Asset Pricing John H. Cochrane June 12, 2000 1 Acknowledgments ThisbookowesanenormousintellectualdebttoLarsHansenandGeneFama. Mostofthe ideas in the book developed from long discussions with each of them, and trying to make senseofwhateachwassayinginthelanguageoftheother. Iamalsogratefultoallmycol- leaguesinFinanceandEconomicsattheUniversityofChicago,andtoGeorgeConstantinides especially,formanydiscussionsabouttheideasinthisbook.IthankGeorgeConstantinides, Andrea Eisfeldt, Gene Fama, Wayne Ferson, Owen Lamont, Anthony Lynch, Dan Nelson, Alberto Pozzolo, Michael Roberts, Juha Seppala, Mike Stutzer, PietroVeronesi, an anony- mous reviewer, and several generations of Ph.D. students at the University of Chicago for manyusefulcomments. I thanktheNSFandtheGraduateSchoolofBusinessfor research support. Additionalmaterialandbothsubstantiveandtypographicalcorrectionswillbemaintained at http://www-gsb.uchicago.edu/fac/john.cochrane/research/papers Comments and suggestions are most welcome This book draft is copyright c John H. ° Cochrane1997,1998,1999,2000 JohnH.Cochrane GraduateSchoolofBusiness UniversityofChicago 1101E.58thSt. ChicagoIL60637 7737023059 [email protected] June12,2000 2 Contents Acknowledgments 2 Preface 8 PartI. Asset pricingtheory 12 1 Consumption-basedmodelandoverview 13 1.1 Basicpricingequation 14 1.2 Marginalrateofsubstitution/stochasticdiscountfactor 16 1.3 Prices,payoffsandnotation 17 1.4 Classicissuesinfinance 20 1.5 Discountfactorsincontinuoustime 33 1.6 Problems 38 2 Applyingthebasicmodel 41 2.1 Assumptionsandapplicability 41 2.2 GeneralEquilibrium 43 2.3 Consumption-basedmodelinpractice 47 2.4 Alternativeassetpricingmodels: Overview 49 2.5 Problems 51 3 ContingentClaimsMarkets 54 3.1 Contingentclaims 54 3.2 Riskneutralprobabilities 55 3.3 Investorsagain 57 3.4 Risksharing 59 3.5 Statediagramandpricefunction 60 4 Thediscountfactor 64 4.1 Lawofonepriceandexistenceofadiscountfactor 64 4.2 No-Arbitrageandpositivediscountfactors 69 3 4.3 Analternativeformula,andx incontinuoustime 74 ∗ 4.4 Problems 76 5 Mean-variancefrontierandbetarepresentations 77 5.1 Expectedreturn-Betarepresentations 77 5.2 Mean-variancefrontier: IntuitionandLagrangiancharacterization 80 5.3 Anorthogonalcharacterizationofthemean-variancefrontier 83 5.4 Spanningthemean-variancefrontier 88 5.5 AcompilationofpropertiesofR ,Re andx 89 ∗ ∗ ∗ 5.6 Mean-variancefrontiersform: theHansen-Jagannathanbounds 92 5.7 Problems 97 6 Relationbetweendiscountfactors,betas,andmean-variancefrontiers 98 6.1 Fromdiscountfactorstobetarepresentations 98 6.2 Frommean-variancefrontiertoadiscountfactorandbetarepresentation 101 6.3 Factormodelsanddiscountfactors 104 6.4 Discountfactorsandbetamodelstomean-variancefrontier 108 6.5 Threeriskfreerateanalogues 109 6.6 Mean-variancespecialcaseswithnoriskfreerate 115 6.7 Problems 118 7 Implicationsofexistenceandequivalencetheorems 120 8 Conditioninginformation 128 8.1 Scaledpayoffs 129 8.2 Sufficiencyofaddingscaledreturns 131 8.3 Conditionalandunconditionalmodels 133 8.4 Scaledfactors:apartialsolution 140 8.5 Summary 141 8.6 Problems 142 9 Factorpricingmodels 143 9.1 CapitalAssetPricingModel(CAPM) 145 4 9.2 IntertemporalCapitalAssetPricingModel(ICAPM) 156 9.3 CommentsontheCAPMandICAPM 158 9.4 ArbitragePricingTheory(APT) 162 9.5 APTvs.ICAPM 171 9.6 Problems 172 PartII. Estimatingandevaluatingassetpricingmodels 174 10 GMMinexplicitdiscountfactormodels 177 10.1 TheRecipe 177 10.2 InterpretingtheGMMprocedure 180 10.3 ApplyingGMM 184 11 GMM:generalformulasandapplications 188 11.1 GeneralGMMformulas 188 11.2 Testingmoments 192 11.3 Standarderrorsofanythingbydeltamethod 193 11.4 UsingGMMforregressions 194 11.5 Prespecifiedweightingmatricesandmomentconditions 196 11.6 Estimatingononegroupofmoments,testingonanother. 