 1995 289

Table 1. AverageYieldLoss,HerbicideMateriatCost Increase md BenefitImssperAcre for the SevenCorn HerbicideCancsdiationsin North Carolina

Atmzine Banvel Dual A&B A&D B&DBan But

A, B&D ●

Ban Bsn Bso Bstt Ban

weedDensity Percent Reductionin Yield

Low 0.46 0.00 0.02 0.61 0.47 0.09 0.62Medium 0.82 0.01 0.09 1.12 0.86 0.24 1.16High 2.05 0.02 0.11 2.78 2.09 0.47 2.84

Initisl Post-ban Herbicide Material Gst ($/Acre)

cost 6.75 7.87 6.68 6.63 7.84 7.81 6.61 7.80costIncrease 1.12 -0.07 -0.12 1.09 1.06 -0.14 1.05

weedDensitv Benefit LCISSin Com Market ($/Acre)

Low 1.99 -0.06 -0.08 2.24 1.96 0.04 22?

Mahum 2.67 -0.05 0.04 3,22 2.69 0.31 3.25Hi-gb 5.03 -0.03 0.09 647 5.06 0.75 6.50

a A = Atraztne A&D = Atrazinc and DustB = Banvel B&D = Bsmvel and DustD= Dual A, B&D = Atmzine, Bsovel, and DualA&B = Atrszine and Banvel

Table 2. Avemge Yield Loss, Herbicide Material Cm.t Iocrcsse , snd Benefit LOSSper Acrefor the Three Soybean Herbicide Cancellations in North Carolina

hSSO snd Dust

Lasso Bsn Dusl Bsn Bsn

Wed Density Percent Reduction in Yield

Low 0.36 0.04 0.10Medium 1.27 o,~z 1.42

High 3.26 0.74 5.11

initial Post-bsn Herbicide Material Cost ($/Acre)

cost 9.59 9.35 9.46 9.14

Qst Increase -0.24 -0.13 -0.45

Weed Density Benefit Loss in Corn Market ($/Acre)

LOw 0.29 -0.08 -0.30

M.dum 1.64 0.19 1.65

High 4.63 0.96 7.24

1. Both average herbicide material cost pertreatment acre and average cost reduction per acreare given.x The net change depends on the pricesof substitute herbicides and the level of control thegrower stnvcs to achieve. Changes in avemgcherbicide material costs range from –$0, 12 pcr acrefor the Dual ban to $1.12 pcr acre for the Atrazineban. Average material costs are lower after Dualbans bccausc Dual is a relatively cxpenslve ($ 11,42per acre), and its substitutes are relatively cheaper($3,62 for Atrazine, $9,58 for Lasso, and $3.89 forPrinccp) (L1u 1993). A ban on Atrame yields theopposite result as the Dual ban. The changes inabsolute value also depend on the initial shares oftwo herbicides to be canceled. For example, the

relatwely high percentage change in average controlmaterial costs for the Atrazine ban is due in part toa big initial share of Atrazinc (31. 17’?’’o). Therelatively small percentage change in averagecontrol material costs for the Dual ban is because ofa small initial share of Dual (2.977,).

Results for material cost reductions per acreon soybeans in North Carolina mnge from a –$0.45to –0. 13 (table 2). Iierbicidc material costsdecrease for either of the two single bans or the banof both together because Lasso and Dual arerelatively expensive herbicides compared to theirsubstitutes,

290 LCU,Car[son and Hoag: Trade-OffA na[~,sis of Herbicide Withdrawals on Agricultural Production

Economic Assessment of Benefit Losses

Knowing changes in price and quantity ofoutput, it is possible to estimate the social welfarelosses from a regional herbicide cancellation oncorn and soybean consumption and production. Theloss can be approximately estimated by (Griliches,1957; Danielson et al., 1993)”:

AICS+lWj = AQ.(1 +0,5.AP)4J{), (11)

where

A[CS+P5’1 is the local changes inconsumer plus producer surplus per acre,

AQ is the change in yield per acre pluschanges in control costs in bushelequivalents,

PO is the base corn price, which for the12-year period (1979- 1990) was $2.45per bushel,’” and

AP is the regional output price changefrom the particular herbicide cancellationin percent.1’

All variables can be estimated from thechanges in control cost and yield outlined inprevious sections except for the change in outputprice (AP). To estimate the change in price, it isassumed that the changes in both price and quantitybefore and after each cancellation are limited,Therefore, both linear fimction and constantelasticity can be assumed.’z With theseassumptions, the solution for Pj is’~

