Forthcoming in A. Heyes, ed., Law and Economics of the Environment, Edward
Elgar Publishing Ltd, 416 pp. (July 2001)
Montréal
Novembre 2000
Série Scientifique
Scientific Series
Law versus Regulation:
A Political Economy Model of
Instrument Choice in
Environmental Policy
Marcel Boyer, Donatella Porrini
Le CIRANO est un organisme sans but lucratif constitué en vertu de la Loi des compagnies du Québec. Le
financement de son infrastructure et de ses activités de recherche provient des cotisations de ses
organisations-membres, d’une subvention d’infrastructure du ministère de la Recherche, de la Science et de la Technologie, de
même que des subventions et mandats obtenus par ses équipes de recherche.
CIRANO is a private non-profit organization incorporated under the Québec Companies Act. Its infrastructure and
research activities are funded through fees paid by member organizations, an infrastructure grant from the
Ministère de la Recherche, de la Science et de la Technologie, and grants and research mandates obtained by its
research teams.
Les organisations-partenaires / The Partner Organizations
•École des Hautes Études Commerciales
•École Polytechnique
•Université Concordia
•Université de Montréal
•Université du Québec à Montréal
•Université Laval
•Université McGill
•MEQ
•MRST
•Alcan Aluminium Ltée
•AXA Canada
•Banque Nationale du Canada
•Banque Royale du Canada
•Bell Québec
•Bombardier
•Bourse de Montréal
•Développement des ressources humaines Canada (DRHC)
•Fédération des caisses populaires Desjardins de Montréal et de l’Ouest-du-Québec
•Hydro-Québec
•Imasco
•Industrie Canada
•Pratt & Whitney Canada Inc.
•Raymond Chabot Grant Thornton
•Ville de Montréal
© 2000 Marcel Boyer et Donatella Porrini. Tous droits réservés. All rights reserved.
Reproduction partielle permise avec citation du document source, incluant la notice ©.
Short sections may be quoted without explicit permission, provided that full credit, including © notice, is given to
the source.
ISSN 1198-8177
Ce document est publié dans l’intention de rendre accessibles les résultats préliminaires
de la recherche effectuée au CIRANO, afin de susciter des échanges et des suggestions.
Les idées et les opinions émises sont sous l’unique responsabilité des auteurs, et ne
représentent pas nécessairement les positions du CIRANO ou de ses partenaires.
This paper presents preliminary research carried out at CIRANO and aims at
encouraging discussion and comment. The observations and viewpoints expressed are the
sole responsibility of the authors. They do not necessarily represent positions of CIRANO
or its partners.
Instrument Choice in Environmental Policy
*
Marcel Boyer
†
, Donatella Porrini
‡
Résumé / Abstract
Nous analysons les conditions sous lesquelles les approches légale et
réglementaire peuvent être comparées dans le cadre d’un modèle d’économie
politique de l’implémentation de la politique environnementale. La première
partie de l’article décrit les caractéristiques essentielles des divers instruments à
comparer, à savoir un régime de responsabilité légale élargie aux prêteurs et un
régime de réglementation incitative, instruments typiquement utilisés aux
États-Unis et en Europe. Dans la deuxième partie, un modèle formel d’économie
politique est développé. La possibilité d’une capture de l’agence de
réglementation est introduite sous forme réduite par la surévaluation de la valeur
sociale de la rente informationnelle des entreprises. Nous montrons qu’un régime
de réglementation incitative peut être plus ou moins performant en termes de
bien-être qu’un régime de responsabilité élargie, stricte et solidaire. Nous
analysons en profondeur trois facteurs principaux de cette comparaison, à savoir
le différentiel de coût entre les niveaux faible et élevé de la protection
environnementale, le coût social des fonds publics et le facteur de surévaluation.
We analyze the conditions under which a legal intervention can be
compared to a regulatory framework in the context of a political economy model
of environmental policy. The first part of the paper describes the characteristics
of the different instruments we want to compare: first, an assignment of legal
liability, focusing on the case of extended lender liability, and second, an
incentive regulation framework. We briefly describe the application of those
instruments in the United States and Europe. In the second part a formal economy
model is presented where the possibility of capture of the regulatory agency is
modeled in a reduced-form fashion through an overvaluation of the social value
of the informational rent of the firms. We show that compared with an extended,
strict, joint and several liability system, a regulatory system may perform better or
worse from a welfare point of view. Three factors underlying this comparison are
discussed in some depth, namely the differential cost between low and high levels
*
Corresponding Author: Marcel Boyer, CIRANO, 2020 University Street, 25
th
floor, Montréal, Qc, Canada
H3A 2A5
Tel.: (514) 985-4002
Fax: (514) 985-4035
email: boyerm@cirano.umontreal.ca
We want to thank Claire Domenget, Patrick González, Yolande Hiriart, Ejan Mackaay and Stephen Shavell for their
comments. Donatella Porrini acknowledges financial support from MURST 1998 Project “Regulation and
Self-Regulation” (Università LIUC, Castellanza, Italy). We remain of course solely responsible for the content of this
article.
†
Department of Economics, Université de Montréal and CIRANO
overvaluation factor.
Mots Clés :
Politique environnementale, responsabilité élargie, CERCLA, capture des
régulateurs, instruments
Keywords:
Environmental policy, extended liability, CERCLA, regulatory capture,
instruments
1 Introduction
Fromalawand economicspointofview,theregulationofenvironmentally riskyactivities isan
alternativetoasystemofliabilityassignment. \Regulationandtortlawarealternativemethods
(though often used in combination) for preventing accidents. The former requires a potential
injurer to take measures to prevent the accident from occurring. The latter seeks to deter the
accidentbymakingthepotentialinjurerliableforthecostsofaccidentshoulditoccur." (Landes
and Posner, 1984, p. 417). We want here to review and characterize, in an incomplete
infor-mation political economy framework, the conditions under which an environmental regulation
approach issuperiorto an environmentalliabilityone.
We develop in this paper a formal analysis of the comparison between di erent policy
in-strumentstoimplementagivensetofenvironmentalprotectionobjectives, 1
includingapolitical
economy explanation of the choice of instruments. 2
The rst instrument we consideris the
as-signment of a CERCLA type liability, 3
that is, a strict, retroactive, joint and several liability
on theowners and operators of the rm responsiblefora catastrophic environmental disaster.
More precisely,wemodelan extended lenderliabilityrulewherebyprivatebanks nancingthe
responsible rm are consideredas liableoperators if the latter is unableto cover the damages
and compensation from its own assets. The second instrument in our comparison consists in
a regulation framework. After the environmental legislation of the 70s in the United States,
the federal government played an extensive role in regulating air pollution, water pollution,
hazardous and solid waste disposal, as wellas pesticideuse, among other environmental risks.
More precisely we consider here an incentive regulation system based on a menu of contracts
and subjectto capturebytheregulated rms.
BoyerandLa ont(1999)arguethattwotypesofmeaningfulcomparisonsofinstrumentsare
possible. Inthe rsttype,oneconsidersexogenousconstraintson instrumentsandthenvarious
constrained instruments can be compared. In the second type, instruments, equivalent in the
completecontractingframework,canbemeaningfullycomparedgivensomeimperfectionsinthe
economyoutsidethecontrolofthesocialorconstitutionalplanner. 4
Theoriginofthisimperfect
The extended lenderliabilityoption
Thecommonlawtortsystem,administeredbythecourtsandgovernedprincipallybystatelawin
the UnitedStates, provides a mechanism for creating incentivesforcare and forcompensating
victims, property losses and health injuries by a strict liability system. 5
Alongside the tort
system, there exist a systemof privateand public insurancebothfor theliabilityof rmsand
for theconsequences on individuals. In the80s, theUnitedStates Congress enacted CERCLA
andcreatedaSuperfundforthequickande ectivecleanupofdangerouswastesites. 6
Itgavethe
EnvironmentalProtectionAgency(EPA)thepowertobringdamageactionsto recovercleanup
costs againsttheowners andoperators of thefacilitydirectlyresponsibleforreleases.
