清管學(xué)院微觀經(jīng)濟(jì)學(xué)講義下tsinghua2005mba lecture15markets with asymmetric informat

Lecture15MarketswithAsymmetricInformationLecture15Slide2TopicstobeDiscussedQualityUncertaintyandtheMarketforLemonsMarketSignalingMoralHazardThePrincipal-AgentProblemLecture15Slide3TopicstobeDiscussedManagerialIncentivesinanIntegratedFirmAsymmetricInformationinLaborMarkets:EfficiencyWageTheoryLecture15Slide4IntroductionWewillstudyhowimperfectinformationinfluencesresourceallocationandthepricesystem.Lecture15Slide5

Example:

“Lemons”ProblemInJune,1999,IwasavisitingscholarinUniversityofSydney.Iboughtaheateratthepriceof$40.WhenIleavetheSydney,Iwanttoselltheheatertothesecond-handstore.Canyouguesshowmuchthestorewanttopaymefortheheater?(note:Theheaterisgoodinquality)

Lecture15Slide6Example:

“Lemons”ProblemThestoreonlywantedtopaymeonly$2fortheheater.Why?Lecture15Slide7Explanationsforthelemon

Alemonsproblemarisesduetoasymmetricinformationaboutproductquality.Forexample,intheusedheatermarketbuyersdeterminethepricestheyarewillingtopayaccordingtotheirbeliefregardingtheaveragequalityoftheheatersofferedforsale.Sellers,incontrast,knowthequalityoftheirheatersandbasetheaskingpricesonthequalityoftheindividualheaterstobesold.Inthisframework,thesellerofthehighestqualityheaterwillneverbeabletogetapricewhichproperlyreflectsthatquality.Asaresult,thissellerwillnotofferherheaterforsaleinthemarket.Butnow,thesellerofthesecondhighestqualityheaterwillnotbeabletogetapricethatreflectsthequalityoftheheatersheisoffering,andwillalsowithdrawitfromthemarket.Thisprocessmaycontinueuntilonlythelowestqualityheatersareofferedinthemarket.Lecture15Slide8QualityUncertainty

andtheMarketforLemonsThelackofcompleteinformationwhenpurchasingausedcarincreasestheriskofthepurchaseandlowersthevalueofthecar.Lecture15Slide9TheMarketforUsedCarsAssumeBuyersandsellerscandistinguishbetweenhighandlowqualitycarsTherewillbetwomarketsQualityUncertainty

andtheMarketforLemonsTheLemonsProblemPHPLQHQLSHSLDHDL5,00050,00050,000Themarketforhighandlowqualitycarswhenbuyersandsellerscanidentifyeachcar10,000DLDMDM75,00025,000Withasymmetricinformationbuyerswillfinditdifficulttodeterminequality.Theylowertheirexpectationsoftheaveragequalityofusedcars.DemandforlowandhighqualityusedcarsshiftstoDM.DLMDLMTheincreaseinQLreducesexpectationsanddemandtoDLM.Theadjustmentprocesscontinuesuntildemand=DL.

Lecture15Slide11TheMarketforUsedCarsWithasymmetricinformation:Lowqualitygoodsdrivehighqualitygoodsoutofthemarket.Themarkethasfailedtoproducemutuallybeneficialtrade.Toomanylowandtoofewhighqualitycarsareonthemarket.Adverseselectionoccurs;theonlycarsonthemarketwillbelowqualitycars.QualityUncertainty

