Adverse Selection, Moral Hazard,Adverse Selection, Moral Hazard,

and Grower Compliance and Grower Compliance with Bt Corn Refugewith Bt Corn Refuge

Paul D. MitchellPaul D. MitchellAgricultural and Applied EconomicsAgricultural and Applied Economics

University of Wisconsin-MadisonUniversity of Wisconsin-Madisonandand

Terry HurleyTerry HurleyApplied Economics, University of MinnesotaApplied Economics, University of Minnesota

Seminar Presented to the Environmental Economics and Seminar Presented to the Environmental Economics and Natural Resources Group at Wageningen University, Natural Resources Group at Wageningen University,

November 16, 2006November 16, 2006

Published as Mitchell and Hurlel (2006) In Published as Mitchell and Hurlel (2006) In Economics and Regulation of Economics and Regulation of Agricultural Biotechnologies Agricultural Biotechnologies (Just, Alston and Zilberman, eds.), Kluwer (Just, Alston and Zilberman, eds.), Kluwer

Academic Publishers.Academic Publishers.

Overview of PresentationOverview of Presentation Motivate Bt Corn IRM Compliance ProblemMotivate Bt Corn IRM Compliance Problem Describe Model SetupDescribe Model Setup Develop principal-agent model to assess a Develop principal-agent model to assess a

fine program for Bt corn IRM compliancefine program for Bt corn IRM compliance Endogenize technology fee (price of Bt corEndogenize technology fee (price of Bt cor

n), fine and audit raten), fine and audit rate Growers have private information for their Growers have private information for their

Willingness to Pay (adverse selection) and Willingness to Pay (adverse selection) and Compliance Effort (moral hazard)Compliance Effort (moral hazard)

Present empirical resultsPresent empirical results

BackgroundBackground Bt Corn (maize) controls two major Bt Corn (maize) controls two major

pests: European corn borer (ECB) and pests: European corn borer (ECB) and corn rootworm (CRW)corn rootworm (CRW)

Each estimated to cost about $1 billion anEach estimated to cost about $1 billion annually in yield losses and control costsnually in yield losses and control costs

Bt corn: Maize engineered to contain DNA Bt corn: Maize engineered to contain DNA from bacterium from bacterium Bacillus thuringiensisBacillus thuringiensis (Bt) (Bt)

Plant tissues express Bt toxin so that pests Plant tissues express Bt toxin so that pests killed when eat/damage plantskilled when eat/damage plants

BackgroundBackground Bt corn for ECB available since 1995Bt corn for ECB available since 1995 Several “events” registered (MON 810, Several “events” registered (MON 810,

Bt 11, DBT 418, Event 176, CBH 351)Bt 11, DBT 418, Event 176, CBH 351) YieldGard Corn Borer (MON 810/Bt 11) moYieldGard Corn Borer (MON 810/Bt 11) mo

st popular Bt corn availablest popular Bt corn available YieldGard Rootworm (MON 863) for corn rYieldGard Rootworm (MON 863) for corn r

ootworm available in 2003ootworm available in 2003 Both YieldGard events available alone or Both YieldGard events available alone or

stacked in the same hybrid (+ RR/HT)stacked in the same hybrid (+ RR/HT) More companies releasing CRW Bt cornMore companies releasing CRW Bt corn

USDA-NASS Adoption DataUSDA-NASS Adoption DataIR 2005 IR 2006 Stack 2005 Stack 2006

IL 25 24 5 19

IN 11 13 4 12

IA 35 32 11 18

KS 23 23 10 12

MI 15 16 5 10

MN 33 28 11 16

MO 37 38 6 7

NE 39 37 12 15

ND 21 29 15 20

OH 9 8 2 5

SD 30 20 22 34

TX 21 27 9 13

WI 22 22 6 10

US 26 25 9 15

Bt Maize in EUBt Maize in EU Spain: 60,000 ha (12%) in 2004, dropped Spain: 60,000 ha (12%) in 2004, dropped

in 2005 (weather), up again in 2006in 2005 (weather), up again in 2006 France: 1000 ha in 2005, 5000 ha 2006; France: 1000 ha in 2005, 5000 ha 2006;

likely higher due to undocumented likely higher due to undocumented imports from Spain (SW France has ECB imports from Spain (SW France has ECB pressure)pressure)