We want to concentrate here on an important aspect of a liability system that makes all
owners and operators retroactively, strictly, jointly and severally liable, namely the extension
of liabilityto the lenders. In spite of a secured interest exemption clause protecting nancial
institutions holding indicia of ownership on the rm's assets, 7
the United States courts have
repeatedly considered secured lenders as owners or operators under CERCLA,insofar as their
involvement intheoperationsof the rmexceededthelevelwarrantedto securetheirinterest. 8
A lenders'liabilitysystem wasde ned by thecourtsdecisions, forinstance inthefollowing
landmark cases involving the bankruptcy of the primary responsible rm: USA v. Mirabile, 9
USAv. MarylandBank&Trust, 10
USAv. Fleetfactors, 11
andBergsoeMetalv. EastAsiatic. 12
Butthesecasesappearedtoarticulatepotentiallycon ictingrulesofliabilityregardingthetype
and degree of involvement making the lenders jointly liable with the responsible rms. 13
To
clarifythisconfusedsituation,theEPA issuedin1992thesocalledFinal Rule 14
underwhicha
lender would be liableforcleanup costsif itparticipated inthe management of the borrower's
operationsbyexercisingmanagementcontrolovereitherthedaytodayoperationsofthefacility
or over its environmental compliance e orts. In the years following theEPA's nalrule, some
court decisionswere basedonthisstatement. 15
Butinthe1994 caseKelleyv. EPA, 16
theD.C.
CircuitCourtofAppealheldthatCongressinenactingCERCLAdidnotgivetheEPAauthority
to e ect theimpositionof liabilityand therefore invalidatedtheEPA's nalrule. 17
In 1996, theAsset Conservation, Lender Liability, and Deposit Insurance Protection Act 18
According tothisAct,lendersandsecuredcreditorsmust`participateinthemanagementofthe
facility'tobeheldliableasan`owneroroperator'ofacontaminatedsiteandasecuredcreditor's
simple nancingtransactionsshouldnotimplyajoint,several,retroactiveandstrictliabilityfor
environmentalcontamination. But,whiletheActprovideswelcomedreliefforsecuredcreditors,
it does not completely insulate lenders and duciaries from environmental liability and the
questionremainsregardingwhichprecise stepsmustbetaken to ensurealimitation ofliability
forlenders. 19
Inadditiontothelenderliabilityruledevelopedthroughthejurisprudence,theCERCLA
li-abilitysystemraisesotherissues. First,suingallthepotentiallyresponsiblepartiesortargeting
some `deeppocket'onestorecoverresponse, cleanupcostsanddamages, aswellascoordinating
numerous parties withcon icting interests and nding an agreement on a cost allocation plan
maygeneratevery hightransactioncosts. Second,theinvolvement ofmanypotentially
respon-sible parties impliesthatthe distribution among pollutingparties of the needed compensation
costs cancreate incentiveproblemssuchastheallocationof resourcesto legalstrategies rather
than to accident prevention. 20
Inadditiontothistransactioncostproblem,theCERCLAliabilitysystemwasnotsupported
by a signi cant development inthe insurance market. 21
The main problems are the following.
The standard insurance policies do not t the CERCLA retroactive liability system because
they do not cover claims made before or after the validity period of the insurance contract.
Moreover,becauseboththepremiumandthedeductibleinthepoliciesareextremelyhigh,only
a few insurance companiesin the United States issue them and many lending institutionsopt
forsel nsurance. 22
In Europe, a uni edregime of liability forenvironmental damages is still in themaking. 23
Theproblemofharmonizingdi erentnationallegalregimesfromthestandpointofbothmarket
integration and environmental protection that cuts across traditionaladministrative and legal
boundariesraisesdiÆcultissues. 24
Inthiscontext,theWhite Paper on Environmental Liability
of February 2000 aims at determining who shouldpay forthe cleanup and restoration costs of
theenvironmentaldamageresultingfromhumanacts. Thequestionwhetherthecostsshouldbe
opted essentially for a strict (no fault)liabilitysystem that is e ective only forfuture damage
wherepolluterscanbeidenti ed,damageisquanti ableanda causalconnection canbeshown.
Given the generalrulethat thepollutermust always be the rst actor a claim is addressed to,
the White Paper nevertheless recommends a form of extended liability rule. It states that the
persons who exercises control (the `operators') of an activity by which the damage is caused
shouldbetheliablepartyandit speci esthatlendersnotexercisingoperationalcontrolshould
not be liable. Furthermore, in the nal part of the White Paper that deals with the overall
economic impact of environmental liability in the European Community, it is stated that the
liabilitysystemgenerally protects economic operatorsin the nancialsectors, unless theyhave
operationalresponsibilities.
TheEC White Paper liabilitysystemwhilesimilarto theUnitedStatessystemdi ersfrom
it on manyimportant aspects. First, both of them arebased on a strict liabilityregime inthe
sense thatliabilitycomesfrom the causallink betweentheactor and thedamage and whether
the actor's behavior wasproperor negligent is irrelevant. Second, whilethe CERCLA system
is applied retroactively, the EC White Paper provides a non-retroactive application. 25
Third,
onlya mitigatedjoint and several liabilityregime isprovidedintheEuropean caseinthesense
that a party isallowed to provide convincing argumentsthat itis onlypartiallyliable. Fourth,
instead of covering every damage including the damage to natural resources, the European
system covers only traditional damages, such as personal injury and damage to property, and
the decontamination of sites. Fifth, the objective of the United States system of recovering
the environmental damage from liableparties is supported also bythecreation of a Superfund
whileno such fundis establishedbythe WhitePaper. Sixth, theset ofactors whocan be held
liableisthesame,namelythe`operators'of the rm,andbothsystemsspecifythatlendersnot
exercisingoperationalcontrolshouldnotbeheldliable,thesocalledsecuredinterestexemption
rule. 26
The regulation option
An alternative instrumentto implement theenvironmental policyisa regulatory systemwhere
andcontrolprocedureofsettingandimplementingpollutionstandards. Amorerecentapproach
restsonincentivemarketbasedinstruments,asemissiontaxation,marketablepermitsando set
trading. 27
In the command and control approach based on a mandatory technology or abatement
standard,theregulatorsuchtheUnitedStatesEnvironmentalProtectionAgencycan orderthe
rmsto limittheiremissions, toemitno morethanaspeci edamountofa pollutantand/orto
installaparticularabatement technology. Theregulatormonitorsoverthetimethecompliance
of rmswith the standards and emission limitsthrough the conduct of inspections, actions in
federal courtsand negotiated settlementswithpolluters. 28
Theincentivemarketbasedinstrumentsarealternativetoolsthataretypicallybasedonthe
menuofcontractsframeworkoronasystemofmarketablepermits. Thelatteressentiallyworks
in thefollowingway: theregulatorgrantsaplant orpublicutilityanumberofpermitsto emit
a given amount of a pollutant;if thefacilityis able to reduce its emissions, preferablythrough
the useofnewertechnologies, itcan sell itsremaining emissionpermitsto anotherfacilitythat
is unableto meet its quota.
Looking at the United States experience, air pollution control under the federalClean Air
Act (1970-1990) followed in its early stages a command and control approach but with the
increasing knowledge and experimentation of market based solutions switched to markets of
pollution`rights'. GiventhatthemaingoaloftheCleanAirActwastheattainmentofnational
ambient air qualitystandards, the Congressasked the EPA to establish theNational Ambient
Air Quality Standards (NAAQSs) for pervasive air pollutants. Later on, euent taxes and
marketableemissionrightsweretakenintoconsiderationinordertoovercometheshortcomings
of thecommandandcontrolinstrumentsintermsofmonitoring,enforcementcapabilities,their
high levelof administrativecosts.
IntheEC,theEuropeanEnvironmentalAgency(EEA)hasalimitedregulatoryrolefortwo
reasons. First,theEEAexercisesmainlytheroleofprovidingobjective,comparableandreliable
information that member States or the Community at large may use to develop measures to
protect the environment, to evaluate the results of said measures, and to educate the public
of addressing directly and credible the environmental problems of the Community. Therefore
in every single member State, the regulation follows the national legislation and the choice of
instrumentsis speci c to eachstate.
Liabilityversusregulation
To comparethe two policy instrumentswe can follow alawand economicsapproach analyzing
their impacts interms of socialwelfare. 29
This kind of analysis balances the bene ts from the
risky activitieswith thecostsofprecautionarycare, theexpected levelof damages(probability
and severity),theadministrativeexpensesassociatedwiththese policies,andthenetsocialcost
of theinformational rents.
Astrict liabilitysystemistypicallyappliedto riskscreatedbyabnormallyhazardous
activ-ities and against defendantsforall injuries caused by theirconduct. The victim lesan action
claiming acausallinkbetweenthedefendant's conductand theplainti 'sinjuryordiseaseand
the system relies on a case by case adjudication. Strict liabilityregime has the advantage of
internalizingenvironmentalrisksbothfrom theincentiveand thecompensationpointsof view.