andtheMarketforLemonsLecture15Slide12ExampleThereare100usedcarsavailableforsaleinthetownofManchester.Theownersofthelow-qualityusedcarsvaluetheircarsat$4,000,whilethehigh-qualitycarownersvaluetheircarsat$8,000.Thereare100potentialbuyerswhovaluelow-qualitycarsat$5,000andhigh-qualitycarsat$10,000.Thepriceabuyeriswillingtopayisafunctionoftheaveragequalityofacartheyexpectedtobeofferedonthemarket:theyhavenoinformationoncarquality,andthereforebuyersexpecttheretobea50percentchancethatacarisalemon(low-quality).Wecanconcludethat:a)Buyersarewillingtopay7,500,andonlylow-qualitycarswillbeofferedforsale;b)Buyersarewillingtopay7,500,andbothtypesofcarswillbeofferedforsale;c)Buyersarewillingtopay5,000,andonlylow-qualitycarswillbeofferedforsale;d)Buyersarewillingtopay10,000,andbothtypesofcarswillbeofferedforsale;e)Buyersarewillingtopay10,000,andonlyhigh-qualitycarswillbeofferedforsale;Lecture15Slide13ImplicationsofAsymmetricInformationMedicalInsuranceQuestionIsitpossibleforinsurancecompaniestoseparatehighandlowriskpolicyholders?Ifnot,onlyhighriskpeoplewillpurchaseinsurance.Adverseselectionwouldmakemedicalinsuranceunprofitable.TheMarketforInsuranceLecture15Slide14ImplicationsofAsymmetricInformationAutomobileInsuranceQuestionsWhatimpactdoesasymmetricinformationandadverseselectionhaveoninsuranceratesandthedeliveryofautomobileaccidentinsurance?Howcanthegovernmentreducetheimpactofadverseselectionintheinsuranceindustry?TheMarketforInsuranceLecture15Slide15ImplicationsofAsymmetricInformationTheMarketforCreditAsymmetricinformationcreatesthepotentialthatonlyhighriskborrowerswillseekloans.QuestionHowcancredithistorieshelpmakethismarketmoreefficientandreducethecostofcredit?Lecture15Slide16ImplicationsofAsymmetricInformationTheImportanceofReputationandStandardizationAsymmetricInformationandDailyMarketDecisionsRetailsalesAntiques,art,rarecoinsHomerepairsRestaurantsLecture15Slide17ImplicationsofAsymmetricInformationQuestionHowcantheseproducersprovidehigh-qualitygoodswhenasymmetricinformationwilldriveouthigh-qualitygoodsthroughadverseselection.AnswerReputationLecture15Slide18ImplicationsofAsymmetricInformationQuestionWhydoyoulookforwardtoaBigMacwhentravelingeventhoughyouwouldneverconsiderbuyingoneathome.HolidayInnonceadvertised“NoSurprises”toaddresstheissueofadverseselection.Lecture15Slide19LemonsinMajorLeagueBaseballAsymmetricinformationandthemarketforfreeagentsIfalemonsmarketexists,freeagentsshouldbelessreliable(disabled)thanrenewedcontracts.Lecture15Slide20PlayerDisabilityAllPlayers 4.73 12.55 165.4Renewedplayers 4.76 9.68 103.4Freeagents 4.67 17.23 268.9DaysSpentonDisabledListperSeasonPrecontract Postcontract PercentageChangeLecture15Slide21FindingsDaysonthedisabledlistincreaseforbothfreeagentsandrenewedplayers.Freeagentshaveasignificantlyhigherdisabilityratethanrenewedplayers.Thisindicatesalemonsmarket.LemonsinMajorLeagueBaseballLecture15Slide22QuestionIfyouareateamowner,whatstepswouldyoutaketoreducetheasymmetricinformationforfreeagents?LemonsinMajorLeagueBaseballLecture15Slide23MarketSignalingTheprocessofsellersusingsignalstoconveyinformationtobuyersabouttheproduct’squalityhelpsbuyersandsellersdealwithasymmetricinformation.Lecture15Slide24MarketSignalingStrongSignalTobeeffective,asignalmustbeeasierforhighqualitysellerstogivethanlowqualitysellers.ExampleHighlyproductiveworkerssignalwitheducationalattainmentlevel.Lecture15Slide25MarketSignalingASimpleModelofJobMarketSignalingAssumeTwogroupsofworkersGroupI:Lowproductivity--AP&MP=1GroupII:Highproductivity--AP&MP=2TheworkersareequallydividedbetweenGroupIandGroupII--APforallworkers=1.5Lecture15Slide26MarketSignalingASimpleModelofJobMarketSignalingAssumeCompetitiveProductMarketP=$10,000Employeesaverage10yearsofemploymentGroupIRevenue=$100,000(10,000/yr.x10)GroupIIRevenue=$200,000(20,000/yr.X10)Lecture15Slide27MarketSignalingWithCompleteInformationw=MRPGroupIwage=$10,000/yr.GroupIIwage=$20,000/yr.WithAsymmetricInformationw=averageproductivityGroupI&IIwage=$15,000Lecture15Slide28MarketSignalingSignalingWithEducationtoReduceAsymmetricInformationy=educationindex(yearsofhighereducation)C=costofattainingeducationallevelyGroupI--CI(y)=$40,000yGroupII--CII(y)=$20,000yLecture15Slide29MarketSignalingSignalingWithEducationtoReduceAsymmetricInformationAssumeeducationdoesnotincreaseproductivityDecisionRule:y*signalsGIIandwage=$20,000Belowy*signalsGIandwage=$10,000SignalingYearsofCollegeValueofCollegeEduc.0$100KValueofCollegeEduc.YearsofCollege1234560123456$200K$100K$200KGroupIGroupIICI(y)=$40,000yOptimalchoiceofyforGroupIHowmucheducationshouldapersonobtain?Theeducationdecisionisbasedonbenefits/costcomparison.B(y)B(y)y*y*B(y)=increaseinwageassociatedwitheachlevelofeducationCII(y)=$20,000yOptimalchoiceofyforGroupISignalingYearsofCollegeValueofCollegeEduc.0$100KValueofCollegeEduc.YearsofCollege1234560123456$200K$100K$200KCI(y)=$40,000yOptimalchoiceofyforGroupIB(y)B(y)y*y*Benefits=$100,000CostCI(y)=40,000y$100,000<$40,000y*y*>2.5ChoosenoeducationCII(y)=$20,000yOptimalchoiceofyforGroupIBenefits=$100,000CostCII(yO)=20,000y$100,000<$20,000y*y*<5Choosey*Lecture15Slide32SignalingCost/BenefitComparisonDecisionruleworksify*isbetween2.5and5Ify*=4GroupIwouldchoosenoschoolGroupIIwouldchoosey*RulediscriminatescorrectlyLecture15Slide33SignalingEducationdoesincreaseproductivityandprovidesausefulsignalaboutindividualworkhabits.Lecture15Slide34WorkingintotheNightQuestionHowcanyousignaltoyouremployeryouaremoreproductive?Lecture15Slide35MarketSignalingGuaranteesandWarrantiesSignalingtoidentifyhighqualityanddependabilityEffectivedecisiontoolbecausethecostofwarrantiestolow-qualityproducersistoohighLecture15Slide36Example

