Standards and Incentives in Safety Regulation [Elektronische Ressource] / Felix Reinshagen. Betreuer: Klaus Schmidt

Felix Reinshagen

Sujets

• Management

Informations

• Publié par: ludwig-maximilians-universitat_munchen
• Publié le: 01 janvier 2011
• Nombre de lectures: 30
• Langue: Deutsch

Extrait

Standards and Incentives
in Safety Regulation
Inaugural-Dissertation
zurErlangungdesGrades
Doctoroeconomiaepublicae(Dr. oec. publ.)
anderLudwig-Maximilians-UniversitätMünchen
2010
vorgelegtvon
FelixReinshagen
Referent: Prof. Dr. KlausM.Schmidt
Korreferent: Prof. RayRees
Promotionsabschlussberatung: 1. Juni2011Acknowledgements
Firstandforemost,Iwouldliketothankmysupervisor,KlausM.Schmidt,for
the support and counsel that he has given me while writing this dissertation.
IamalsoverythankfultoRayReesandSvenRady,whoencouragedme–espe-
ciallyduringthedifficultstart–andagreedtoactassecondandthirdexaminer.
Ihavereceivedvaluablefeedbackanddirectionfromanumberofpeople.Among
others,IwouldliketomentionJenniferArlen,BennoBühler,MatthiasDischinger,
FlorianEnglmaier,GeorgGebhardt,DanielGöller,SandraLudwig,MartinSchnei-
der,CasparSiegert,AnnoStolper,ChristinaStrassmairandRobertUlbricht.
Thanksgotomynumerousofficemates,who–atsometimeoranother–hap-
pily shared oneof thelargenumberof offices that I occupied duringmy doc-
toralcareer,namelySilviaAppelt,JoachimKlein,NicolasKlein,JohannesMaier,
JanSchikora,CasparSiegertandAnnoStolper.
IamalsogratefultotheDepartmentofEconomicsforadmittingmetothedoc-
toralprogram,despitemyunusualbackgroundasalawyer. Ireceivedfinancial
supportfromtheDeutscheForschungsgemeinschaft(DFG) throughGRK801,
whichIgratefullyacknowledge.
Manythanksgotomyfamily,whosupportedmeduringmylongeducation,but
especiallytomylategrandmother,IrmgardGauss,whogenerously“toppedup”
my scholarship grant. Finally, I wish to thank ChristinaStrassmair– her love,
supportandadvicewasinvaluableduringthelastyears.
2Contents
Preface 6
1 StandardsandIncentivesunderMoralHazardwithLimitedLiability 11
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2 SetupoftheModel . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3 BenchmarkCase: Incentivesonly . . . . . . . . . . . . . . . . . . . 16
1.4 JointUseofIncentivesandStandards . . . . . . . . . . . . . . . . . 18
1.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.6 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2 RegulationandLiabilityasComplementsandSubstitutes 28
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.2 SetupoftheModel . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.3 TheFirst-BestCareLevels . . . . . . . . . . . . . . . . . . . . . . . . 34
2.4 CareLevelswithLiabilityonly . . . . . . . . . . . . . . . . . . . . . 35
2.5 TheOptimalCombinationofLiabilityandRegulation . . . . . . . 37
2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.7 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3 MinimumEducationRequirementsforProfessions 47
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.2 ObjectiveFunctionoftheAgent . . . . . . . . . . . . . . . . . . . . 51
3.3 MarketEquilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3Acknowledgements 4
3.4 RegulatoryInterventions . . . . . . . . . . . . . . . . . . . . . . . . 56
3.5 WelfareAnalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.5.1 AchievingtheFirst-Best . . . . . . . . . . . . . . . . . . . . . 60
3.5.2 ComparisonsintheSecondBest . . . . . . . . . . . . . . . . 62
3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.7 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.7.1 ProofofProposition3.6 . . . . . . . . . . . . . . . . . . . . . 67
3.7.2 Derivationofequation3.5.2 . . . . . . . . . . . . . . . . . . . 69
a3.7.3 Examplewithcostfunctionc(k)= . . . . . . . . . . . . . . 69k
Bibliography 71List of Figures
2.1 ObservableEffortwithSubstitutes . . . . . . . . . . . . . . . . . . . 40
2.2 ObservableEffortwithComplements . . . . . . . . . . . . . . . . . 41
3.1 Timingofthemodel . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.2 Payoffswithoutregulation . . . . . . . . . . . . . . . . . . . . . . . . 57
3.3 Payoffswithminimumhumancapital . . . . . . . . . . . . . . . . . 59
3.4 Payoffswithminimumhumancapital(notcostless) . . . . . . . . 62
5Preface
Theoptimalregulationofsafetyisapopulartopic,bothforacademicsandfor
thewider public. On theone hand, technologicalprogress has multipliedthe
destructivepotentialofhumanaction. Thedebatesaboutnucleartechnology
or genetically modified organisms give daily proof to this statement. On the
otherhand,thesameprogresshasincreasedhumancapabilitytoguardagainst