But it has some practical disadvantages: inmany cases the victimsare widely dispersedwith
noneofthemsuÆcientlymotivatedtoinitiatealegalaction,harmmayappearonlyafteralong
delay, speci callyresponsiblepolluters may be diÆcultto identify,determiningthe causallink
maybediÆcult,inconsistentverdictsmayemerge,delaysincourtproceedingsmaybeverylong
and the systemmaybe morepro table forlawyers and experts thanforthe victims.
Ontheotherhand,aregulationsystemistypicallycharacterized bya centralizedstructure.
Its advantages are based on the fact that it is wellsuited to set policies regarding the
de ni-tion and implementationof standards. The centralizedsearch facilities,thecontinualoversight
of problems and a broad array of regulatory tools can make the regulation system capable of
systematically assessing environmental risksand of implementing a comprehensive set of
poli-cies. But,regulatory agencies may be not very exibleinadapting to changingconditionsand
centralized commandstructurerelyingon expertadvicemaybe subjectto politicalpressureas
wellasto collusionand capturebytheregulated rms. 30
privateknowledgeoftheparties regardingthecausalfactorsofaccidentprobabilityandtherisk
of captureorcollusion.
The cost of a liabilitysystem includes theadministrative expenses incurredbythe private
and the public parties, namely the cost of optimally controlling the probability of accidents,
the legal expenses and the public expenses for maintaining legal institutions. The cost of the
regulatory systemincludesthepublicexpensesformaintainingtheregulatory agenciesand the
privatecosts of compliance. One advantage of the liabilitysystem is that a signi cant part of
the administrative costsisincurredonlyifasuitoccurs. Ontheother handtheadministrative
costs of a regulation system are incurredwhether or notthe harm occurs because the process
of regulationiscostly byitself andtheregulatorneedsto collect informationabouttheparties,
their activitiesand therisks.
A second element of thecomparison refersto whobears thecost of environmental damage.
In aregulation system,thecosts areusuallydirectlyorindirectlycovered bythepublicparties
when duecare wasexercised bythe rmsaccording to thestandards de nedbythe regulatory
agency. Ina liabilityregime, these costsare imposed on theresponsible privateparties, ifand
when a suit occurs, given their capacity to pay and their limited liability. Both systems may
require some form ofcompulsoryinsuranceforthe losses inexcess of theassetsof the rmbut
the liabilitysystem can also rely on an extended liabilityassignment according to which most
or all deep pocket stakeholders (suppliers, partners and nanciers) of the rm may be made
strictly,jointlyand severallyresponsibleforthedamages.
Athirdimportantelementofthecomparisonisthedistributionofknowledgeamongparties
regardingthebene tsofactivities,thecostofreducingrisksandtheprobabilityandtheseverity
of accidents. Sometimes the nature of the activities carried out by the rms is such that the
privatepartieshavebetterknowledgeofthebene tsandcostsofreducingrisks. Insuchacasea
liabilitysystemhastheadvantageofmakingtheprivatepartiesresidualclaimantsofthecontrol
of riskswhilearegulation systemsu ersfrom thelackofinformationleadingto overestimation
orunderestimationofthecostsandbene tsoftherisks(probabilityand/orseverity). Butitmay
also happenthat theregulatorhasbetter knowledgeof those risksbecause ofthe possibilityof
centralizing informationanddecisions,inparticularwhenabetterknowledgeoftheriskfactors
Afourthrelevantfeatureinthecomparisonisthepossibilityofcaptureandcollusionbetween
the enforcers and the parties. The enforcers may be in uenced by external pressure in both
systems, butone may reasonablyargue that thecourts are less likelyto be capturedthan the
regulating agencies.
On the basis of theses di erences between the two policy instruments, we will present in
the next section a model based on the stylized features of an extended lender liabilitysystem
and ofanincentiveregulationsystemwheretheasymmetricinformationbetweenparties(moral
hazard) andthe possibilityof captureare explicitlypresent.
2 The model
We consider a two period context where a rm can, in each period, invest an amount F to
generate a lowpro t levelof L
withprobabilityora highpro t levelof H
with probability
(1 ), withexpected pro t = L
+(1 ) H
. 32
Thestochastic revenuesare i.i.d. and the
discountrateiszero. The rmcanchooseself-protectionactivitiesethatreducetheprobability
p(e) of a major environmental accident generating damages of d > H
. Therefore if a major
environmental disaster occurs in period 2, it sends the responsible rm into bankruptcy. We
will assume thatthe self-protectionactivities are exerted in period 1 and that an accident can
happeninperiod 2only, ifitdoesoccur. The self-protectionactivities can be at thehigh level
e h
or at thelow levele `
; we willassume forsimplicity that thecost of the low level e `
is zero
and that the (di erential)cost of the highlevel is . Letp(e h ) =p h and p(e ` ) =p ` . We will assume that <(p ` p h
)d and therefore it is socially optimal ina rst best sensethat the
rm chooses thehighlevelof self-protectionactivities.
We willassume forsimplicitythatthe rmhasno equityand must borroweach periodthe
fullamountF inorder to remaininbusiness. Weconsidertwo regimes. Inthe rst regime,the
rm interacts witha privatebankerwho istheresidualliablepartyforenvironmentaldamages
caused bythe rm,thatis fordamagesabove theassets ofthe rm. The rmis assumedto be
risk neutral butwith limited liability. The bank is assumed to be a deep pocket private bank
whoselimitedliabilityisirrelevant. Inthesecondregime,the rminteractswitharegulatorwho
the rmborrows fromtheprivatebank. Under theregulatoryregime,weassume forsimplicity
thatthe rmborrowsfromtheregulator. Clearly,arealregulatordoesnot nancethe rmbutin
hercomplexrelationshipwiththe rm,shewouldworryaboutthe nancialviabilityofthe rm
and also theimpactof nancial contracts onincentivesforself-protectionactivities. Creatinga
direct nancial linkbetweenthe regulatorand the rm isa reducedform representationof the
structural relationshipbetween theregulator,the rmand the nancial markets.
We want to concentrate here on the prevention of environmental accidents and so, the
in-formation structurewe consideris asfollows: althoughtherealizedpro tlevel isobservable by
everyone, thelevelof self-protectionactivities is aprivateinformationof the rmand is
there-fore observable neither bythe regulatornor by thebank. The timingof theinterplaybetween
the principal(eitherthepublicregulatorortheprivatebank) and the rmisasfollows inboth
regimes considered. The principal o ers a nancial contract to the rm making explicit the
payments to be made in each period if the rm is nanced. If the contract is accepted, the
rm invests F and chooses the carelevel e. The pro t levelof the rst periodisthen observed
and a payment is made to the principal according to the nancial contract. In period 2, the
rm is re nancedornot andif re nanced, itinvests F again,thepro t level isobserved and a
catastrophic accident occursornot. Apaymentismadeto theprincipalaccording tothe
nan-cial contract and,if an accident occurs, cleanup costsare distributedaccording to theliability
system inforce.
We willcharacterize and compare three solutions. The benchmark solutionwillcorrespond
to the casewhere abenevolent regulator, notsubject to capture,chooses the nancialcontract
o eredtothe rminordertomaximizeautilitariansocialwelfarefunction. Thesecondsolution
will be obtained when a privatebank, under an extended lender liabilitysystem, chooses and
o ers a nancial contract that maximizes its own expected pro t functionin which the
infor-mational rent ofthe rmis notpresent. Thethirdsolutionwillbe obtainedwhen thecaptured
regulator chooses the nancial contract o ered to the rm. In so doing, she maximizes a
dis-tortedsocialwelfarefunctioninwhichtheinformationalrentofthe rmwillbeovervalued. Ina
sense, therearethree possibleprincipalsinthiscontext: thebenevolent regulator,thecaptured
3 Moral hazard in environmental protection
Clearly, the asymmetric information structure and the limited liabilityof the rm makes the
internalizationofexternalitiesadiÆcultproblem. Ifamajorenvironmentaldisasteroccurs,the
rm willbe `judgment-proof' for damages above its value, that is here, its pro t level. Under
limited liability,moralhazard variablescannotbecostlesslycontrolledbyimposingappropriate
penaltiesontheriskneutral rmandthelatterwillingeneralbeabletocaptureaninformational
rent. Accident-preventingactivities bythe rmmustthenbeinducedbyhigherrewardsrather
than sti penaltiessincethelimitedliabilityconstraintimposes a limitonthose penalties.