~ p

1+1-Ac.AP3=P,)+l - E E PC,

Y,,(12)

l+n T— “—E Y,

where

E is the regional supply elasticity of cornor soybeans,

q is the absolute value of the regionaldemand elasticity of’the commodity, and

all other variables are definedas in previous equations,

Once P3 is estimated, the total percentagechange m price can be estimated, We useinformation about average yield loss (AC) in NorthCarolina and assume that the percentage change inyield from each of the cancellations for a given cropis the same at every location in the four stateregion, The same assumption is made for thechanges in herbicide material costs. With theseassumptions, Equation 12 can be used to calculatethe price of output after cancellations and tocompute the consumer and producer loss in benefitsfor North Carolina.’4

The estimated losses in consumer plusproducer surplus in North Carolina for the sevencorn herbicide bans, under low, medium and highweed density, are given in the lower part of table 1.The producer plus consumer surplus lossescomputed by this method reflect local herbicide usepatterns and weeds to give local yield and controlcost changes rather than national changes for eachof the ten hypothetical cancellations. For cornproduction, these total cost changes range from lessthan –$0.06 per acre for a Banvel ban with lowweed density to $6,50 per acre for the Atrazine,Banvel and Dual ban with high weed density. Weused a demand elasticity of –.21 and a supplyelasticity of ,48 based on Gardiner et al, (1989), butfound that our total surplus changes were notsensitive to demand and supply elasticitychanges.’s All losses are expected to be positive.Negative values in table 1 were not significantlydifferent (statistically) from zero. Only primaryweed species in fields are considered in this studydue to limited density information, This omissionexplains the small negative values in table 1.

From table 1, it is clear that A(CS + PS) isincreased as weed density increases, This isexpected because herbicide efficacy is moreimportant when there are high weed densities thanfor low weed densities. The largest single benefitloss was for Atrazine. However, multiplecancellations had stronger impacts than canceling allrelated herbicides independently, For example theAtrazine and Banvel bans alone result in losses of$5.03 and -.03, or a sum of $5.00, but the A&Bcolumn shows a loss of $6.47 per acre.

J. Agr, und Appked Econ , .JuIv 1995 291

Similar computations were made for the threesoybean herbicide cancellations and are given intable2. The basic findings aresimilar to those forcorn herbicide cancellations; production lossincreases as weed density increases, The effects ofbanning Lasso and Dual together IS stronger thanbanning both of them independently with high weeddensity. The elasticities used in the soybean weedcalculation are based on Gardiner ct al,’s (1989)estimates for soybeans of –0,42 for dernmd and0.60 for supply. Again, changing these values toreflect regional demand and supply elast lcities doesnot significantly alter results.

As discuwed in the previous section, whethera cancellation makes sot;lety betier or worse offdepends on both risk reductions and losses toconsumers and producers, The results estimated inthis section are just the losses in benciits in theagricultural commodity markets, Potentialgroundwater quality improvement from a herbicidecancellation is another benefit that could enter mtoassessments of whether a herblcidc should becanceled.

Changes in Groundwater Quality

One of the primary purposes of herbicidecancellations has been to improve groundwaterquality. As discussed in previous sections, however,a herbicide cancellation dots not necessarilyguarantee groundwater quality improvement. Thisdepends on the groundwater pollution potential ofthe herbicide to bc canceled and those of itssubstitutes. Given an estimate of potentialsubstitutes for several potential cancellations, thechanges in groundwatcr quality can also bcestimated, One way to estimate the groundwaterquality improvement due to herbicldc cancellationsis to estimate changes in pcsticldc leachingpotential, especially for large-scale estimates (state,regional or national levels) when it is almostimpossible to include soil information into thecalculations. Changes in pesticide leaching potentialat the state lCVC1were cstirmtcd by using (Wcber1990; Danielson et al, 1993; Weber and Warren1993):

APLP%=A~.100PLP

,.A,’’”PLP, -&”PLP, (13)

_ ,=1 ,=1— .— “loo ‘,={~S,’’VPLP,,=1

where

PLP, is pesticide leaching potential scoreror herbicide i,

S,”is herbicide i’s treatment share beforethe cancellation, and

S,” is herbicide i’s treatment share afterthe cancellation.