Portugal: 750 ha in 2005Portugal: 750 ha in 2005 Czech Republic: 1500 ha in 2006Czech Republic: 1500 ha in 2006 Germany: trials since 2004, 1000 ha in Germany: trials since 2004, 1000 ha in

2006 for commercial use2006 for commercial use

Resistance ProblemResistance Problem Pests can develop resistance to Bt, Pests can develop resistance to Bt,

especially the more regularly and exespecially the more regularly and exclusively it is usedclusively it is used

Documented resistance to Bt in fielDocumented resistance to Bt in field and lab for different speciesd and lab for different species

No documented cases of field resistNo documented cases of field resistance to Bt corn or Bt cottonance to Bt corn or Bt cotton

ResistanceResistance RegulationRegulation Bt crops registered by EPA under FIFRA as Bt crops registered by EPA under FIFRA as

Plant Incorporated Protectants (PIP’s)Plant Incorporated Protectants (PIP’s) Under FIFRA, EPA can only register PIP’s, Under FIFRA, EPA can only register PIP’s,

not enforce regulations, but can impose renot enforce regulations, but can impose registration requirements on registrantsgistration requirements on registrants

EPA requires Insect Resistance ManagemeEPA requires Insect Resistance Management (IRM) plan for registration and requires nt (IRM) plan for registration and requires registrants to enforce IRM planregistrants to enforce IRM plan

EPA has not made similar requirements foEPA has not made similar requirements for other pest control methodsr other pest control methods

High Dose/Refuge IRM High Dose/Refuge IRM StrategyStrategy

Bt corn must express a high dose (25 X LC99)Bt corn must express a high dose (25 X LC99) Plant non-Bt corn refuge to generate non-Plant non-Bt corn refuge to generate non-

exposed adults to mate with the few resistant exposed adults to mate with the few resistant adults from the Bt corn (500:1), and so to adults from the Bt corn (500:1), and so to dilute the resistance gene in next generationdilute the resistance gene in next generation

RefugeRefuge size requirementsize requirement 20% for most of the Corn Belt 20% for most of the Corn Belt 40% if spraying for other pests40% if spraying for other pests 50% in southern corn-cotton counties50% in southern corn-cotton counties

RefugeRefuge proximity requirement proximity requirement within ½ milewithin ½ mile

Compliance Problem for IRMCompliance Problem for IRM

Farmers have little incentive to voluntarily Farmers have little incentive to voluntarily manage resistance by planting refugemanage resistance by planting refuge Refuge decreases profit in short runRefuge decreases profit in short run Pest susceptibility treated as a common Pest susceptibility treated as a common

property resource: ”Tragedy of the Commons”property resource: ”Tragedy of the Commons”

Compliance surveys find a variety of Compliance surveys find a variety of compliance levels among farmerscompliance levels among farmers

Annual Industry (ABSTC) Survey of Annual Industry (ABSTC) Survey of farmers with more than 200 acres cornfarmers with more than 200 acres corn In 2003, 92% met size requirement and In 2003, 92% met size requirement and

93% met distance requirement93% met distance requirement In 2000, 87% met size requirement and In 2000, 87% met size requirement and

82% met distance requirement82% met distance requirement Not report % satisfying both requirementsNot report % satisfying both requirements

CSPI, via Freedom of Information Act, CSPI, via Freedom of Information Act, obtained USDA 2002 crop acreage obtained USDA 2002 crop acreage datadata Farmers report Acres of corn and Acres BtFarmers report Acres of corn and Acres Bt Only able to evaluate Only able to evaluate sizesize requirement requirement Found substantially less complianceFound substantially less compliance

% of Farms Planting Bt Corn % of Bt Corn Acres

StateNon-

complying w/ 100% BtNon-

complying w/ 100% Bt

IL 14% 9% 15% 7%

IN 11% 10% 13% 12%

IA 18% 13% 24% 14%

KS 33% 24% 34% 20%

MI 46% 38% 47% 33%

MN 18% 13% 25% 15%

NE 22% 14% 27% 14%

OH 38% 37% 56% 54%

SD 33% 21% 35% 19%

WI 18% 16% 28% 21%

10 Sts 21% 15% 26% 15%

Small farmers more likely violate size requirement

Large farmers have most non-complying acres (and more likely violate distance requirement)