Given thatthepro t levelisobserved byallparties,theprincipalis ableto o era nancial
contract wheretherepaymentlevelisafunctionofthepro tlevel. Butbecausethelevelof
self-protectionactivitiesisnotobserved,therepaymentlevelmustbeindependentofthoseactivities.
So wewillassume thatthe nancial contract stipulatesthatin period t, theprincipalwilllend
theamountF andaskforrepaymentlevelsofR t L ifrealizedpro tis L andR t H ifrealizedpro t is H
. A nancial contract is therefore a 4-tuple of repayments forloans of F in each period:
(R 1 L ;R 1 H ;R 2 L ;R 2 H
). The objective function of theprincipal will depend on the setting, that is,
on whether theprincipalis abenevolentregulator, acaptured regulatororaprivatebank,and
whether the principal haspriorityor not over the rm's pro t in case of bankruptcy. We will
assume here that if an accident occurs, the rm must pay for the damages at least up to the
maximalamount madepossiblebyitslimitedliability. Sinced> H
,itmeansthatallits pro t
willbe taken awayifan accident occursandno payment isthen madeto theprincipal.
Under ourassumptions, the fullinformation rst best allocationentails clearlya high level
of self-protectionactivities e h
and aloan/investment F in bothperiodsi
2 2F p h
d 0; (1)
aconditionwhichissatis edifwehaveabenevolentregulatororacapturedregulator. Thetwo
regulators di er by their treatment of the rm's informational rent but since the rent is zero
under full information, this di erential treatment has no impact. In the absence of extended
lender liability,theprivatebanklendsineach periodi
leading to overinvestment because of the partial, rather than full, internalization of the
exter-nality. With the extended liabilityof the deep pocket private bank, the full information rst
bestallocationis alsoachieved at theNashequilibriumof thegameplayedbythe rmand the
bank. The bank'sliabilityinducesit to fullyinternalizethe externalityand being riskneutral,
it prefers the optimal level of e ort e h
. Hence, one may suggest that a possible solution to
thefullinternalizationoftheexternalitycreatedbyenvironmentalaccidentsisto makethe
pri-vate bank responsiblefordamages if thejudgment-proof rm it nances causesa catastrophic
environmentalaccident. 33
But when the principal, whether it is the benevolent regulator, the captured regulator or
the privatebank,su ers from agency problemsin its relationshipwiththe rm,thepossibility
of achieving the rst best must be quali ed. As mentioned before, we consider in this paper
that the rm's pro t is observable by the regulators and the private bank but that they all
face a moral hazard problem regarding the level of the rm's accident preventing activities.
We will characterize rst the social optimum to be used as a benchmark. This benchmark
corresponds to the case of the benevolent regulator who maximizes the proper social welfare
functionbutinsodoingmusttake intoaccounttheprivateinformationofthe rmregardingits
self-protectionactivities. Next,wewillcharacterizetheNashequilibriumobtainedforthegame
involving the rm and the privatebank under the extended lender liabilityregime. Then, we
willcharacterizethesolutionobtainedwhen theregulatoris captured. Finally,wewillcompare
the three solutions and derive some propositions on therelative socialeÆciency of the regime
of incentive regulationimplementedbya capturedregulatorand theregime of extendedlender
liability.
The socialoptimumundermoral hazard.
Because of asymmetric information, thefull information rst best allocation is not achievable
anymore. Theproperbenchmarkforouranalysisisthesocialoptimumundermoralhazard
be-causeeventhebenevolentregulatorwhoseobjectiveistomaximizesocialwelfaremusttakeinto
account the agency costs. We willassume thatthe socialwelfare function(SWF) is utilitarian
and that there is a socialcost of publicfunds(1+) comingfrom distortionsdue to taxation:
rm to the benevolent regulator acting here as a nancier together with the cost F invested
by the benevolent regulator will enter the social welfare function with a weight of (1+), in
the rstcase becausethey allowa reductionintaxation and inthesecondcase because the
in-vestment F mustbe nanced throughtaxation, directlyorindirectly. We willassumealso that
the expected damage of an accident enters thesocial welfare functionwith a weight of (1+)
becausethegovernmentwillhavetocoverthatcostinonewayoranotherand nanceitthrough
taxation. Giventhat the rm's netutility(rent)is notobservable by thebenevolent regulator,
and therefore nottaxable, thisnetutilityenters thesocialwelfarefunction witha weight of 1.
It willtherefore be eÆcient that the benevolent regulator recuperates any observable pro t of
the rm. 35
The rm willbe left with its unobservable informational rent which will thenhave
a weight of 1 in the social welfare function. The socially optimal program of the benevolent
regulator willtherefore minimize the rent left to the rm because of its smaller weight in the
social welfare function. The existence of a social cost of public funds is an important and
re-alistic feature ofregulatory frameworks. It makes incomedistribution relevant, althoughin an
unusualsense, forenvironmentalprotection. Werethat cost equalto zero, theregulator would
not care whetherthe rmmakes monopoly pro ts orcapturesigni cant informationalrentsas
long astheeÆcient productionlevelisrealized. Theexistence ofapositive togetherwiththe
assumption of a regulator acting as nancier will allow us to develop a tractable yet realistic
modelof instrumentchoice inenvironmental protectionpolicy.
The social optimum under moral hazard maximizes the expected social welfare under the
incentive compatibility, limited liabilityand individual rationality constraints of the privately
informed rm. The rmwillchooseahighlevelofself-protectionactivitiesi it ndspro table
to incurthedi erentialcost ,thatis, i its expected netutilityinthe islarger withe=e h than withe=e ` ,thatis i : (1 p h )[ (R 2 L +(1 )R 2 H )] (1 p ` )[ (R 2 L +(1 )R 2 H )]
whichcan be rewrittenas
R 2 L (1 )R 2 H p ` p h ; (3)
which is the incentive compatibility constraint to be satis ed if the principal wants to induce
repay-ment levels notexceedthe corresponding pro t levels. Finally,the rm'sindividualrationality
constraint is that its net utility be non negative (assuming an exogenous utilitynormalized at
zero). The rm's expected net utility is 0 under e = e `
and given by its informational rent
undere=e h : R 1 L (1 )R 1 H +(1 p h ) R 2 L (1 )R 2 H . Distortionscreated
by the presence of moral hazard will occur only when the combination of the limited liability
constraintsandtheincentivecompatibilityconstraint(3)requiretogiveupa(costly)renttothe
rm. The existence of a socialcost of public fundsrequires that R 1 L = L , R 1 H = H and that
(3) be satis edwithastrict equality: because>0,itissociallybetter to usethepro tofthe
rm to reduce the generaldistortionary taxes. Therefore, the netutilitylevelor informational
rent ofthe rm is R +(1 p h ) p ` p h >0: (4)
Under the socially optimal nancial contract, the benevolent regulator collects an expected
amount of +(1 p h )(R 2 L +(1 )R 2 H ) p h [(d R 2 L )+(1 )(d R 2 H )] 2F: (5) Proposition1: If 1+ R+ (p ` p h )d; (6)
thesocialoptimum(thebenevolentregulatorsolution)ischaracterizedbyahighlevelofaccident
preventing activities andan investment F inbothperiodsi
2 2F p h d 1+ R0: (7)
If (6) is notsatis ed, thesocial optimumis characterized bya low levelof accident preventing
activities and an investmentin bothperiodsi
2 2F p `
d0: k (8)
Proof: Let us rst derive the social welfare when e h
is induced. We must solve the following
program Maxf(1+)[R 1 L +(1 )R 1 H +(1 p h )(R 2 L +(1 )R 2 H ) 2F] (1+)p h (d ) +[( R 1 (1 )R 1 )+(1 p h )( R 2 (1 )R 2 )]g 9 > > > > > = > > > > > ; (9)
subjectto the incentivecompatibilitycondition(3)and theindividualrationalitycondition R 1 L (1 )R 1 H +(1 p h )( R 2 L (1 )R 2 H )0: (10)
The solutionentails R 1 L
+(1 )R 1 H
=and (3)satis edwithastrict equalitybecause ofthe
di erent weightsinthe SWF.Accordingly,the socialwelfare(9)can be writtenas
(1+)[ 2F+(1 p h )( p ` p h ) p h (d )] +[(1 p h ) p ` p h ]; hence as (1+)[2 2F p h d ] [(1 p h ) p ` p h ]; that is, (1+)[2 2F p h d ] R: (11)
Therefore, investment shouldtake place if (7) issatis ed. If e=e `
, no rent is left to the rm,
the socialwelfarebecomes
(1+)(2 2F p `
d) (12)
and investmentmusttake placeinbothperiodsif(12)ispositive. Comparingthesocialwelfare
levels (11) and (12), we obtainthate=e h mustbe inducedif 1+ R+ (p ` p h )d; (13)
where the right hand side is the incremental value and the left hand side is the incremental
cost, includingthesocialcostof theinformationalrent,of thehighlevelof accident preventing
activities. This completesthe proof. }
Proposition 1 di ersfrom the rst best fullinformation rulebecause of the presence of R,
the rent to be given up to the rm when thebenevolent regulator wantsto induceahigh level
of accident prevention activities. The benevolent regulator cannot avoid giving up that rent
to inducea high levelof accident preventing activities and willtherefore take into account the
net social cost of that rent, namely R. If that cost is large, the benevolent regulator may
prefer, in maximizing the SWF, to induce a low level of care e `
generating a high probability
p `
investment decision and deciding on the optimal level of care activities, the socialcost of this
rent must be accounted for. Asdecreases, thenet socialcost of the rm's informationalrent
decreases andcondition(13)converges totheconditionfore=e h
under fullinformation. As
increases,thenetsocialcostofgivinguparenttothe rmgoesupandcondition(13)converges
to thecondition,to bederivedbelow,fore=e h
under extendedlenderliabilitysince=(1+)
converges to 1as !1.