The pesticide leaching potential (PLP) wascalculated as:

PLP = T,,. uR “ F/Koc , (14)

where

PLP is herbicide leaching potential,

Koc is pesticide retention by soil index,

T,,j is the half-hfe of pesticide in thefield for the region,

R is the rate of pesticide applied for theregion, and

F is the fraction of pcsticidc hitting thesoil, which depends on crop canopy wc(Weber and Warren 1993; Danielson etal. 1993).

Onc of the advantages of this type of index isthat it is very simple and can be understood easilyby farmers. Also, it can be used easily to estimatepotential change in leaching for large geographicalareas for potential cancellations (Danielson ct al.,1993). The fraction of pesticide hitting the soil andthe application rate give a precise estimate of theamount of pesticide hitting the soil. One of the

292 Liu, Curlson and [{qg 7t mle-ofl AnalMIs of [Ierlmde Wl(hdrawals on Agrlcultura[ Production

disadvantages of this model is that it still does notgive an absolute estimate of the amount or thepercent of a pesticide that will reach groundwater.This is a relative measure of leaching acrossherbicides.

The reductions in PLP corresponding to theseven corn and three soybean herbicidecancellations are estirndted for four states and givenin tables 3 and 4.16 The percentage reductions mPLP are quite different among differentcancellations in each state and among differentstates for a given cancellation, The changes in PLPare different within each state because Initial shares,substltutablhty of other herbicides, and lcachmgpotentials are different for each herbicide, Thepercent changes in PLP vary for different statesbecause initial shares and substitute herbicides aredifferent for different states, For example, theperccnwgc changes in PLP are much larger for thecarwelkmon of Atrazinc than those for thecancellation of Banvel for each state. This isbecause the initial share of Atrwinc for each stateN much larger than that of 13anvel, For thecancellation of Atrazine, the percentage change mGeorgia ISmuch larger than that in Virginia becausethe share of Atrazinc in Georgia IS 53.06 percentcompared to 30.38 percent in Vmgmla (Liu, 1993),In addition, there are different substitute herbicidesfor the same cancellation in each of the states. Thehigh percentages of substitution of Pnncep forAtrazine in Virginia leads to very limlted chdngcs inPLP because Pnncep’s PLP is almost as high asAtrdzinc’s.

The range of reduction in PLP among allstates for the seven corn herbicide bans is from–2.29 to 75,61 percent and from –9.20 to 60.63 forthe three soybean bans, depending on theassumptions used for estimatmg the shares ofsubstitute herbicides, The largest PLP reductionfrom a single herbicide cancellation is for bans ofAtrazine, The smallest PLP reduction is for bans ofDual. The PLP even increases in three of fourstates for the ban of Dual, This change is notstatistically significant but it is in the direction ofdlustratmg the National Research Council’sconclusion that: “Cancellations sometimes lead toincreased human risk, as replacement pesticides areused at high rates.” Whether a cancellation willincrease or decrease groundwater pollution potential

(or PLP) depends on the initial shares of theproduct to be canceled and its substitutes as well ason the relative pesticide leaching potential of eachrelated pesticide.

From tables 3 and 4, it IS clear that the bothcost and PLP effects of the ban of severalherbicides at the same time are not equal to the sumof the effects of each herbicide ban consideredindependently. The major reason for this M thatpotential subshtute herbicides arc different acrossherbicide products, If two products are canceledthey can no longer act as replacements, and totaleffects are larger with multiple cancellations.

Trade-offs of Costs and Benefits of PotentialHerbicide Cancellations

Knowing the potential benefit losses inagricultuml markets and the potential benefitsgained from cnvironrnentd quality improvementbecause of the possible ban of a herbicide isimportant in making decisions to cancel herbicides.1Iowever, studying these effects separately, as domany studies, does not allow comparisons of costand benefit changes including herbicidesubstitutions, and thus may not accurately refleeteconomic or environmental impacts (Ferguson ct al.1992). According to FIFRA, decision makers mustformulate policlcs regarding the use of agriculturalpesticides so as to balance the benefits against

environmental risks. Therefore, both A (PS + CS)and changes in PLP must be analyzed together.

