House Prices and Consumer Welfare

HousePricesandConsumerWelfare*PatrickBajariUniversityofMichiganandNBERC.LanierBenkardStanfordUniversityandNBERandJohnKrainerFederalReserveBankofSanFranciscoThisversion:April,2005Abstract:Wedevelopanewapproachtomeasuringchangesinconsumerwelfareduetochangesinthepriceofowner-occupiedhousing.Inourapproach,anagent’swelfareadjustmentisdefinedasthetransferrequiredtokeepexpecteddiscountedutilityconstantgivenachangeincurrenthouseprices.Wedemonstratethat,uptoafirst-orderapprox-imation,thereisnoaggregatechangeinwelfareduetopriceincreasesintheexistinghousingstock.Thisfollowsfromasimplemarketclearingconditionwherecapitalgainsexperiencedbysellersareexactlyoffsetbywelfarelossestobuyers.Weshowthatthisresultholds(approximately)eveninamodelthataccountsforchangesinconsumptionandinvestmentplanspromptedbycurrenthousepricechanges.Therecan,however,bechangesinwelfareduetoadditionstothestockofhousing,ortochangesinthepriceofrenovatingandupgradingtheexistingstockofhousing.FortheUnitedStates,weesti-matethewelfarecostofhousepriceappreciationtobeanaverageof$127perhouseholdperyearoverthe1984-1998period.*WethankMichaelBoskin,TimBresnahan,TimErickson,GautamGowrisankaran,BobHall,EddieLazear,CarolynSargent,TomSargent,DannyBen-Shahar,JeroenSwinkels,RandyVerbrugge,andseminarparticipantsatWashingtonUniversityinSt.Louis,SITE,theHooverInstitution,theAREUEAmeetings,andtheBureauofLaborStatisticsforusefulcomments.Allremainingerrorsarethoseoftheauthors.BajariandBenkardthanktheNationalScienceFoundationandtheHooverInstitution’sNationalFellowsProgramforgenerousresearchsupport.TheviewsexpressedinthispaperarethoseoftheauthorsandnotnecessarilythoseoftheFederalReserveSystem.12I.IntroductionTherehasbeentremendoushousepriceappreciationintheUnitedStatesduringthelatterpartofthe1990s.Atthenationallevel,housepriceshaverisenatanaverageannualrateof7percentsince1995.Someregions,particularlythosewithaheavyconcentrationofhigh-techfirms,sawdoubledigitincreasesinhousepricesduringthepeakofthemostrecenteconomicexpansion(seeTable1).Thispriceappreciationsignificantlystrengthenedthebalancesheetsofexistinghomeowners.ArecentstudybyCanner,Dynan,andPassmore[3]estimatedthathouseholdsextractedapproximately$130billioninequityin2001andthefirsthalfof2002.Onthenegativeside,however,the7percentannualhousepriceappreciationinthelate1990’ssignificantlyoutpacedthegrowthinpercapitaincomeof2percentforthenationasawhole.Localgovernmentsinmetropolitanareaswiththesharpestpriceappreciation,suchasAustin,Boston,andSanFrancisco,routinelycitehousingaffordabilityastheirprincipleconstraintonfutureeconomicgrowth.Moregenerally,householdsinthelowertailoftheincomedistributionarenowlessabletoaffordhousingthantheywereadecadeago.Thislackofaffordabilityistroublingtopolicymakers,ashomeownershipisoftenassociatedwithavarietyofpositivesocialoutcomes(seeGreenandWhite[13]andDiPasqualeandGlaeser[7]).Thispaperisaboutthewelfareeffectsofchangesinthepriceofhousing.Thesheersizeofhousingintheaveragehousehold’sconsumptionbundlemakesitimperativetounderstandtheeffectsofchangingpricesonconsumerwelfare.Sheltercomprises32percentofthetotalconsumerpriceindex(CPI).Owner-occupiedhousingisalsoanassetwhichoccupiesalargeportionofthetypicalhousehold’sportfolio.Accordingtothe1998SurveyofConsumerFinances,thevalueofstocksownedbyhouseholdswas$7.8trillion,whilethevalueofprimaryresidenceswas$9.4trillion.Whiletherehasbeenalargeamountofresearchinterestonhowtoconstructaccuratehousepriceindices,therehasbeenrelativelylittleworkonmeasuringthewelfareeffectsofhousepricechanges.Mostmeasuresofhousingaffordabilityarecalculatedaspercentagesofatargetpopulationwithsufficientincometoaffordthemedian-pricedhouseatthecurrentinterestrate.Thesemeasures,however,aredifficulttointerpretbecausetheyabstractawayfrom,amongstotherthings,householdpreferencesandhouseholdwealth.Inaddition,affordabilityindexesdisplayvolatilitythatisinheritedfromthevariablesusedintheirconstruction,suchasinterestratesandprices.Bycontrast,homeownership3rates,thevariablesthataffordabilityindexesaresupposedtopredict,areamongtheleastvolatileoftheurbaneconomicseries(seeGlaeserandShapiro[12]).Thebodyofworkmostcloselyrelatedtooursisdevotedtodevelopingbettermeasuresofthecostofhousingandhousepriceinflation.CurrentpracticesattheBureauofLaborStatistics(BLS)reflecttheimportantresearchofKearl[16],DoughertyandVanOrder[10],andPoterba[17].Inconstructingitsowner-occupiedshelterindex,theBLSaimstomeasurechangesintheusercost,orthefairrentalvalueofowner-occupiedhousing.Theusercostcanbeinterpretedastheamountneededtocompensateahouseholdforaunitdecreaseinhousingservices,inagiventimeperiod,holdingassetownershipfixed.Whiletheusercostisagoodmeasureofthecostofhousingservices(whichistheobjectiveoftheBLS),therearetworeasonsthatmakeitunsatisfactoryasameasureofthewelfareeffectsofhousepricechanges.First,theusercostcalculationassumesthathouseholdscanadjusttheirconsumptionofhousingservicescostlessly,andthereforedoesnotaccountforbehaviorchangesinperiodsfollowingchangesinprices.Iftherewerenocosttoadjustingconsumptionofhousingservices,thenallconsumptionad