The Nash equilibriumwhen the rm faces aprivatebank.
We now considerthecase wherethe rm facesa privatebankerwho isliableforenvironmental
damages caused by the rm when the latter is unable to cover those damages from its assets,
here its pro ts. Clearlyas in the above case of a benevolent regulator actingas nancier, the
privatebank can o er a care-inducing contract to the rm but inso doing will concede a rent
to the rm asexpressed by (3). Otherwise,thebank can capture thewholepro t. The bank's
expected pro tunder acontract inducingahigh levelof careactivities e h is,using(3), +(1 p h )( p ` p h ) p h (d ) 2F that is 2 2F p h d (1 p h ) p ` p h ; (14)
whileunder thealternative contract inducingthelowlevelof caree `
,its pro tis
2 2F p `
d: (15)
Proposition2: Underextended lenderliability,theprivatebankinducesahighlevelofaccident
preventing activities less often than thebenevolent regulator does. When the bank decides to
induce e=e h
conceding a rent Rto the rm,it lendsless oftenthan thebenevolent regulator
does. When, inspite of lenderliability,thebank opts to induce thelow level of care activities
e `
leavingno rent to the rm,it lendsasoften asthebenevolent regulatordoesinthatcase. k
Proof: Comparing(14) and (15), theprivatebankopts forinducinge=e h i R+ <(p ` p h )d (16)
because
1+
R+ <R+ :
Considering (14)and using(4), theprivatebankwilllend withe h
i
2 2F p h
d R0: (17)
Comparing(17)and(7),obtainedinthecaseofthebenevolentregulator,showsthattheprivate
bank lends less often thanthe benevolent regulator does. Similarly,since (15) and (8) are the
same thentheprivatebanklendsasoftenasthebenevolentregulatordoesinthatcasesinceno
rent isleft to the rmand the bankinternalizescompletelythecost of an accident. }
TheintuitionbehindProposition2 isthatundertheextendedlenderliability,thecostof an
accidentforthebankisthesameasforthebenevolentregulatorandsothecomparisonbetween
thetwosolutionsrestsontheirdi erentevaluationofthe rm'srentwhene=e h
isinduced. For
the bank,thecost of therent is equalto thevalue of therent itself Rwhileforthe benevolent
regulator the net cost is smaller, namely 1+
R, because she considers the socialvalue of that
rent in the SWF. This makes the bank less willing than the benevolent regulator not only to
lend but also to induce a high level of accident preventing activities. Hence this unavoidable
informational rent leads to insuÆcient nancing and too little care activities induced by the
bank. If the bank chooses to induce e = e `
, there is no rent and therefore the bank lends as
often asthebenevolent regulator.
The biasedoptimumwhenthe regulatoriscaptured
If theregulatoris captured,shewillina sensebene tfrom the rm'srentone way oranother,
that is, throughbribes, collusive interests, perks, future employment opportunities and so on.
It willbeasifsheputstoo muchweight (overvaluation) onthe rm's informationalrent inthe
objective function, thatis,asifshe undervaluesthesocialcost of thatrentincomparison with
thebenchmarkcaseofthebenevolentregulator. Thiswillmakethecapturedregulatorlesskeen
to reduce thisrentto its minimum.
We will assume that the rent R of the rm, when e = e h
is induced, enters the captured
regulator's objective function with a weight of K, where 1 < K < (1+). The captured
is induced, (1+)[ 2F +(1 p h )( p ` p h ) p h (d )] +KR;
that is,using(3),
(1+)[2 2F p h d ] (1+ K)R (18) and, when e ` isinduced, (1+)[2 2F p ` d]: (19) We have:
Proposition3: The captured regulator induces a high level of accident preventing activities
more often than the benevolent regulator does. When she induces e = e h
conceding a rent
to the rm, she lends more often than called for, conditionallyon e= e h
, by the second best
optimalinvestmentrule. Whenthecapturedregulatorinducesalowlevelofaccidentpreventing
activities e `
leavingnorentto the rm,shelendsasoftenascalledfor,conditionallyone=e `
,
bythesecond bestoptimalinvestment rule.k
Proof: Comparing(18) and (19), we obtainthatthecaptured regulatorinducese=e h i 1+ K 1+ R+ <(p ` p h )d
whichcompared with(6) showsthat thecapturedregulatorinduces e h
too oftensince
1+ K 1+ R+ < 1+ R+ : When e h
is induced,thecaptured regulator'sobjective function(18) ispositivei
2 2F p h d 1+ K 1+ R0: Since 1+ K 1+ < 1+
, the social cost of the rent is undervalued and therefore, the capture of
the regulator leads to overinvestment, conditionally on e = e h
, in the environmentally risky
activities. When e `
is induced, we observe from (12) and (19) that theinvestment rules of the
captured regulatorand ofthe benevolentregulatorare thesame. }
4 The choice of instruments
LetSWF CR
bethevalueoftheSWF(9)withthesolution,levelofcareactivitiesandinvestment
valueoftheSWFwiththesolution,levelofcareactivitiesandinvestment rule,implementedby
the private bankunder extended lender liability,as de nedin proposition2. Let usde ne the
correspondences;; asfollows: SWF CR > SWF PB i 2() SWF CR = SWF PB i 2() SWF CR < SWF PB i 2 ():
We now turn to the characterization and illustrationof those correspondences. 36
But rst, let
us recallthemainresult ofour analysissofar.
When e = e `
is induced, the investment rules are the same in the three cases considered,
namely thebenevolent regulator, theprivatebank and thecapturedregulator.
When the benevolent regulatoris the principal,we obtainthat theprobabilitythat e=e h
will be induced is theprobability that 1+ R+ is less than (p ` p h
)d and the probability
that an investment 2F willbe made in the rm is the probability that 1+
R+ is smaller
than2 2F p h
d. Thisisthebenchmarkcasecorrespondingto thesecondbestsolution,that
is, thesocialoptimumunder moralhazard.
Whentheprivatebankistheprincipal,theprobabilitythate=e h
willbeinducedisnowthe
probabilitythatR+ islessthan(p `
p h
)dandtheprobabilitythataninvestment2F willbe
made inthe rm isnowtheprobabilitythatR+ issmallerthan2 2F p h
d. Compared
with thebenchmark case, thissolutionrepresentsalossof socialwelfarebecauseof notenough
incentive forcare,hencetoomanyaccidents,andnot enoughinvestmentintheenvironmentally
risky operations of the rm. The loss in welfare is due to the bank's overvaluation of the net
socialcostoftheinformationalrentcapturedbythe rm,thatistothebank'sfailuretoconsider
the socialvalue of therent.
Whenthecapturedregulatoristheprincipal,wethenobtainthattheprobabilitythate=e h
willbeinducedistheprobabilitythat
1+ K 1+ R+ islessthan(p ` p h
)dandtheprobability
that an investment 2F will be made in the rm is then the probabilitythat
1+ K 1+
R+ is
smallerthan2 2F p h
d. Again,comparedwiththebenchmarkcase,thissolutionrepresentsa
the rm. Thelossinwelfareisdueto thecaptured regulator'sundervaluationof thenetsocial
costoftheinformationalrentcapturedbythe rm,thatisthecapturedregulator'soverweighing
of thesocialvalue ofthe rent.