% within % all Bt farms

Acres Bt Farms

Non-comply

100% Bt

Non-comply

100% Bt

Non-comply

100% Bt

≥ 200 56,150 7,090 3,720 13% 7% 8% 4%

< 200 37,380 12,620 10,300 34% 28% 13% 11%

All 93,530 19,710 14,020 21% 15% 21% 15%

% within % all Bt acres

Acres Bt Acres

Non-comply

100% Bt

Non-comply

100% Bt

Non-comply

100% Bt

≥ 200 14.15 3.36 1.81 24% 13% 21% 11%

< 200 2.01 0.85 0.66 42% 33% 5% 4%

All 16.15 4.21 2.47 26% 15% 26% 15%

EPA (via FIFRA) required registrants to EPA (via FIFRA) required registrants to develop more aggressive compliance develop more aggressive compliance program in Dec. 2002, months before CSPI program in Dec. 2002, months before CSPI published its analysispublished its analysis

Compliance Assurance ProgramCompliance Assurance Program Randomly audit farmers for complianceRandomly audit farmers for compliance Non-complying farmers receive extra educaNon-complying farmers receive extra educa

tion and are guaranteed a compliance audit tion and are guaranteed a compliance audit the next yearthe next year

If farmer caught non-complying twice, bannIf farmer caught non-complying twice, banned from buying Bt ed from buying Bt

Is the punishment is enforceable? Is the punishment is enforceable? Registrants have licensed many seed compaRegistrants have licensed many seed compa

niesnies Is the ban is an effective deterrent?Is the ban is an effective deterrent? We examine a Fine Program as alternativeWe examine a Fine Program as alternative

Pertinence to EUPertinence to EU

Reducing genetic contamination of Reducing genetic contamination of non-GMO crops the issue in EU, not IRMnon-GMO crops the issue in EU, not IRM

Coexistence requirements include a Coexistence requirements include a buffer strip planted around Bt maizebuffer strip planted around Bt maize

This buffer also serves as a refuge to This buffer also serves as a refuge to slow the development of resistanceslow the development of resistance

What are the incentive issues for EU What are the incentive issues for EU growers of Bt maize and coexistence growers of Bt maize and coexistence requirements?requirements?

Fine ProgramFine Program OverviewOverview

Growers register when they buy Bt corn, just as Growers register when they buy Bt corn, just as with current Grower Agreementswith current Grower Agreements

Growers audited with probability Growers audited with probability If audited and not complying (cheating on If audited and not complying (cheating on

refuge requirement), grower pays the fine refuge requirement), grower pays the fine FF ($/ac)($/ac)

A non-complying grower pays the fine A non-complying grower pays the fine FF with with probability probability and nothing withand nothing with probabilityprobability (1(1 –– ))

Company chooses Bt corn technology fee Company chooses Bt corn technology fee T T (price), the audit probability (price), the audit probability , and the fine , and the fine FF

Timeline of EventsTimeline of Events

1.1. Company announces Bt corn price Company announces Bt corn price TT, audit probability , audit probability , and fine , and fine FF

2.2. Grower decides whether to buy Bt Grower decides whether to buy Bt corn or conventional corncorn or conventional corn

3.3. Company audits those who buy Bt Company audits those who buy Bt corn and imposes fines on growers corn and imposes fines on growers not complying with refuge not complying with refuge requirementrequirement

Model OverviewModel Overview1.1. Define grower returns and then Define grower returns and then

formulate participation and incentive formulate participation and incentive compatibility constraintscompatibility constraints

2.2. Reformulate and describe constraints Reformulate and describe constraints (Proposition 1 and 2 and Corollaries)(Proposition 1 and 2 and Corollaries)

3.3. Formulate and describe company’s Formulate and describe company’s (principal’s) optimization problem(principal’s) optimization problem

4.4. Note special case (Proposition 3)Note special case (Proposition 3)

5.5. Empirical AnalysisEmpirical Analysis

Model SetupModel Setup

Grower returns ($/ac) for conventional corGrower returns ($/ac) for conventional cornn

cvcv = = pypy – – KK

pp non-random price of cornnon-random price of cornyy random potential (pest free) yirandom potential (pest free) yi

eldeldKK non-random production costnon-random production cost

Model SetupModel Setup

Grower returns for planting all Bt cornGrower returns for planting all Bt corn

btbt = = pypy (1 + (1 + ) – ) – K K

random yield gain for Bt cornrandom yield gain for Bt corn

Grower returns with the Tech Fee Grower returns with the Tech Fee TT ($/ac)($/ac)