Of primary interest to policymakers is theextent to which the costs of pesticide cancellationstrdnslate into a reduction in groundwatercontamination potential or pesticide lcaehingpotential, Changes in the PLP are not easilyconverted into changes in monetary values andtherefore cannot be directly integrated intocost/benefit analysis, IIowcver, the change in PLPcan be directly compared to changes in costs andbenefits from a cancellation, The results for NorthCarolina are expressed graphically in Figures 2 and3, which show the relationship between averagelosses of benefits per acre and the estimatedpercentage reductions in pesticxde leaching potentialfor the seven corn herbicide cancellation and thethree soybean herbicide cancellations.

J Agr and Applied EcmI Julv 199.$

Table 3. Corn Herbicide Lacbing Potential (x11X3)and Percentage Cbqms

Herbicide Lcacbing Potential Before and After cancellations

293

S&we ti]tisd Ban A BrIn B BmD Bm A&B Ban A&D Bsn B&D Ban A, B&D ‘

GA 73.’70 25.30 72.75 74.96 22.73 20.71 74.01 17.97NC 63.88 36.38 62.68 63.74 34.59 3~.52 62.54 31,29

Sc 71.87 41.60 71.35 7~34 40.94 39.31 71.83 38.80VA 78,84 59.32 77.12 80.65 57.10 57.26 78.93 55.23

Percent Reduction of Herbicide Leaching Potentials

GA 65.67 129 -1.70 69.16 71.!KI -0.41 75.61NC 43.06 1.89 ().~z 45.85 49.09 2.11 51.02Sc 42.11 0.71 -0.66 43.03 45.30 0.05 46.01VA 24.76 2.19 .:.29 27.57 27.37 -0.11 29.95

a A = Atmzine A&D = Atrazine md DualB = Banvel B&D = Banvel and DualD= Dual A. B&D = Atrarine, Banvel, aod DualA&B = Atraz.ine end Benvel

Table 4. Sovbcan Herbicide Lc-aching Potentisd (X100) snd Percentage Cbsngw

Herbicide Leading Potential Before and After Cancellations

Sate Initial IJLSSOBan Dusl Ban biSSO and DualBan

GA 12.29 10.55 10.85 8.72NC 16.61 13.43 15.43 9.09Sc 13.57 11.39 11.57 8.74VA 23.24 12.67 25.37 9.15

Percent Reduction of Herbicide Leaching Potentials

GA 14.16 11.72 29.11NC 19.14 7.12 45.26Sc 16.06 14.71 35.58VA 45.46 -9.20 60.63

Losses in benefits per acre increase as weeddensity increases for each herbicide cancellation.The basic tendency is an increase m the reductionsin PLP associated with an increase in benefit losses,but there are some notable exceptions. As shown inFigures 2 and 3, Atrazine and Dual bans Icad tolarger reductions in groundwater contaminationpotential than do Atrazine and Banvcl bans, but thebenefit losses for Atrazine and Dual bans are lessthan those for Atrazine and Banvel bans,

Summary and Conclusions

Ten hypothetical herbicide bans, seven forcorn and three for soybeans, were cvaludt~dutilizing herbicide and weed ddta in North Carolina.Both the impact on producer and consumer surplusand the environmental impact on groundwatercontamination potential were examined, For theseven corn herbicide cancellations, producer andconsumer benefit losses per acre ranged from

-$0.08 to $6.50 per acre in North Carohna,depending on weed density and replacementherbicides. For the three soybean herbicidecancellations, producer and consumer benefit lossesper acre ranged from -$0.30 to $7.24. For theseven corn herbicide cancellations, pesticideleaching potential (PLF’) reductions (used asmeasurements of groundwatcr quality improvement)m North Carolina ranged from 0.22 to 51.02percent. Reductions in PLP for the three soybeanherbicide cancellations ranged from 7.12 to 45.26percent, At the state level, the reductions PLPamong four states (Georgia, North Carolina, SouthCarolina and Virginia) ranged from -2.29 to 75.61percent for the seven corn hcrblcide cancellationsand ranged from –9,20 to 60,63 for the threesoybean herbicide cancellations.

The effects of herbicide cancellations ongroundwater quality can be very significant, but acancellation does not guarantee groundwater quality

294 [’w c’lv’lsotl [1!1[/Iloug lrad~,-off Anulv.sts ofllerhwlde Wtthdruwal.? (J!I Agncul(ura[ Productmn

Figure 2. Relative Changes In PLP and pcr Acre Consumer Plus Producer Losses in Benefits for Seven1hypothetical Corn IIerbicidc Canccllatlons In North Carolma.