Theinvestment rulesinthethree contextsarethesame ife=e `
isinduced. Butthey di er
ife=e h
isinduced. Moreovertheconditionsunderwhiche h
isinduceddi erbetweenthethree
contexts.
In thebenevolent regulator(BR) solution:
e=e h i 1+ R+ (p ` p h )d;
that is,using(4), i
(1+)(p ` p h )d 1+ 1 p h p ` p h : (20)
Investment 2F willtake place i
2 2F p h d 1+ R0; that is,i (1+)(2 2F p h d) 1+ 1 p h p ` p h : (21)
Intheprivatebank (PB) solution:
e=e h i R+ (p ` p h )d; thatis,i (p ` p h ) 2 d 1 p h : (22)
Investment 2F willtake place i
2 2F p h d R0; that is,i (p ` p h )(2 2F p h d) 1 p : (23)
In thecapturedregulator(CR)solution: e=e h i 1+ K 1+ R+ (p ` p h )d; that is,i (1+)(p ` p h )d K+ (1 p h )(1+ K) p ` p h : (24)
Investment 2F willtake place i
2 2F p h d 1+ K 1+ R0; that is,i (1+)(2 2F p h d) K+ (1 p h )(1+ K) p ` p h : (25)
Supposethat=0. Inthatcase, theBRimplementsthe rst bestsolutionsince,althoughthe
rm cancaptureaninformationalrent, thatrenthasnosocialcost. SotheBRinducese h since by assumption < (p ` p h
)d and nances the rm i 2 2F p h
d 0. The PB
solution isindependent of thevalueof . Using(4), weobtainthat thePB inducese h ,leaving a rent R to the rm,i p ` p h 1 p h (p ` p h )d:
Otherwise, thePBinduces e `
,leavingno rent to the rm. Conditionallyon e=e `
,theprivate
bank lends asoften ascalled forbythe rst bestrulesince there is no rent. The social lossin
welfarein thiscase is thewelfare lossdueto inducinge `
ratherthan e h
. Moreover, even when
the PBprefersto inducee h
,itdoesnotfollowthat it nancesthe rm. In fact, if
>(2 2F p h d) p ` p h 1 p h ;
the PBwillnot nancethe rm withe=e h
,contraryto the rstbestrule. It maystill nance
the rm with e = e ` . But (2 2F p h d) p ` p h 1 p h > p ` p h 1 p h (p ` p h )d, as in Figure 1 below, i 2 2F p `
-(=0and 2 2F p ` d>0) 0 A p ` p h 1 p h (p ` p h )d B (2 2F p h d) p ` p h 1 p h C (p ` p h )d D In A, we have e PB = e h and e BR = e h
, and nancing occurs in both the BR and PB
solutions: no welfarelossin thePBsolution.
In Band C, we have e PB =e ` and e BR = e h
, and nancingoccurs inboth the BRand
PB solutions: thewelfarelossin thePBsolution corresponds to thehigherthan eÆcient
level ofaccidents. In D, we have e PB = e ` and e BR = e `
, and nancing occurs in both the BR and PB
solutions, since 2 2F p `
d 0, independentlyof which is notpaid since e= e `
:
there is nowelfarelossinthePBsolution. 37
If 2 2F p `
d<0, thenwe have thefollowing:
-FIGURE 2 (=0and 2 2F p ` d<0) 0 A 0 p ` p h 1 p h (p ` p h )d B 0 (2 2F p h d) p ` p h 1 p h C 0 (p ` p h )d D 0 In A 0 , we have e PB = e h and e BR = e h
, and nancing occurs in both the BR and PB
solutions: no welfarelossin thePBsolution.
In B 0 ,we have e PB =e h and e BR =e h
,but nancingoccursonlyintheBRsolution: the
welfarelossinthePB solutioncorrespondsto notrealizing theinvestment.
In C 0 , we have e PB = e ` and e BR = e h
InD 0 ,wehavee PB =e ` ande BR =e `
,and nancingoccursinneithertheBRnorthePB
solutions: no welfarelossin thePBsolution.
Similarly,comparingthesolutionsunderBRandCRforthecase2 2F p ` d>0,weobtain: -FIGURE 3 (=0and 2 2F p ` d>0) 0 E (p ` p h )d G p ` p h 1 p h K(1 p ` ) (p ` p h )d H In E, we have e CR = e h and e BR = e h
, and nancing occurs in both the BR and CR
solutions: no welfarelossin theCRsolution.
In G, we have e CR
=e h
, leaving arent to the rm,and e BR
=e `
,leavingno rent to the
rm, and nancing occursin boththe CRand theBRsolutions. The welfareloss inthe
CR solution corresponds to a levelof accidents that is too low (fromthe level of care e h
generatinga rent forthe rm)!
InH,wehavee CR =e ` ande BR =e `
,and nancinginboththeCRandtheBRsolutions:
no welfarelossintheCRsolution.
Ratherthanproceedwithageneralanalysisofthecaseswith>0,letusconsiderthefollowing
illustrative numericalexample:
L =5; H =10; =0:5; F =5; p ` =0:1; p h =0:05; d=20; K=1:2
for which =7:5and (p `
p h
)d=1. For thiscase, we can graph thefrontiers(20) to (25) as
-FIGURE4 [2 2F p ` d0] 6 0 1 1 (p ` p h )d (22)(20)(24) (23) (21) (25)
Conditions (20), (22) and (24) relate to the decisionabout the level of care activities, namely
e BR
=e h
i is to the left of (20) fora given , that is,i is below (20) fora given ,
e PB
=e h
i is to the left of (22), irrespective of thevalue of , and e CR
=e h
i is to
the left of (24) for a given , that is,i is below (24) fora given . Conditions(21), (23)
and (25) relate to thedecisionaboutthe investment inthe rm,namely BRinvests withe h
i
isto theleftof (21),PBinvestswith e h
i isto theleftof (23)and CRinvests withe h
i is to theleft of (25). Therefore, fortheexampleconsidered, all three principalsinvestif
indeed they decide to induce e h
from the rm. When they inducee `
, they all follow thesame
rule since there is then no rent left to the rm, namely they all invest i 2 2F p `
d 0.
If we assume that this last condition is satis ed, then there will always be investment in the
rm. The di erence between the three solutions comes from the di erent decisions regarding
theinducement ofcareactivities. ConsiderFigure4. Fora situation( ;)to theleftof(22),
situation( ;)between(22)and (20),e BR =e CR =e h bute PB =e `
,whiletheyall investin
the rm: inthisregion, thecapturedregulatorsolutionispreferredtotheprivatebanksolution.
For a situation ( ;) between (20) and (24), e BR = e PB = e ` but e CR = e h
, while they all
investinthe rm: inthisregion,theprivatebanksolutionispreferredtothecapturedregulator
solution, even if there willbe more accidents in theformer solution. The larger number, more
preciselythehigherprobability,ofaccidentsismore thancompensatedbythefactthatthere is
no (costly) rent left to the rm. Finallyfora situation( ;) above (24), allthree principals
inducee `
and investinthe rm: thethree solutions arethesame. Therefore:
Proposition4: If2 2F p `
d0,that is,ifthe rmis sociallypro table witha lowlevel of
care, thenthe`extendedlenderliability'regimeandthe`regulatorsubjecttocapture'regimeare
equivalentinstrumentsforimplementingtheenvironmentalpolicyifthedi erentialcostbetween
high and low levels of accident prevention activities is relatively smallor relatively large (as a
function ofthesocial costof publicfunds), that is,if
2() " 0; (p ` p h ) 2 d 1 p h # [ 2 6 4 (1+)(p ` p h )d K+ (1 p h )(1+ K) p ` p h ; 1 3 7 5 [ f (1+)(p ` p h )d 1+ 1 p h p ` p h g:
The `regulator subjectto capture'regime is abetter instrumentifthe di erentialcost between
high and low levels of accident prevention activities is inthe lower intermediate range, thatis,
if 2() 0 B @ (p ` p h ) 2 d 1 p h ; (1+)(p ` p h )d 1+ 1 p h p ` p h 1 C A :
The `extendedlenderliability'regime isabetterinstrumentifthedi erentialcostbetweenhigh
and lowlevels ofaccident prevention activities isinthehigherintermediaterange,that is,if
2 () 0 B @ (1+)(p ` p h )d 1+ 1 p h p ` p h ; (1+)(p ` p h )d K+ (1 p h )(1+ K) p ` p h 1 C A : k
If investing in the rm is not socially desirableor pro table unless e= e h
is induced, that
is, if2 2F p `
d<0,we obtaina con gurationrepresentedinFigure 5, wherep `
=0:3, for
-FIGURE5 [2 2F p ` d<0] 6 0 1 5 (p ` p h )d (22) (23) (21) (25) (20) (24) ^ ^ ^ ^
We observe the following. Fora situation ( ;) to theleft of (23), all three principalsprefer
to inducee h
and invest inthe rm: the three solutions arethe same. To the right of (23), the
privatebank doesnotinvestanymorebecause istoo high. Forasituation( ;)between
(23)and(21), e BR =e CR =e h
andbothinvestinthe rm: inthisregion,thecapturedregulator
solutionispreferred totheprivatebanksolution. Totheright of(21), thebenevolentregulator
doesnotinvestanymore. Althoughshewouldprefertoinducee h
[for( ;)between(21)and
(20)], is too high and therefore the rent to be left to the rm is too costly. Note that as
expected,thenoinvestment triggervalueisdecreasingin: thehigherthesocialcost ofpublic
funds, thefaster the benevolent regulatorstops investing as thedi erentialcost increases.