= = btbt – – T T

Model SetupModel Setup

Returns for a complying grower who plantReturns for a complying grower who plants required refuge s required refuge rr

cpcp = = rrcvcv + (1 – + (1 – rr))btbt

with the Tech Fee with the Tech Fee TT= = cpcp – – (1 – (1 – rr))TT

Model SetupModel SetupGrower’s maximum per acre willingness to Grower’s maximum per acre willingness to

pay pay WW ($/ac) for Bt corn is private/hidden in ($/ac) for Bt corn is private/hidden informationformation

E[E[UU ((cpcp –– WW )] = E[)] = E[UU ((cvcv)])]Hidden information concerning Hidden information concerning WW creates creates

adverse selection when choosing adverse selection when choosing TTParticipation ConstraintParticipation Constraint

E[E[UU ((cpcp –– WW )] )] ≥ ≥ E[E[UU ((cvcv)])]WW ≥ ≥ (1 – (1 – rr))TT

Buy Bt corn if WTP ≥ priceBuy Bt corn if WTP ≥ price

[ ( )] [ ( )]comply cheatE U E U

[ ( (1 ) )] (1 ) [ ( )]

[ ( )]

cp r Bt

Bt

E U T E U T

E U T F

Hidden information concerning Hidden information concerning compliancecompliance

effort (% refuge) creates moral effort (% refuge) creates moral hazardhazard

Company uses the fine program to Company uses the fine program to solvesolve

Incentive Compatibility Incentive Compatibility ConstraintConstraint

Model SetupModel Setup

Propositions 1 and 2Propositions 1 and 2

If utility is continuous and strictly increases in income, If utility is continuous and strictly increases in income, The ICC can be expressed as The ICC can be expressed as WW ≥ ≥ ZZ ((,,FF,,TT ), ),

where where ZZ (() is a function depending on grower utility) is a function depending on grower utility Distribution Distribution GG ((WW ) “common knowledge”) “common knowledge” Using Using GG ((WW ), the cdf of ), the cdf of WW, the constraints can be exp, the constraints can be exp

ressed as probabilitiesressed as probabilities Probability of participation: Probability of participation: = 1 – = 1 – GG ((1 – ((1 – rr))TT )) Probability of compliance: Probability of compliance: = 1 – = 1 – GG ((ZZ ((,,FF,,TT ))))

Propositions 1 and 2Propositions 1 and 2 Participation (Rationality) constraintParticipation (Rationality) constraint

WW ≥ ≥ (1 – (1 – rr))T T → → = 1 – = 1 – GG ((1 – ((1 – rr))TT )) Incentive compatibility constraintIncentive compatibility constraint

WW ≥ ≥ ZZ ((,,FF,,TT ) ) → → = 1 – = 1 – GG ((ZZ ((,,FF,,TT )))) Both put lower bound on Both put lower bound on WW and which one and which one

binds implies different grower behaviorbinds implies different grower behavior

(1 – (1 – rr))T T < < ZZ ((,,FF,,TT ))→ → (some buyers cheat) (some buyers cheat)(1 – (1 – rr))T T ≥≥ ZZ ((,,FF,,TT ))→ v→ v≥≥all buyers comply)all buyers comply)

(1 – )T

Buy Bt corn (probability = )

Z(, F, T)

Comply (probability = )

Grower willingness to pay W

Buy Bt corn, do not comply (probability = – )

Do not comply

Do not buy Bt corn

(1 – )T

Buy Bt corn (probability = )

Z(, F, T)

Comply (probability = )

Grower willingness to pay W

Do not comply

Do not buy Bt corn

Corollary 0Corollary 0

Probability of participation Probability of participation = 1 – = 1 – GG ((1 – ((1 – rr))TT ))

Decreases in technology fee Decreases in technology fee TT (i.e., downward sloping demand curve)(i.e., downward sloping demand curve)

Independent of fine Independent of fine FF and audit probabil and audit probability ity

Proposition 2/Corollary 1Proposition 2/Corollary 1Probability of compliance Probability of compliance = 1 – = 1 – GG ((ZZ((,,FF,,TT ))))