8Weed Density

7 law

6 Medium3t15 HIsI)

&

~4a63n.

~’

1z

: 0,

/-1 D B B&D A A&B A&D A B&B

022 1.89 2.11 43.06 45.a5 49.09 51.02

FkJrceni Ft8dUCl10n6 In PLP

A = Atrazine Ban A&D = Atrazme and Dual BanB = Banvel Ban B&D = Banvel and Dual BanD= DwJ Ban A, B&D = Atraine, Banvel, and Dual BanA&B = Atrazme and Banvel Ban

Figure 3, Relative Changes in PLP and per Acre Consumer Plus Producer Losses in Benefits for ThreeHypothctlcal Soybean 1Icrblcide Cancellations in North Carol ins.

9

8

171

0

5-

4 -

3

21

1’

lJ‘1

Weed C9n9i~

❑ Low

II❑ Mdium

❑ HiQh

-1i

Dud Ban bsso ml DA md Liwm Ban

-2 ‘7.12 19.14 45.26

Percsnl Reducllons In PLP

J Agr and Applied Econ Julti, 1995 295

improvement, and the effects of a mulhplccancellation are different from the summation of theeffects of independent cancellations, Therefore, ifthere are several herbicides under consideration forcancellation, effects of both single and multlplecancellations can be studied as dlustrated m thisstudy. Weed density has very strong affects onbenefit losses, but agricultural output demand andsupply elasticities do not in the Southeast. To

Improve the precision of welfare loss estimates inagricultural commodity markets due to herbicidecancellations, it is necessary to study the yield andcost effects m each state because of theheterogeneity of weed conditions across states. Thetrddeoff methodology developed and applied to theSoutheast could bc applied to other regions andcrops.

References

Aller, L., T. Bennet, J, H, Lehr and R, Petty. “DRASTIC: A System To Evaluate the Pollutlon Potential

of Hydrogeologic Setting by Pesticide.” pp. 141-158 in Evu/uutlotJ of Pes~icides in Groundwater(eds. W. Y. Garner, R. C. IIoncycutt, and H. N. Nigg), American Chermcal Society, Washington,D,C., 1986,

Batie, S,, W, Cox, and P, Diebel, Managing Agricheinical Ckntarninution qf Ground Water: StateStrategies, Capital Resources Pohcy Studies, Center for Policy Research, National Governors’Association, Washington, D.C, State Policy Report 1989.

Bouzaher, A,, P. G,, Lakshminarayan, A, Carriquiry, P, Gassman, J, F. Shogrtm, and R. Cabe. “EvaluatingRegional Ground- and Surface-Water Quality: An Apphcatlon of Mctamodelillg.” Paper presented atAAEA Annual Meetings. Baltimore, MD, August 9-12, 1992,

Carlson, G. A, and M, E, Wetzstem. “Pesticides and Pest Management,” pp. 268-318 in Agricultural andEnvironmental Resource Economics (eds,, G. A, Carl son, D, Zilbcrman and J, A. Miranowski), Oxford

University Press: Ncw York, 1993.

Cheng, H. H. and W, C, Koskinen, “Proccsscs and Factors Affecting Transport of Pesticides to GroundWater.” pp. 2-13 in Evaiuatlo~/ @ Pesticides in Groundwazer (eds., W. Y. Garner, R. C. I{oneycuttand H. N. Nigg), American Chemical Society, Washington, D.C., 1986.

Coble, D. H. and D. A, Mortensen. “The Threshold Concept and Its Application to Weed Science.” WeedTechnology. 6(1992): 191-5,

Danielson, L. E,, G, A, Carlson, S. Liu, J. B. Weber, R. L, Warren. Ground Water Contamination andCosts of Pesticide Restrictions in the Sautheus~ern Caaslal Plain. Water Resources Research Institute,North Carolina State University, Raleigh, NC, November, 1993.

Ferguson, W. L., L, J, Moffitt, and R, M. Davw “Short-Run Welfare Implication of Restricting FungicideUse in Vegetable Production.” J, of Agribusiness, 8(1992): 41-50.

Gardiner, W, H., V. O. Roningen, and K, Liu. Elasticities in the Trade I.iheralization Database. EconomicResearch Service, U,S, Department of Agriculture. Washington, DC, 1989,

Gardner, B. L. The Econonlics of Agrlculturol Policies, Macmdlan Publishing Corporation: New York.1987.