Forasituation( ;) between(21)and(25), e CR
=e h
and thecapturedregulatoristheonly
principal still interested in investing in the rm: in this region, the private bank solution is
cost of the rent left to the rm. In thisregion of parameter space, it is socially better not to
nance the rm. Finallyfora situation ( ;) above (25), noneof thethree principalsinvest
in the rm: the three solutions are the same. We obtain a comparison similarto the case of
proposition4 butwithdi erent boundaries. We have:
Proposition5: If 2 2F p `
d 0, that is, if the rm is socially pro table only with a high
level of care, then the`extended lender liability' regime and the `regulator subject to capture'
regime areequivalent instrumentsforimplementing theenvironmentalpolicy ifthedi erential
cost betweenhighandlowlevels ofaccidentpreventionactivitiesisrelativelysmallorrelatively
large, that is,if
2 ^ () 0; (p ` p h )(2 2F p h d) 1 p h [ 2 6 4 (1+)(2 2F p h d) K+ (1 p h )(1+ K) p ` p h ; 1 3 7 5 [ f (1+)(2 2F p h d) 1+ 1 p h p ` p h g:
The `regulator subjectto capture'regime is abetter instrumentifthe di erentialcost between
high and low levels of accident prevention activities is inthe lower intermediate range, thatis,
if 2 ^ () 0 B @ (p ` p h )(2 2F p h d) 1 p h ; (1+)(2 2F p h d) 1+ 1 p h p ` p h 1 C A :
The `extendedlenderliability'regime isabetterinstrumentifthedi erentialcostbetweenhigh
and lowlevels ofaccident prevention activitiesis inthehigherintermediaterange,that is,if
2 ^ () 0 B @ (1+)(2 2F p h d) 1+ 1 p h p ` p h ; (1+)(2 2F p h d) K+ (1 p h )(1+ K) p ` p h 1 C A : k
Proposition6: The regionin( ;)-space over whichthecapturedregulatorsolutionis better
thantheextendedlenderliabilitysolutionisindependentofK whiletheregionin( ;)-space
over whichthe captured regulator solutionis worse thanthe extended lender liabilitysolution
5 Conclusion and policy implications
We analyzedinthispaperasimplemodelallowinga comparisonbetweendi erentinstruments
for implementing an eÆcient environmental protection policy. More precisely,we considered a
moral hazard context in which rms can take preventing actions to reduce the probability of
environmental disasters. Those actions being a private informationof the rm concerned will
giverisetoinformationalrentswhosenetsocialcostispositivebecauseoftheexistenceofasocial
cost of publicfunds. We rst characterized theoptimal rulesa benevolent welfare maximizing
regulator would choose regarding the level of care activities a given rm should be exercising
and the condition for nancing the rm. Those rules di er from the rst best rules because
the maximizationof socialwelfarerequires thatthe regulatorminimizesthe informationalrent
of the rm. We thencompared the benevolent regulatorsolution rulesto those rulesa private
bank would applyunder the extended lenderliabilityof the CERCLA type and to those rules
a captured regulator, overestimating the contribution of the informational rent of the rm to
socialwelfare, wouldchoose.
The comparison of thethree sets of rules led usto identifythe region inthe ( ;)-space
wherethecapturedregulatorrulesandtheCERCLA-liableprivatebankrulesareequivalentto
thebenevolentregulatorrules,theregionwherethecapturedregulatorrulesarebetterinterms
ofsocialwelfareattainedtotheCERCLA-liableprivatebankrules,and nallytheregionwhere
the CERCLA-liable privatebankrulesare better.
Our main results are summarized in proposition 4 for the case where the rm is socially
pro table when thelevelof care ishighorlow and inproposition5forthe casewherethe rm
is sociallypro table onlyifthelevel ofcare ishigh. Ingeneral, thecapturedregulatorsolution
is better if the deterministic characteristic location of the economy in the ( ;)-space falls
in the lower intermediate range, that is if 2 () [or 2 ^
()] and theCERCLA-liable
privatebank solutionis better if the deterministic characteristic location of the economy falls
in thehigherintermediate range,that isif 2 ()[or 2 ^
()]. It is interesting to note
that the former region is independent of the capture parameter K while the latter region is
boundaryof the regionin ( ;)-space wherethe CERCLA-liable privatebank solutionrules
arebetterthanthecapturedregulator'srules[condition(24)inproposition4andcondition(25)
in proposition5]moves up.
Themainconclusionofthispaperisthereforethatchoosingbetweenaregulationframework
and a legal framework to implementan environmentalprotection policy is notan easy matter.
Butouranalysisprovidessome preliminarystepsindeterminingawaytoanalyzesuchachoice.
So the answer to the question regarding which instrumentsshould be employed by the policy
makers isthatacasebycaseexaminationisrequired. But,someoftheimportantdeterminants
oftherelativeeÆciencywithwhichdi erentpolicyinstrumentsmaximizesocialwelfarefunction
References
[1] BianchiA.(1994), `TheHarmonizationof Laws on LiabilityforEnvironmentalDamagein
Europe: AnItalian Perspective',Journal of EnvironmentalLaw 6,21-42.
[2] Boyd J. and D. Ingberman (1997), `The Search of Deep Pocket: is "Extended Liability`
Expensive Liability?',Journal ofLaw, Economics, and Organization13,233-258.
[3] Boyer M. (1979), `Les e ets de la reglementation', Canadian Public Policy / Analyze de
Politiques4,469-474.
[4] Boyer M., Lewis T.R. and W.L. Liu (2000), `Setting Standards for Credible Compliance
and LawEnforcement', Canadian Journalof Economics,(forthcoming).
[5] Boyer M. and J.J. La ont (1996), `Environmental Protection, Producer Insolvency and
LenderLiability',inA.Xepapadeas(ed.),EconomicPolicyfortheEnvironmentandNatural
Resources, EdwardElgar, 1-29.
[6] Boyer M. and J.J. La ont (1997), `Environmental Risk and Bank Liability', European
Economic Review41,1427-1459.
[7] BoyerM. andJ.J.La ont(1999), `TowardaPoliticalTheoryoftheEmergence of
Environ-mentalIncentiveRegulation',RandJournalof Economics41, 137-157.
[8] Boyer M.,PorriniD.(2000),`ThePoliticalEconomyofLawversusRegulationin
Environ-mentalPolicy', mimeo,Universite de Montreal andUniversitadiMilano.
[9] BuchananJ.M.andG.Tullock(1975),`Polluters'Pro tsandPoliticalResponse',American
Economic Review65,976-978.
[10] Buchanan J.M. (1969), `External Economies, Corrective Taxes and Market Structure',
AmericanEconomic Review59, 174-177.
[12] ChaddC.M.,Satinoven T., BergesonL., NeumanR.W.and C.R.Briant (1991),`Avoiding
Liability for Hazardous Waste: RCRA, CERCLA, and Related Corporate Law Issues',
Corporate Practice Series57, Washington DC,Bureauof National A airs.
[13] Chiancone A.and D. Porrini(1998), Lezioni di Analisi Economica del Diritto, IIIed.,
Gi-appichelli.