Non-decreasing in the audit probability Non-decreasing in the audit probability Non-decreasing in the fine Non-decreasing in the fine FF Non-decreasing (non-increasing) in technolNon-decreasing (non-increasing) in technol

ogy fee ogy fee TT if if

Note: risk neutral or CARA utility Note: risk neutral or CARA utility → → dd/d/dTT ≥ 0 ≥ 0

(1 ) [ '( ) [ '( )]

(1 ) [ '( (1 ) ( )] ( )0Bt Bt

r Bt r r

E U T E U T F

E U T py

Company ProblemCompany Problem Choose audit probability Choose audit probability , fine , fine FF, and te, and te

chnology fee chnology fee TT to maximize expected net to maximize expected net revenue, subject to the participation and revenue, subject to the participation and incentive compatibility constraintsincentive compatibility constraints

Company endogenizes both purchase anCompany endogenizes both purchase and compliance probabilitiesd compliance probabilities Pr[buy] = Pr[buy] = = 1 – = 1 – GG ((1 – ((1 – rr))TT )) Pr[comply] = Pr[comply] = = 1 – = 1 – GG ((ZZ ((,,FF,,TT ))))

Which binds? (1 – Which binds? (1 – rr))T T <(>) <(>) ZZ ((,,FF,,T T )) Creates two different functionsCreates two different functions

(1 – )T

Buy Bt corn (probability = )

Z(, F, T)

Comply (probability = )

Grower willingness to pay W

Buy Bt corn, do not comply (probability = – )

Do not comply

Do not buy Bt corn

(1 – )T

Buy Bt corn (probability = )

Z(, F, T)

Comply (probability = )

Grower willingness to pay W

Do not comply

Do not buy Bt corn

Company ProblemCompany Problem Company returns are sum of net revenue frCompany returns are sum of net revenue fr

om sales and fine collectionom sales and fine collection If If (1 – (1 – rr))T T < < ZZ ((,,FF,,T T ), ), > >

Expected net sales revenue (Expected net sales revenue ( – – rr)()(TT – – cc)) Expected fine revenue (Expected fine revenue (– – ))FF Expected monitoring cost Expected monitoring cost kk(())

, ,max ( )( ) ( ) ( )r

F Tv T c v F k

Company ProblemCompany Problem If (1 – If (1 – rr))T T > > ZZ ((,,FF,,TT ), ), > > All growers who buy comply, so All growers who buy comply, so = =

Expected net sales revenue Expected net sales revenue (1 – (1 – rr)()(TT – – cc )) Expected fine revenue = 0Expected fine revenue = 0 Still have to monitor and threaten fine for iStill have to monitor and threaten fine for i

ncentive compatibility, but ncentive compatibility, but kk ’(’() defines ) defines * then set * then set FF so that (1 – so that (1 – rr))T T > > ZZ ((,,FF,,TT ) hol) holdsds

, ,max (1 )( ) ( )r

F TT c k

OptimizationOptimization

The company/principal maximizes the The company/principal maximizes the upper envelope of the two functionsupper envelope of the two functions

Various relationships possible Various relationships possible depending on the parameters depending on the parameters cc, , GG((∙∙), ), , , UU((∙∙), ), rr, etc., etc.

Both functions concave, so separately Both functions concave, so separately maximize each and compare solutionsmaximize each and compare solutions

0.00

0.50

1.00

1.50

2.00

2.50

0 5 10 15 20 25 30

Technlogy Fee T

Co

mp

any

Net

Rev

enu

e V

Complete Compliance >

Non-compliance Permitted <

Proposition 3Proposition 3

If growers are risk neutral, the If growers are risk neutral, the optimization problem separatesoptimization problem separates

kk) ) = = 0 defines the optimal audit rate 0 defines the optimal audit rate , , regardless of regardless of GG ((WW ), the distribution ), the distribution of grower willingness to payof grower willingness to pay

The principal’s objective need only be The principal’s objective need only be optimized with respect to optimized with respect to FF and and TT, , treating treating as a parameter defined by as a parameter defined by kk) = 0 ) = 0

Model SummaryModel Summary

Conceptually, a solution (Conceptually, a solution (,,FF,,TT) exists for ) exists for the company’s optimization problemthe company’s optimization problem