Gianessi, L, P., R. J, Kopp and C, A. Puffer, “Regulating Pcsticlde Use: Social Costs, Policy Targeting andEconomic Incentives,” Resources for the Future Discussion Paper QE89-21. Washington, D.C., 1989.

296 Liu, Carlson and Hoag: Trade -OflA nulws of Herbicide Withdrawals on Agricultural Production

Gianessi, L. P., R. J. Kopp, P. Kuch, C. Puffer, and R. Torla, “Welfare Implications of Restricted TriazineHerbicide Use in the Chesapeake Bay Region”. Marine Res. Econ, 5(1988): 243-58.

Griliches, Z, “Hybrid Corn: An Exploration in the Economics of Technical Change.” Economeo”ica.25(1957): 501-22.

Gustafson, D. “Groundwater Ubiquity Score: A Simple Method for Assessing Pesticide Leachability.”Environ. Toxicol, and Chem, 8(1989): 339-57,

Harper, C, R. and D, Zilberman, “Pest Externalities from Agricultural Inputs.” Amer. J. Agr. Econ.71(1989): 692-702.

Helling, C, S. and T, J. Gish. “Soil Characteristics Affecting Pesticide Movement into Ground Water.”pp. 14-38 in Evaluation qf Pesticides in Groundwaw (eds,, W. Y. Garner, R. C. Honeycutt and H.N. Nigg) American Chemical Society, Washington, D.C., 1986.

Hoag, D. L. and A. G. Hornsby. “Coupling Groundwater Contamination to Economic Returns whenApplying Farm Pesticides.” J. Environ. Qual~ty 21( 1992): 579-86.

Housenger, J. E. “Decision Making in the Risk-Benefit Assessment Process at EPA.” pp. 15-19 inProceedings oj’ the National Pesticide Impact Assessmen~ Workshop (cd,, Stephen J, Toth, Jr,),Rdleigh, NC, February, 1992.

Jennings, A, L. “Components of an EPA Pesticide Benefits Assessment.” pp. 61-64 in Proceedings qftheNational Pesticide Impact As,ressment Workshop (cd. Stephen J, Toth, Jr.), Raleigh, NC, February,1992.

Jury, W, A., D. Focht and W, J. Farmer. “Evaluation of Pesticide Groundwater Pollution Potential fromStandard Indices of Soil-Chemical Adsorption and Biodegradation.” J Environ. Quality 16(1987):422-8,

Just, R. E., D, L, Hueth and M. Phillips. “Benefits Estimates for MANE13.”Unpublished paper, Universityof Maryland, College Park, MD, 1987,

Kennedy, R,, R, Lowrey, A, Bernstein, F. Rueter, 11.Cole, and H. F. Smyth. “A Benefit Cost System forChemical Pesticides.” Environmental Protection Agency Report EPA-540/9-76-00 1, Strategic StudiesUmt, Washington, D,C., 1975,

Kopp, R. J. and A. J. Krupnick. “Agricultural Policy and the Benefits of Ozone Control,” Amer. J. Agr.Econ. 69(1987): 956-62.

Lichtenberg, E, and D, Zilberman. “The Welfare Economics of Regulation in Revenue-Supported Industries:The Case of Price Supports in U.S. Agriculture.” Amer. Econ. Rev. 76(1986): 1135-1141.

Liu, S. “Estimating the Effects of Herbicide Withdrawals on Agricultural Production and GroundwdterQuality.” Unpublished Ph,D Thesis, North Carolina State University, Raleigh, NC, 1993,

Lm, S, and G. A, Carlson, “Production Quality Information and Regulation of Pesticide Residues in Food,”Paper presented at AAEA annual Meetings, Baltimore, MD, August 9-12, 1992,

J Agr and Applied Econ, JuIv. 199S 297

Liu, S. and G, A. Carlson, “EX Ante Estimation of Pesticide Substitutes from a Pesticide Withdrawal.”Unpublished Paper, North Carolina State University, Raleigh, NC, 1994.

Modena, S, A., G. G. Wilkerson and H. D. Coble, “HERB Version 3,0 User’s Manual.” Research Report131, Department of Crop Science, North Carolina State University, Raleigh, NC, 1991.

National Agricultural Pesticide Impact Assessment Program (NAPIAP), 1993. The Importance of Pesticideand Other Pest Management Practices in U.S. Cotton Production, NAPlAP Report, Number l-CA-93,USDA, Washington, DC.