Iffinancialaidwereexpandedandmorestudentsattendedcollegeasaresult,wouldyouexpectthatmorestudentswouldgetMaster’sofBusinessAdministrationdegrees?Why?Lecture15Slide37MoralHazardMoralhazardoccurswhentheinsuredpartywhoseactionsareunobservedcanaffecttheprobabilityormagnitudeofapaymentassociatedwithanevent.Lecture15Slide38MoralHazardDeterminingthePremiumforFireInsuranceWarehouseworth$100,000Probabilityofafire:.005witha$50firepreventionprogram.01withouttheprogramLecture15Slide39MoralHazardDeterminingthePremiumforFireInsuranceWiththeprogramthepremiumis:.005x$100,000=$500Onceinsuredownerspurchasetheinsurance,theownersnolongerhaveanincentivetoruntheprogram,thereforetheprobabilityoflossis.01$500premiumwillleadtoalossbecausetheexpectedlossis$1,000(.01x$100,000)Lecture15Slide40TheEffectsofMoralHazardMilesperWeek0$0.5050100140CostperMile$1.00$1.50$2.00D=MBMC’Withmoralhazardinsurancecompaniescannotmeasuremileage.MCto$1.00andmilesdrivenincreasesto140miles/week--inefficientallocation.MCMCisthemarginalcostofdriving.WithnomoralhazardandassuminginsurancecompaniescanmeasuremilesdrivenMC=MBat$1.50and100miles/week--efficientallocation.Lecture15Slide41ReducingMoralHazard