Analytically tractable solutions exist with Analytically tractable solutions exist with uniform uniform GG ((WW ) and risk neutral grower) and risk neutral grower

Special cases (risk neutral or CARA utility) Special cases (risk neutral or CARA utility) imply simpler optimization problemimply simpler optimization problem

More realistic distributions for More realistic distributions for WW require require numerical methods to find solutionnumerical methods to find solution

Empirical AnalysisEmpirical Analysis

Parameterize the model for Rock County Parameterize the model for Rock County Wisconsin and corn rootworm Bt cornWisconsin and corn rootworm Bt corn

Prime Wisconsin corn-soybean area with Prime Wisconsin corn-soybean area with new invasion of rotation resistant new invasion of rotation resistant western corn rootworm and new Bt corn western corn rootworm and new Bt corn availableavailable

Grower survey data (Langrock and Grower survey data (Langrock and Hurley) for similar Minnesota location Hurley) for similar Minnesota location

Grower ReturnsGrower Returns

cvcv = = pypy – – K K pp = $2.25/bu, = $2.25/bu, KK = $200/ac = $200/ac yy random: beta distribution, mean = 150 random: beta distribution, mean = 150

bu/ac, CV = 30%, min = 0, max = 240bu/ac, CV = 30%, min = 0, max = 240

btbt = = pypy (1 + (1 + ) – ) – KK random: beta distribution, mean 3%, random: beta distribution, mean 3%,

5%, 7%, CV = 100%, min 0, max 15%, 7%, CV = 100%, min 0, max 1 yy and and independent independent

Grower PreferencesGrower Preferences

CARA Utility: CARA Utility: UU (() = 1 – exp(–) = 1 – exp(–RR)) RR = 0.005174 so risk premium 25% of = 0.005174 so risk premium 25% of

E[E[]] Effort = proportion of refuge plantedEffort = proportion of refuge planted

Comply Comply = = rr or Not Comply or Not Comply = 0 = 0 Grower willingness to pay pdf Grower willingness to pay pdf GG ((WW ))

Based on survey of Langrock and HurleyBased on survey of Langrock and Hurley GG ((WW ) = lognormal, mean = $8.59/ac, ) = lognormal, mean = $8.59/ac,

standard deviation = $20.60/acstandard deviation = $20.60/ac

Company ReturnsCompany Returns Marginal cost of Bt corn vs conventional Marginal cost of Bt corn vs conventional

corn corn cc = 0, 3, 6 = 0, 3, 6 Define Define kk (() so ) so kk ’(’() = 0 defines ) = 0 defines

reasonable reasonable *, since this defines optimal *, since this defines optimal * if * if > > or if risk neutral grower or if risk neutral grower

Calibrate with hybrid seed corn Calibrate with hybrid seed corn certification certification

1.1. kk (() so ) so kk ’(’() = 0 defines ) = 0 defines = 0.04 = 0.042.2. Average cost = $1.20/ac at Average cost = $1.20/ac at = 0.04 = 0.043.3. 25% cost increase if audit rate doubles25% cost increase if audit rate doubles kk (() = 1.5 – 15) = 1.5 – 15 + 187.5 + 187.522