National Governors’s Association (NGA). “Agriculture and Water Quality.” Washington, D.C., 1989.

National Research Councd, Regulating Pes~icides in Food: the Delane.v Paradox, Washington, D.C.,National Academy Press, 1987.

Nielsen, Elizabeth G, and Linda K. Lee, 1987. The Magnitude and Costs of Ground Water Contaminationfrom Agricultural Chemicals: A National Perspective. Agricultural Economic Report No, 576,Resources and Technology Division, USDA, ERS, Washington D.C.. October,

North Carolina Agricultural Statistics, 1979-1990. North Carolina Department of Agriculture, Raleigh, NC.

North Carolina School of Agriculture and Life Sciences. 1992 North Carolina Agricultural ChemicalsManual. North Carolina State University, Raleigh, NC.

Osteen, C, D. “Commodity and Chemical Assessments.” pp. 659-60 in Proceedings of lhe NalionalPesticide Impact Asses.snzent Workshop (cd., Stephen J. Toth, Jr,), Ralclgh, NC, February, 1992.

Osteen, C. and F. Kuchler. “Pesticide Regulatory Decisions: Production Efficiency, Equity andInterdependence,” Agribu.~i}~ess 3(1987):307-22.

Smith, E, D,, R. D. Knutson, C, R, Taylor and J. B. Penson. “Impacts of Chemical Use Reduction on CropYields and Costs.” Agricultural and Food Policy Center, Department of Agricultural Economics,Texas A&M University, College Stwion, TX, 1990.

Stemeroff, M., J. Groenewegen and R. Krystynak. “The Benefits of 2,4-D: An Economic Assessment.”Canadian Farm Econ. 23( 1993]:3- 19.

Strobel, M. et al. “An Examination of the Spatial and Intertemporal Aspe&s of Basis Determination,”NCR- 134 Conference on Applied Price Analysis, Chicago, IL, 1992.

Taylor, C, R., Ronald D. Laecwcll and Hovav Talpaz. “Use of Extraneous Information with an EconometricModel to Evaluate Impacts of Pesticide Withdrawals,” W.J. Agr. Econ 4(1979):1-7.

U. S. Department of Agriculture (USDA), Economic Research Service, “Inputs, Situation and OutlookReport.” Washington, D,C., selected years.

U. S, Environmental Protection Agency. “Agricultural Chemicals in Ground Water: Proposed PesticideStrategy.” Washington, D,C,: Ofticc of Pesticides and TOXICSubstances, 1987.

U. S. Environmental Protection Agency. National Pesticide Wellwatcr Survey. Office of PesticidePrograms, U.S. Environmental Protection Agency, Washington, D,C., 1990.

298 l>lu, (’(U7SOHawl lIo[lg Trade-off A ndvsts o/ Iferlucde WI(hdruwal.s on Agricultural Production

Weber, J.El. “Potential Problems for NorthCarolinaG round Water from Herbicides: A Ranking Index”,Proceedings of Weed Sciencc Society, North Carohnd 8(1990): 30-46,

Weber, J, B. and R, L. Warren, “I lerbicide Bchavlor in Sods: A Pcsticidc/Soil Ranking System forMimrnizing Ground Water Convdminatlon,“ Procccdmgs of Northeastern Weed Science Society,47(1993):144-57.

Wilkerson, G. C’,, S. A, Modcna and 11. 1>,C’oblc. “HERB: Decision Model for Posterncrgencc WeedControl in Soybean.” Agron. J, 83(1991): 413-17.

IIndnotes

1. For feed grains, there are prlcc SUppOrtprograms that operate to hold producer prices to about 20 percentabove world equihbriurn prices (Gardner 1987). This distortion together with set-aside requirements canalter the changes in producer and consurncr surplus from an herbicide cancellation, Explicit accounting forthe mdepcndent impacts of price supports for corn arc beyond the scope of’this study. See Lichtenbcrg andZdberman (1986) for methods.

2. If a pcstlclde withdrawal results in a quallty Improvcrncn[ of the product---for example, pestlcidcreslduds in food are decreased with cancellation---then derruand for the product could increase and it ispossible that consumers could gain from a pcsticldc cancellation. Therefore, there coutd be consumer gainsin the agrlculturtal commodity market from a pesticide cancellation (Liu and Carlson 1992).