[14] ColeD.H. andP.Z. Grossman (1999), `Whenis Command-and-ControlEÆcient?
Institu-tions, Technology, and the Competitive EÆciency of Alternative Regulatory Regimes for
EnvironmentalProtection',WisconsinLawReview, 887-938.
[15] CooterR.and T. Ulen (1999),Lawand Economics, IIIed.Addison Wesley.
[16] Cropper M.L. and W.E. Oates (1992), `Environmental Economics: A Survey', Journal of
Economic Literature XXX,1675-740.
[17] DionneG. and S.Spaeter(1998),`EnvironmentalRisks andExtended Liability: TheCase
of GreenTechnologies', mimeo,
Ecole desHEC (Montreal).
[18] GobertK.andM.Poitevin(1998),`EnvironmentalRisks: ShouldBanksbeLiable?',mimeo,
CIRANO,Universite de Montreal.
[19] Gobert K. (1999), `Responsabilite des creanciers en matiere environnementale', mimeo,
CIRANO.
[20] HahnR.W. (1990),`The Political Economyof EnvironmentalRegulation: Towards a
Uni-fyingApproach', PublicChoice65, 21-47.
[21] Henderson, D.A. (1997), `Congressional Reform of Lender and Fiduciary Liability under
CERCLAand RCRA:Is Fleet FactorsFinallyDead?',Banking LawJournal,210-218.
[22] HeyesA. (1996),`Lender Penalty forEnvironmental Damageand theEquilibriumCost of
Capital',Economica 63, 311-323.
[24] La ont J.J. (1995), `Regulation, Moral Hazard and Insurance for Environmental Risk',
Journal ofPublicEconomics 58,319-336.
[25] La ont,J.J.andD. Martimort(1999), `SeparationofRegulatorsAgainstCollusive
Behav-ior', RandJournalof Economics30,232-262.
[26] Landes W. and R. Posner (1984), `Tort Law as a Regulatory Regime for Catastrophic
PersonalInjuries', Journalof Legal Studies13,417-434.
[27] LandesW.andR.Posner(1987),TheEconomicStructureofTortLaw,HarvardUniversity
Press, Cambridge, 1987.
[28] Lewis T.R. (1996), `Protecting the Environment when Costs and Bene ts are Privately
Known',RandJournalof Economics27, 819-847.
[29] Menell P.S. (1991), `The Limitations of Legal Institutions for Addressing Environmental
Risks', Journalof Economic Perspectives5,93-113.
[30] Noll R. (1983), `The Political Foundations of Regulatory Policy', Journal of Institutional
and TheoreticalEconomics 139,377-404.
[31] Pitchford R. (1995), `How Liable Should a Lender Be?', American Economic Review 85,
1171-1186.
[32] Poli S.(1999), `ShapingtheEC Regime onLiabilityforEnvironmentalDamage: Progress
orDisillusionment?',EuropeanEnvironmentalLawReview11, 299-309.
[33] Posner R.(1998),Economic Analysis ofLaw, V ed., LittleBrown andCompany.
[34] Segerson K. (1996), `Issues in the Choice of Environmental Instruments', inJ.B. Braden,
H.Folmer,T.S.Ulen(eds), EnvironmentalPolicywithPoliticalandEconomicIntegration,
Brook eld: EdwardElgar.
[35] ShavellS.(1984),`LiabilityforHarmsversusRegulationofSafety',JournalofLegalStudies
[36] Shavell S. (1987), Economic Analysis of Accident Law, Harvard University Press,
Cam-bridge.
[37] Simons R.P. (1994), `The Consequences of the Overturning of the EPA Lender Liability
Rule', Journalof CommercialLending4, 43-47.
[38] Staton D.S.(1993),`EPA'sFinalRuleonLenderLiability: LendersBeware', TheBusiness
Lawyer49, 163-185.
[39] Strasser K.A. and D. Rodosevich (1993), `Seeing the Forest for the Trees in CERCLA
Liability',YaleJournal onRegulation10, 493-560.
[40] YoheG.W.(1976), `Polluter'sPro tsandPoliticalResponse: DirectControlVersusTaxes:
NOTES
*.WewanttothankClaireDomenget,PatrickGonzalez,YolandeHiriart,EjanMackaayandStephen
Shavellfortheircomments. Weremainofcoursesolelyresponsibleforthecontentofthisarticle.
1.SeeCropperand Oates(1992), Segerson(1996) and Lewis(1996).
2.SeeBuchanan and Tullock(1975), Yohe (1976), Boyer(1979), Noll(1983), Hahn(1990), La ont
(1995) and Boyer and La ont (1999).
3.CERCLA:ComprehensiveEnvironmentalResponse, Compensationand LiabilityAct1980, 1985,
1996.
4.This is the case in Buchanan's (1969) example of a polluting monopolist when the subsidies
required to correctthemonopolisticbehaviorare notavailable: Pigouvian taxesarethen
dom-inatedbya quota which implementsthe secondbest tax.
5.SeeCalabresi(1970), Landes and Posner (1984, 1987), Shavell(1987), Menell (1991) and Boyer,
Lewisand Liu(2000).
6.TheSuperfundenabledthegovernmenttobeginthecleanupofprioritysitesplacedontheNational
PriorityList(NPL) withmoney generatedprincipallyby taxeson crude oil,corporate income,
petrochemical feedstocks,and motorfuels.
7.`The term \owner or operator" does not include a person, who, without participating in the
management of a vessel or facility, holdsindiciaof ownership primarilyto protect his security
interestinthevesselorfacility' (42USAC., par. 9601, (21), 1988)
8.See Strasser and Rodosevitch (1993), Pitchford (1995), Heyes (1996), Boyer and La ont (1996,
1997), Boyd and Ingberman (1997), Dionne and Spaeter (1998), Gobert and Poitevin (1998)
and Gobert(1999).
9.15,Environmental Law Reports 20,994 (E.D.Pa1985).
10.632 F. Supp. 573 (d. Md. 1986).
11.901 F. 2d1550 (11th Circuit 1990)cert. denied, 498USA 1046 (1991).
12.910 F. 2d668 (9th Circuit1990).
13.SeeChaddetalii(1991).
14.National Oil and Hazardous Substances Pollution Contingency Plan; Lender Liability under
15.Publicker Indus., Inc. v. USA(In reCuyahoga Equip. Corp.), 980 F. 2d 110, 118-119 (2d Cir.
1992); Kelley ex rel. Mich. Natural Resources Comm'n v. Tisconia, 810 F. Supp. 901, 907
(W.D. Mich. 1993); Ashaland Oil, Inc. v. Sonford Prods. Corp., 810,F. Supp. 1057, 1060 (d.
Minn. 1993), and others.
16.Kelleyv. USEPA,15 F. 3d1100 (DCcircuit1994), cert. denied;American Bankers Assoc. and
Others v. Kelley, 115 S.Ct. 900 (1995).
17.SeeSimons(1994).
18.H11766SubtitleE,xx2501to2505-AssetConservation,LenderLiability,andDepositInsurance
Protection Actof 1996 (HR3610).
19.SeeHenderson(1997).
20.SeeKornhauserand Revesz (1994).
21.SeeStaton (1993).
22.Formore on thisissue,see A. Johnson & Co. v. Aetna Casualty & Sur. Co., 933 F.2d 66 (1st
Cir. 1991); USAFidelity &Guar. Co. v. George W. Whitesides Co.,932 F..2d 1169 (6thCir.
1991).
23.In 1993, the European Commission published the Green Paper on Remedying Environmental
Damage (COM(93)47 nal,Brussels,14may1993,OJ1993 C149/12). It presentedthebroad
concepts on which a liability system could be built and contained a description of the issues
relevant to designing a Community-wide liability system. It led to discussions on the future
EC liabilityregime througha debate aimed at collecting the opinionsof the interested sectors
and parties. It focused on the liability criteria, the de nition of environmental damage, the
right ofnon governmentalorganizationsto bringlegalactions(thelegalstandingdoctrine),the
insurabilityof environmentaldamage, the limitationsof liability,theproblemof reinstatement
of the environment, the possibility of compensation funds nanced by industries. In the same
year, the Commission explored the possibility of joining the 1993 Council of Europe Lugano
Convention, but a decision did not follow because of the intention to issue a speci c White
Paper and a Directive proposal. In November 1997, the Working Paper on Environmental
Liability outlined the key elements of a proposed Directive on environmental liability and in