E[E[]] cc TT FF cheatcheat VV

3%3% 00 8.058.05 ---- ---- 31.5%31.5% 24.6%24.6% 22.0%22.0% 2.142.14

33 19.7019.70 ---- ---- 12.9%12.9% 12.9%12.9% 0.0%0.0% 1.731.73

5%5% 00 9.609.60 ---- ---- 27.1%27.1% 13.7%13.7% 49.6%49.6% 2.342.34

33 14.0614.06 ---- ---- 18.8%18.8% 14.7%14.7% 22.0%22.0% 1.751.75

00 10.5310.53 ---- ---- 24.9%24.9% 8.7%8.7% 65.0%65.0% 2.442.44

7%7% 33 15.5815.58 ---- ---- 16.9%16.9% 9.3%9.3% 44.9%44.9% 1.891.89

66 19.7319.73 ---- ---- 12.9%12.9% 9.8%9.8% 24.2%24.2% 1.501.50

3%3% 00 12.8212.82 ---- ---- 20.6%20.6% 20.6%20.6% 0.0%0.0% 2.122.12

33 19.7019.70 ---- ---- 12.9%12.9% 12.9%12.9% 0.0%0.0% 1.731.73

5%5% 00 8.928.92 ---- ---- 28.9%28.9% 18.0%18.0% 37.8%37.8% 2.262.26

33 19.7019.70 ---- ---- 12.9%12.9% 12.9%12.9% 0.0%0.0% 1.731.73

00 9.879.87 ---- ---- 26.4%26.4% 12.1%12.1% 54.1%54.1% 2.372.37

7%7% 33 14.4914.49 ---- ---- 18.2%18.2% 13.0%13.0% 28.6%28.6% 1.791.79

66 26.0226.02 ---- ---- 9.2%9.2% 9.2%9.2% 0.0%0.0% 1.471.47

Risk NeutralRisk Averse

No Compliance No Compliance ProgramProgram

Results with No ProgramResults with No Program Current tech fee about $20/ac, which Current tech fee about $20/ac, which

consistent for consistent for T T * with * with cc = $3-$6/ac = $3-$6/ac ““Peak Switching” occursPeak Switching” occurs

In some cases, no cheating occursIn some cases, no cheating occurs In some cases, cheating ranges 22%-65%In some cases, cheating ranges 22%-65%

Risk aversion can cause peak switching, Risk aversion can cause peak switching, (increases (increases T T *), else decreases *), else decreases T T **

Risk aversion reduces cheating and Risk aversion reduces cheating and company returnscompany returns

Marginal cost Marginal cost cc and expected loss E[ and expected loss E[] ] increase increase T T **

E[E[]] cc TT FF cheatcheat VV

3%3% 00 16.3616.36 4.0%4.0% ---- 16.0%16.0% 16.0%16.0% 0.0%0.0% 1.901.90

33 22.9122.91 4.0%4.0% ---- 10.8%10.8% 10.8%10.8% 0.0%0.0% 1.591.59

5%5% 00 16.3616.36 4.0%4.0% ---- 16.0%16.0% 16.0%16.0% 0.0%0.0% 1.901.90

33 22.9122.91 4.0%4.0% ---- 10.8%10.8% 10.8%10.8% 0.0%0.0% 1.591.59

00 16.3616.36 4.0%4.0% ---- 16.0%16.0% 16.0%16.0% 0.0%0.0% 1.901.90

7%7% 33 22.9122.91 4.0%4.0% ---- 10.8%10.8% 10.8%10.8% 0.0%0.0% 1.591.59

66 29.0429.04 4.0%4.0% ---- 7.9%7.9% 7.9%7.9% 0.0%0.0% 1.361.36

3%3% 00 16.3616.36 4.0%4.0% ---- 16.0%16.0% 16.0%16.0% 0.0%0.0% 1.901.90

33 22.9122.91 4.0%4.0% ---- 10.8%10.8% 10.8%10.8% 0.0%0.0% 1.591.59

5%5% 00 5.855.85 4.8%4.8% 62.1862.18 40.1%40.1% 23.8%23.8% 40.6%40.6% 2.072.07

33 22.9122.91 4.0%4.0% ---- 10.8%10.8% 10.8%10.8% 0.0%0.0% 1.591.59

00 8.798.79 4.7%4.7% 51.9151.91 29.3%29.3% 14.8%14.8% 49.6%49.6% 2.312.31

7%7% 33 10.0110.01 5.0%5.0% 67.1467.14 26.1%26.1% 16.6%16.6% 36.3%36.3% 1.601.60

66 29.0429.04 4.0%4.0% ---- 7.9%7.9% 7.9%7.9% 0.0%0.0% 1.361.36

Risk Neutral

Risk Averse

Fine revenue capped at monitoring Fine revenue capped at monitoring costcost

Results with fine revenue Results with fine revenue capcap

Optimal audit rate Optimal audit rate * gravitates to 4% to * gravitates to 4% to minimize monitoring costsminimize monitoring costs

Cap generally causes a “peak shift” to Cap generally causes a “peak shift” to complete compliance with higher tech fees complete compliance with higher tech fees and lower participationand lower participation

Eliminates non-compliance, but reduces Eliminates non-compliance, but reduces company revenue and grower use of company revenue and grower use of technologytechnology

Exceptions: low tech fee (high Exceptions: low tech fee (high ) and fine ) and fine F F * * of $50-$70/ac, with lots of cheating (36%-50%)of $50-$70/ac, with lots of cheating (36%-50%)