3. This is a geographical] y stratlficd, ]andorn sample of Pdrmers chosen to give accumte esurnatcs ofpesticide usc at the crop rcportmg district lCVC1of aggregation. For our four states and four years (1985-1988), on average there arc about 770 fwrncrs for corn and 994 Fdrmers for soybeans sampled per year,Since data from this survey is anmmlly sold to most pestlcidc companies, it Mjudged to be rehable survey.

4, Ybcan be estimated as

1.1 ,J=, 1=[ J?IY,, = ~ ~ Y,,$ = Y,).~ ~Sr:.[ 1-D,/l ()()]

!=1 1.1 (=I ,.1

5, Y, can be estimated as

,.[ ,[=, ,=1 ]51

Y,, = x x Y,”s; = Y,)”x Zs;”[1-D,/ I()()],.1 J.j ,=1 ,=1

6, The relative substitution index that reflects the ability of one herbicide (i) to substitute for another tobe canceled (b) is:

where, Ef~WLis the weed control efficacy of herbicide i used to treat weed specie w at application time t;COSL,WMthe herbicide material cost pcr acre trcauncnt for the same herbicide i, weed species w and timet; A,w are the acre treatments of hcrbicldc I and Al,,ware the treatment acres for hcrbicidc b against thesesame weeds at tlmc t; and A~is the sum of all treatments of hcrblclde b across all times and weeds for thiscrop. Treatment shares (SH,) for any combination of hcrbicld:s can then be formed by:

ISII, = [ND, i IND, ,,=1

J Agr and Applied Ilwn Julv, 199S 299

where k is the total available set of substitute herbicides (Liu, 1993),

7. Probably only medium and hgh weed densities make sense because only primary weed species areconsidered in the calculation. Potential damage from secondary weeds is ignored because of informationlimitation.

8. A positive value indicates that control cost will increase if a herbicide is canceled and a negative valueindicates that herbicide material cost will decrease.

9. The losses to society due to “disappearance” of hybrid were estimated by Griliches (195’7)with bothlong-run and short-run supplies of corn (horizontal and vertical supply curves). The ratio of two estimatesis 1.07. Therefore, the difference between these two extreme assumptions is vcly limited, The loss dueto a herbicide cancellation is approximately calculated by the case with a vertical supply curve.

10, The Southeast has a regional market for corn and soybeans which is influenced by many regionalfactors, and is for convenience in this study assumed to be separate from that of the remainder of the UnitedStates (Strobel ct al., 1992),

11. A regional cancellation might have mmor affects on national commodity prices since these are smallshares of national and international markets, lIowever, for regional cancellations in the regional corn marketthere can be larger affects.

12. Linear demand and supply functions are not constant elasticity dcmmd and supply functions. However,if the cha~lges in price and quantity are very limited, the clastlcitics of demand and supply can be consideredapproximately constant within a limited range for the linear functions.

13. With these assumptions and using difference to replace differential, the following two equations canbe derived

[ I’llQ-Q, p,)K@(,-QJ = ~ ~ ~

o

and

[ 11P,)-Q3 ~,,P,-PO= _“

TT’These two equations are derived based on the geometry in Figure 1 and the definitions of the demand andsupply ehsticities, Solving these two equations and equations 3 and 4 simultaneously, four unknowns, K,M, P~ and Q?, can be expressed as functions of q, E, P(),Q(l,Y(),Yl, and Ac.

14, Supply and demand elasticities, base yields, weed types and herbicide use patterns are similar over the

study region, however, there are some differences, Sensitivity of A[CS + PSj results to weed density asshown below led us to present yield, material cost and benefit loss figures specific to North Carolina sincewe did not have starting weed density figures for the three other states.

15, For sensitivity analysis purposes, three elasticities of demand and supply are used in the analysis. Thedemand elasticities for corn used in calculation are –0,21, – 1,21, and -2.21, Three supply elasticities usedin analysis are 0,28, 0,48, and 0,68, Based on this analysis, elasticity effects are limited relative to totaleffects. The largest difference due 10 elasticities of demand and supply is only $0,16 per acre. (Detailedresults are avadable from authors.)

300 Liu, Curlson and [{oag: Trade -Oj’A~jah,sls of[ferbwlde Withdrawal.s ml Agricultural Productwn

16. Sepamte estimates by state only require starting herbicide use and substitution patterns. Therefore, statelevel estimates for the four separate states are presented,