--WarrantiesofAnimalHealthScenarioLivestockbuyerswantdiseasefreeanimals.AsymmetricinformationexistsManystatesrequirewarrantiesBuyersandsellersnolongerhaveanincentivetoreducedisease(moralhazard).QuestionHowcanthismoralhazardbereduced?Lecture15Slide42CrisisintheSavingsandLoanIndustryQuestionHowmanypeopleknowthefinancialstrengthoftheirbank?Whynot?Depositinsurance,moralhazard,andfailuresintheS&LindustryLecture15Slide43CostoftheS&LBailout1,000+failedinstitutions$200billion(1990)Texasalone--$42billion(1990)Agencyexpenditures--$100million(1990)QuestionHowcanthismoralhazardbereduced?CrisisintheSavingsandLoanIndustryLecture15Slide44ExampleEventhoughJoehasanexcellentcomprehensivehealthinsurancepolicy,hehasnowishtoriskaheartattack.Hecontinuestowatchhisdietandexerciseregularlytoreducetheriskofheartdiseaseeventhoughheisinsures.Isthereanyelementofmoralhazardthathishealthinsurancecompanywouldstillneedtobeconcernedabout?Lecture15Slide45ThePrincipal--AgentProblemAgencyRelationshipOneperson’swelfaredependsonwhatanotherpersondoesAgentPersonwhoactsPrincipal PersonwhomtheactioneffectsLecture15Slide46ThePrincipal--AgentProblemCompanyownersareprincipals.Workersandmanagersareagents.Ownersdonothavecompleteknowledge.Employeesmaypursuetheirowngoalsandreduceprofits.Lecture15Slide47案例分析:承包基數(shù)的確定問(wèn)題：委托人應(yīng)當(dāng)如何采取對(duì)策，防止代理人利用自己的職權(quán)侵犯委托人的利益？改革開放以來(lái)國(guó)有資產(chǎn)的大量流失，說(shuō)明了研究這一問(wèn)題的重要性和緊迫性。委托代理問(wèn)題的核心是基數(shù)問(wèn)題的確定。而且這一過(guò)程雙方所擁有的信息是不對(duì)稱的，代理人明顯擁有比委托人更多的信息?；鶖?shù)的確定過(guò)程中的一個(gè)難題是：委托人總想提高基數(shù)，而代理人總想降低基數(shù)，因此雙方僵持不下!Lecture15Slide48案例分析:承包基數(shù)的確定

騎馬比慢？還是跳起來(lái)摘蘋果？

決策思想：在信息不對(duì)稱的情況下，給代理人以決定基數(shù)的很大的發(fā)言權(quán)，（這樣就避免了代理人日后完不成基數(shù)把責(zé)任推給委托人這樣一種常見的“基數(shù)軟約束”的情況）。與此同時(shí)，對(duì)代理人濫報(bào)基數(shù)的行為進(jìn)行制約，利用利益導(dǎo)向的行為引導(dǎo)代理人報(bào)出一個(gè)他通過(guò)努力能夠完成的最高基數(shù)。這樣，就使基數(shù)確定達(dá)到了最佳的狀態(tài)。

Lecture15Slide49具體講：1、委托人與代理人共同確定基數(shù)，最終合同基數(shù)C（Contract)是委托人要求數(shù)D(Demand)與代理人自報(bào)數(shù)S(Self-offered)的加權(quán)平均。在實(shí)際應(yīng)用中，可取權(quán)數(shù)為雙方50％，即算術(shù)平均。合同基數(shù)C全部歸委托人，也即“基數(shù)全交”。2、當(dāng)期末實(shí)際數(shù)A（Actual)超過(guò)合同基數(shù)C時(shí)，超過(guò)部分Q1（％）將作為獎(jiǎng)勵(lì)歸代理人，若案例分析:承包基數(shù)的確定Lecture15Slide50案例分析:承包基數(shù)的確定A<C，則不足部分的一個(gè)同樣的比例將由代理人補(bǔ)足。不妨取Q1＝80％。3、當(dāng)期末實(shí)際數(shù)A超過(guò)代理人自報(bào)數(shù)S時(shí)，將對(duì)代理人收取“少報(bào)罰金”（因?yàn)镾較小導(dǎo)致了加權(quán)平均后的合同基數(shù)C=(D+S)/2也較小，從而使委托人收益減少）。少報(bào)罰金數(shù)量為（S-A)Q2,這里Q2