dangerswhilerisingincomeshaveincreasedthedemandforsafety. Whatfor-
mer generations might have mourned as a “stroke of fate” will today be de-
nouncedasa“preventabletragedy”.
The three chapters in this dissertation all deal with the optimal regulation of
safety. By safetywe meanthepreventionofaccidents, i.e. eventswhichcause
significant harm but occur with a low probability. The last criterion distin-
guishesaccidentsfrombadoutcomesthatareconsidered“normal”.Forexam-
ple, the failure of a company to produce an important innovation will not be
considered as an accident. Still, this definition of accidents might be broader
thantheordinaryuseoftheword.Forexample,considerthecaseofaphysician
whofailstoperformanimportanttest.Thisfailuremightcauseabadoutcome
for the patient, albeit with a low a probability. If the bad outcomeis realised,
ourdefinitionofanaccidentisfulfilled.
Consideranagentwhoseactivitymightleadtoanaccident. Thereareusually
many actions that the agent can take to lower the probability of an accident.
According to the usual criterion of cost-benefit analysis, an action should be
takenifthebenefitsofthisaction–thedecreaseinexpecteddamagefromac-
cidents–aregreaterthanoratleastequaltothecostsoftheaction.
6Preface 7
If theperformanceof theseactionscan be observedby a regulatoror a court,
the optimal regulation of safety is straightforward. All actions that pass the
cost-benefit test should be prescribed by law or regulation. We will call this
approach “standard setting”. But in many cases, the observation of these ac-
tions will be impossibleor at least very expensive. This means that a strategy
of“standardsetting”willbeinfeasibleoratleastveryexpensive. Insuchasit-
uation theregulatorcan stillimplementall preventiveactions which pass the
cost-benefittestifheisabletoimposeontheagentallcosts thatarisefroma
potential accident. This strategy, the use of “incentives”, makes the agent the
“residual claimant” on the risk. It will induce the agent to voluntarily imple-
ment all actions that pass the cost-benefit test. The legal institutionof “strict
liability”,eitheronastatutoryoracontractualbasis,canbeseenasanattempt
tomaketheagenttheresidualclaimant.
In practiceit is often very difficultto shift all potentialcosts of an accident to
theagent.Thefinancialassetsoftheagentmightbefarbelowtheamountthat
is necessary to pay for all accidental damages in theworst case. It might also
be very difficult to find all victims of an accident and to accurately estimate
thedamagestheyhavesuffered. Finally,theagentmightbeabletoescapethe
reach of the law. If the liability of the agent is limited in such a way, he is no
longerinducedtotakeallpreventiveactivitiesthatpassthecost-benefittest.
Soinpracticeneitherstandardsettingnorincentiveswillworkperfectly. Stan-
dard setting can only affect actions which are easily observable, while incen-
tivescanshiftonlypartsofthecostsofaccidentstotheagent.Wewilltherefore
investigatetheoptimalcombinationofthesetwoapproaches. Wewillassume
that preventive actions can be separated into two distinct categories, observ-
able actions and unobservable actions. Examples of actions that we think of
as observable are conformance with technical norms in the constructionof a
plant,orwhethera physicianhasaccess to adequateequipment. Anotherex-
ample of an observableaction is theacquisition of human capital by the per-
sonswho willcontroltherisk,e.g.whetherhehasarelevantacademicdegree
or has passeda prescribedexamination. Examplesof actionsthatwe thinkofPreface 8
asunobservablearethementalalertnessofthepersonwhocontrolstheriskor
whethersafetyproceduresareactuallyfollowedinday-to-daywork.
Given thisdistinction,itwill bea naturaloutcomeofourmodelsthatmostof
the time it is optimal to use standards and incentives at the same time. Our
main interest will be the optimal level of standard setting in such a situation.
Inthefollowingchaptersweshowthatinadequateincentivesforunobservable
actionswillalsoinfluencetheoptimallevelofstandardsforobservableactions.
Often this optimal level will differ from the optimaluse of observable actions
in a first-best world. This has the consequence that a cost-benefit analysis of
standards that ignores the interaction between observable and unobservable
preventiveactionsmightbeflawed.
Thefirstchapterisanapplicationofthisideatocontracttheory.Weconsidera
modelofmoralhazardwithlimitedliabilityoftheagentandeffortthatistwo-
dimensional.Onedimensionoftheagent’seffortisobservableandtheotheris
not.Theprincipalcanthusmakethecontractconditionalnotonlyonoutcome
but also on observable effort. In this chapter we make the assumption that
there is no interaction between the cost