205 11.7 Estimatingthespectraldensitymatrix 205 11.8 Problems 212 12 Regression-basedtestsoflinearfactormodels 214 12.1 Time-seriesregressions 214 12.2 Cross-sectionalregressions 219 12.3 Fama-MacBethProcedure 228 12.4 Problems 234 13 GMMforlinearfactormodelsindiscountfactorform 235 13.1 GMMonthepricingerrorsgivesacross-sectionalregression 235 13.2 Thecaseofexcessreturns 237 13.3 HorseRaces 239 5 13.4 Testingforcharacteristics 240 13.5 Testingforpricedfactors: lambdasorb’s? 241 13.6 Problems 245 14 Maximumlikelihood 247 14.1 Maximumlikelihood 247 14.2 MLisGMMonthescores 249 14.3 Whenfactorsarereturns,MLprescribesatime-seriesregression 251 14.4 Whenfactorsarenotexcessreturns,MLprescribesacross-sectional regression 255 14.5 Problems 256 15 Timeseries,cross-section,andGMM/DFtestsoflinearfactormodels 258 15.1 ThreeapproachestotheCAPMinsizeportfolios 259 15.2 MonteCarloandBootstrap 265 16 Whichmethod? 271 PartIII. Bondsandoptions 284 17 Optionpricing 286 17.1 Background 286 17.2 Black-Scholesformula 293 17.3 Problems 299 18 Optionpricingwithoutperfectreplication 300 18.1 Ontheedgesofarbitrage 300 18.2 One-periodgooddealbounds 301 18.3 Multipleperiodsandcontinuoustime 309 18.4 Extensions,otherapproaches,andbibliography 317 18.5 Problems 319 19 Termstructureofinterestrates 320 19.1 Definitionsandnotation 320 6 19.2 Yieldcurveandexpectationshypothesis 325 19.3 Termstructuremodels–adiscrete-timeintroduction 327 19.4 Continuoustimetermstructuremodels 332 19.5 Threelineartermstructuremodels 337 19.6 Bibliographyandcomments 348 19.7 Problems 351 PartIV. Empirical survey 352 20 Expectedreturnsinthetime-seriesandcross-section 354 20.1 Time-seriespredictability 356 20.2 TheCross-section:CAPMandMultifactorModels 396 20.3 Summaryandinterpretation 409 20.4 Problems 413 21 Equitypremiumpuzzleandconsumption-basedmodels 414 21.1 Equitypremiumpuzzles 414 21.2 Newmodels 423 21.3 Bibliography 437 21.4 Problems 440 22 References 442 PartV. Appendix 455 23 Continuoustime 456 23.1 BrownianMotion 456 23.2 Diffusionmodel 457 23.3 Ito’slemma 460 23.4 Problems 462 7 Preface Assetpricingtheorytriestounderstandthepricesorvaluesofclaimstouncertainpayments. Alow priceimplies a highrateof return, soonecanalsothinkof thetheoryas explaining whysomeassetspayhigheraveragereturnsthanothers. Tovalueanasset,wehavetoaccountforthedelayandfortheriskofitspayments. The effects of time are not too difficult to work out. However, corrections for risk are much moreimportantdeterminantsofanmanyassets’values. Forexample,overthelast50years U.S.stockshavegivenarealreturnofabout9%onaverage. Ofthis,onlyabout1%isdue to interest rates; the remaining 8% is a premium earned for holding risk. Uncertainty, or correctionsforriskmakeassetpricinginterestingandchallenging. Assetpricingtheorysharesthepositivevs. normativetensionpresentintherestofeco- nomics. Does it describe the waytheworlddoes workor thewaythe worldshould work? Weobservethepricesorreturnsofmanyassets. Wecanusethetheorypositively,totryto understandwhypricesorreturnsarewhattheyare.Iftheworlddoesnotobeyamodel’spre- dictions,wecandecidethatthemodelneedsimprovement.However,wecanalsodecidethat theworld iswrong, thatsomeassets are“mis-priced”andpresent tradingopportunitiesfor theshrewdinvestor. Thislatteruseofassetpricingtheoryaccountsformuchofitspopular- ityandpracticalapplication. Also,andperhapsmostimportantly,thepricesofmanyassets orclaimstouncertaincashflowsarenotobserved,suchaspotentialpublicorprivateinvest- mentprojects, newfinancialsecurities, buyout prospects, andcomplexderivatives. Wecan applythetheorytoestablishwhatthepricesoftheseclaimsshould beaswell; theanswers areimportantguidestopublicandprivatedecisions. Asset pricing theory all stems from one simple concept, derived in the first page of the firstChapterofthisbook: priceequalsexpecteddiscountedpayoff. Therestiselaboration, special cases, and a closet full of tricks that make the central equation useful for one or anotherapplication. Therearetwopolarapproachestothiselaboration. Iwillcallthemabsolutepricingand relativepricing. Inabsolutepricing,wepriceeachassetbyreferencetoitsexposuretofun- damental sources of macroeconomic risk. Theconsumption-basedandgeneral equilibrium modelsdescribedbelowarethepurestexamplesofthisapproach. Theabsoluteapproachis mostcommoninacademicsettings, inwhichweuseassetpricingtheorypositivelytogive aneconomicexplanationforwhypricesarewhattheyare,orinordertopredicthowprices mightchangeifpolicyoreconomicstructurechanged. Inrelativepricing,weaskalessambitiousquestion. Weaskwhatwecanlearnaboutan asset’svaluegiventhepricesofsomeotherassets.Wedonotaskwherethepriceoftheother set of assets came from, and we use as little informationabout fundamental risk factors as possible.Black-Scholesoptionpricingistheclassicexampleofthisapproach.Whilelimited inscope,thisapproachoffersprecisioninmanyapplications. 8 Assetpricingproblemsaresolvedbyjudiciouslychoosinghowmuchabsoluteandhow muchrelativepricingonewilldo,dependingontheassetsinquestionandthepurposeofthe calculation. Almostnoproblemsaresolvedbythepureextremes. Forexample,theCAPM anditssuccessorfactormodelsareparadigmsoftheabsoluteapproach. Yetinapplications, theypriceassets“relative”tothemarketorotherriskfactors,withoutansweringwhatdeter- mines the marketor factor riskpremiaandbetas. The latter are treated as freeparameters. Ontheotherendofthespectrum,mostpracticalfinancialengineeringquestionsinvolveas- sumptions beyond pure lack of arbitrage, assumptions about equilibrium “market prices of risk.” Thecentralandunfinishedtaskofabsoluteassetpricingistounderstandandmeasurethe sourcesofaggregateormacroeconomicriskthatdriveassetprices.Ofcourse,thisisalsothe centralquestionof macroeconomics, andthis is a particularlyexcitingtimefor researchers whowant toanswer thesefundamentalquestions inmacroeconomicsandfinance. Alotof empiricalworkhasdocumentedtantalizingstylizedfactsandlinksbetweenmacroeconomics andfinance. For example, expectedreturns varyacrosstimeandacross assets inways that arelinkedtomacroeconomicvariables,orvariablesthatalsoforecastmacroeconomicevents; a wide class of models suggests that a“recession” or “financial distress”factor lies behind manyassetprices. Yettheorylagsbehind; we donot yet haveawell-describedmodel that explainstheseinterestingcorrelations. In turn, I think that what we are learning about finance must feed back on macroeco- nomics. To take a simple example, we have learned that the risk premium on stocks – the expectedstock return less interest rates –is muchlarger than the interest rate, andvaries a gooddealmorethaninterestrates.Thismeansthatattemptstolineinvestmentupwithinter- estratesareprettyhopeless–mostvariationinthecostofcapitalcomesfromthevaryingrisk premium. Similarly,wehavelearnedthatsomemeasureofriskaversionmustbequitehigh, or people would all borrow like crazy to buy stocks. Most macroeconomics pursues small deviationsaboutperfectforesightequilibria,butthelargeequitypremiummeansthatvolatil- ityisafirst-ordereffect,notasecond-ordereffect. Standardmacroeconomicmodelspredict thatpeoplereallydon’tcaremuchaboutbusinesscycles(Lucas1987). Assetpricesarebe- ginningtorevealthattheydo–thattheyforegosubstantialreturnpremiatoavoidassetsthat fallinrecessions.Thisfactoughttotellussomethingaboutrecessions! This book advocates a discount factor / generalized method of moments view of asset pricingtheoryandassociatedempiricalprocedures. Isummarizeassetpricingbytwoequa- tions: p =E(m x ) t t+1 t+1 m =f(data,parameters). t+1 wherep =assetprice,x =assetpayoff,m =stochasticdiscountfactor. t t+1 t+1 9