稱為少報(bào)懲罰系數(shù)，不妨取Q2＝60％。Lecture15Slide51案例分析:承包基數(shù)的確定可以證明：如果委托人采用上述三條對(duì)策，那么，他在確定基數(shù)時(shí)只要提出一個(gè)保底數(shù)就夠了，而代理人則會(huì)自動(dòng)報(bào)出一個(gè)通過(guò)他努力能夠完成的最大基數(shù)。Lecture15Slide52案例分析:承包基數(shù)的確定代理人自報(bào)的5種情況一二三四五(1）代理人自報(bào)數(shù)S60708090100(2)委托人要求數(shù)D6060606060(3)合同確定數(shù)C=(S+D)/26065707580(4)代理人實(shí)際能力（期末完成數(shù)）A8080808080(5)超基數(shù)獎(jiǎng)勵(lì)（A-C）×0.81612840(6)少報(bào)罰金（S-A）×0.6-12-6000(7)代理人凈收益=(5)+(6)46840Lecture15Slide53案例分析:承包基數(shù)的確定分析：從表中可知：在第三種情況，代理人自報(bào)數(shù)為80萬(wàn)，委托人要求數(shù)為60萬(wàn)，合同數(shù)為70萬(wàn)，但代理人實(shí)際完成數(shù)為80萬(wàn)，故超額10萬(wàn)，代理人可得超基數(shù)獎(jiǎng)勵(lì)10×0.8＝8萬(wàn)，又由于在第三種情況，代理人的自報(bào)數(shù)S＝80萬(wàn)，其實(shí)際完成數(shù)A也是80萬(wàn)，也就是代理人沒(méi)有少報(bào)，因此罰金為0。這樣代理人的凈收益為8萬(wàn)元。原則：只有當(dāng)代理人使其自報(bào)數(shù)S等于其實(shí)際能力數(shù)A時(shí)，也即實(shí)事求是時(shí)，其凈收益才為最大。Lecture15Slide54ThePrincipal--AgentProblemThePrincipal--AgentProbleminPrivateEnterprisesOnly16of100largestcorporationshaveindividualfamilyorfinancialinstitutionownershipexceeding10%.Mostlargefirmsarecontrolledbymanagement.Monitoringmanagementiscostly(asymmetricinformation).Lecture15Slide55ThePrincipal--AgentProblemThePrincipal--AgentProbleminPrivateEnterprisesManagersmaypursuetheirownobjectives.GrowthUtilityfromjobLecture15Slide56ThePrincipal--AgentProblemThePrincipal--AgentProbleminPrivateEnterprisesLimitationstomanagers’abilitytodeviatefromobjectiveofownersStockholderscanoustmanagersTakeoverattemptsMarketformanagerswhomaximizeprofitsLecture15Slide57ThePrincipal--AgentProblemThePrincipal--AgentProbleminPublicEnterprisesObservationsManagers’goalsmaydeviatefromtheagenciesgoal(size)Oversightisdifficult(asymmetricinformation)MarketforcesarelackingLecture15Slide58ThePrincipal--AgentProblemThePrincipal--AgentProbleminPublicEnterprisesLimitationstoManagementPowerManagerschooseapublicservicepositionManagerialjobmarketLegislativeandagencyoversight(GAO&OMB)CompetitionamongagenciesLecture15Slide59TheManagersof

NonprofitHospitalsasAgentsArenonprofitorganizationsmoreorlessefficientthatfor-profitfirms?725hospitalsfrom14hospitalchainsReturnoninvestment(ROI)andaveragecost(AC)measuredLecture15Slide60For-Profit 11.6% 12.7%Nonprofit 8.8% 7.4%ReturnOnInvestment1977 1981TheManagersof

NonprofitHospitalsasAgentsLecture15Slide61Afteradjustingfordifferencesinservices:AC/patientdayinnonprofitsis8%greaterthanprofitsConclusionProfitincentiveimpactsperformanceCostandbenefitsofsubsidizingnonprofitsmustbeconsidered.TheManagersof