E[E[]] cc TT FF cheatcheat VV

3%3% 00 16.3616.36 4.0%4.0% 00 16.0%16.0% 16.0%16.0% 0.0%0.0% 1.901.90

33 22.9122.91 4.0%4.0% 00 10.8%10.8% 10.8%10.8% 0.0%0.0% 1.591.59

5%5% 00 0.100.10 4.0%4.0% 293.84293.84 99.6%99.6% 37.3%37.3% 62.6%62.6% 6.226.22

33 0.140.14 4.0%4.0% 296.43296.43 99.3%99.3% 37.9%37.9% 61.8%61.8% 3.463.46

00 0.070.07 4.0%4.0% 431.05431.05 99.8%99.8% 31.4%31.4% 68.6%68.6% 10.6710.67

7%7% 33 0.070.07 4.0%4.0% 433.74433.74 99.8%99.8% 31.8%31.8% 68.1%68.1% 9.369.36

66 0.100.10 4.0%4.0% 436.10436.10 99.6%99.6% 32.2%32.2% 67.7%67.7% 5.075.07

3%3% 00 16.3616.36 4.0%4.0% 00 16.0%16.0% 16.0%16.0% 0.0%0.0% 1.901.90

33 22.9122.91 4.0%4.0% 00 10.8%10.8% 10.8%10.8% 0.0%0.0% 1.591.59

5%5% 00 0.250.25 7.2%7.2% 93.9493.94 97.9%97.9% 38.6%38.6% 60.6%60.6% 2.882.88

33 22.9122.91 4.0%4.0% 00 10.8%10.8% 10.8%10.8% 0.0%0.0% 1.591.59

00 0.140.14 8.7%8.7% 109.41109.41 99.3%99.3% 32.6%32.6% 67.2%67.2% 4.874.87

7%7% 33 0.300.30 8.6%8.6% 110.69110.69 97.1%97.1% 32.9%32.9% 66.1%66.1% 2.122.12

66 29.0429.04 4.0%4.0% 00 7.9%7.9% 7.9%7.9% 0.0%0.0% 1.361.36

Risk Neutral

Risk Averse

No fine revenue No fine revenue capcap

Results with no fine revenue Results with no fine revenue capcap

Two regimes are optimal depending on Two regimes are optimal depending on parametersparameters

Higher tech fee with complete complianceHigher tech fee with complete compliance Eliminates non-compliance, but reduces Eliminates non-compliance, but reduces

company revenue and grower access to Btcompany revenue and grower access to Bt Very low tech fee with cheating 60%-70%Very low tech fee with cheating 60%-70%

Give Bt away so 100% adoptionGive Bt away so 100% adoption Become fine collection company with much Become fine collection company with much

higher returns: high fines with lots of cheating higher returns: high fines with lots of cheating

Discussion/Summary of Discussion/Summary of EmpiricsEmpirics

Inspection-Fine program may work if Inspection-Fine program may work if Compliance Assurance Program not sufficientCompliance Assurance Program not sufficient

Must impose cap on company fine revenue, Must impose cap on company fine revenue, otherwise create perverse incentivesotherwise create perverse incentives

Current Compliance Assurance Program Current Compliance Assurance Program similar to capped fine revenue casesimilar to capped fine revenue case Company monitors, but collects no finesCompany monitors, but collects no fines Tech fee increased, 100% compliance, lower Tech fee increased, 100% compliance, lower

adoption/participationadoption/participation Before imposing Inspection-Fine, let’s see how Before imposing Inspection-Fine, let’s see how

Compliance Assurance Program performsCompliance Assurance Program performs

What’s Next?What’s Next? Welfare analysisWelfare analysis

Monopoly and IRM restrict supply, implying Monopoly and IRM restrict supply, implying welfare losswelfare loss

Cheating offsets these welfare lossesCheating offsets these welfare losses To identify social optimum, must determine To identify social optimum, must determine

social gain from preserving pest social gain from preserving pest susceptibilitysusceptibility

Unify grower willingness to pay and utilityUnify grower willingness to pay and utility Joint distribution Joint distribution GG ((WW,,RR ) implies ) implies UU ((||RR )) Estimate with same survey in manner akin to Estimate with same survey in manner akin to

Love and Bucola 1991; Saha et al. 1994Love and Bucola 1991; Saha et al. 1994