NonprofitHospitalsasAgentsLecture15Slide62ExampleThedemandforautomobilesishighlycyclical,goingupwhentheeconomyisdoingwell,andgoingdownwhentheeconomyisinarecession.Theownersoftheautomobilecompaniesrealizethis,andhavetoconsiderdifferentpaymentschemesknowingthattheprofitseachyearforfirmwilldependonthestrengthoftheeconomyandtheperformanceoftheirmanagers.Whichofthefollowingpaymentschemeswouldyoumendtotheownersandwhy?1).Paythemanagersaflatsalarythatisnottiedtofirm’sperformance.2).Paythemanagersashareoftheprofits.3).Paythemanagersaflatsalaryplusabonusthatistiedtothefirm’sperformance.Lecture15Slide63IncentivesinthePrincipal-AgentFrameworkDesigningarewardsystemtoaligntheprincipalandagent’sgoals--anexampleWatchmanufacturerUseslaborandmachineryOwnersgoalistomaximizeprofitMachinerepairpersoncaninfluencereliabilityofmachinesandprofitsThePrincipal--AgentProblemLecture15Slide64ThePrincipal--AgentProblemIncentivesinthePrincipal-AgentFrameworkDesigningarewardsystemtoaligntheprincipalandagent’sgoals--anexampleRevenuealsodepends,inpart,onthequalityofpartsandthereliabilityoflabor.Highmonitoringcostmakesitdifficulttoassesstherepair-person’sworkLecture15Slide65TheRevenuefromMakingWatchesLoweffort(a=0) $10,000 $20,000Higheffort(a=1) $20,000 $40,000PoorLuck GoodLuckLecture15Slide66ThePrincipal--AgentProblemIncentivesinthePrincipal-AgentFrameworkDesigningarewardsystemtoaligntheprincipalandagent’sgoals--anexampleRepairpersoncanworkwitheitherhighorloweffortRevenuesdependoneffortrelativetotheotherevents(poororgoodluck)Ownerscannotdetermineahighorloweffortwhenrevenue=$20,000Lecture15Slide67ThePrincipal--AgentProblemIncentivesinthePrincipal-AgentFrameworkDesigningarewardsystemtoaligntheprincipalandagent’sgoals--anexampleRepairperson’sgoalistomaximizewagenetofcostCost=0forloweffortCost=$10,000forhigheffortw(R)=repairpersonwagebasedonlyonoutputLecture15Slide68ThePrincipal--AgentProblemIncentivesinthePrincipal-AgentFrameworkChoosingaWagew=0;a=0;R=$15,000R=$10,000or$20,000,w=0R=$40,000;w=$24,000R=$30,000;Profit=$18,000Netwage=$2,000Lecture15Slide69ThePrincipal--AgentProblemIncentivesinthePrincipal-AgentFrameworkChoosingaWagew=R-$18,000Netwage=$2,000HigheffortLecture15Slide70ThePrincipal--AgentProblemConclusionIncentivestructurethatrewardstheeofhighlevelsofeffortcaninduceagentstoaimforthegoalssetbytheprincipals.Lecture15Slide71ThePrincipal--AgentProblemAsymmetricInformationandIncentiveDesignintheIntegratedFirmInintegratedfirms,divisionmanagershavebetter(asymmetric)informationaboutproductionthancentralmanagementLecture15Slide72ThePrincipal--AgentProblemAsymmetricInformationandIncentiveDesignintheIntegratedFirmTwoIssuesHowcancentralmanagementillicitaccurateinformationHowcancentralmanagementachieveefficientdivisionalproductionLecture15Slide73ThePrincipal--AgentProblemPossibleIncentivePlansBonusbasedonoutputorprofitWillthisplanprovideanincentiveforaccurateinformation?Lecture15Slide74ThePrincipal--AgentProblemPossibleIncentivePlansBonusbasedonhowclosethemanagersgettotheirforecastsofoutputandprofitsQf=estimateoffeasibleproductionlevelB=bonusindollarsQ=actualoutputB=10,000-.5(Qf-Q)IncentivetounderestimateQfLecture15Slide75ThePrincipal--AgentProblemPossibleIncentivePlansBonusstilltiedtoaccuracyofforecastIfQ>Qf

;B=.3Qf+.2(Q-Qf)IfQ<Qf;B=.3Qf-.5(Qf-Q)Lecture15Slide76IncentiveDesigninanIntegratedFirmOutput(unitsperyear)2,0004,0006,00010,000010,00020,00030,00040,000Bonus($peryear)8,000IfQf=30,000,bonusis$6,000,themaximumamountpossible.Qf=30,000Qf=10,000IfQf=10,000,bonusis$5,000Qf=2