The Composition of Hedonic Pricing Models A Review

NATIONALASSOCIATIONOFNationalCenterforRealEstateResearchREALTORS®THECOMPOSITIONOFHEDONICPRICINGMODELS:AREVIEWOFTHELITERATUREbyG.StacySirmans,PhDKennethG.BachellerProfessorofRealEstateDepartmentofInsurance,RealEstateandBusinessLawCollegeofBusinessFloridaStateUniversityTallahassee,Florida32306-1110(850)644-8214(Phone)[phone](Fax)[e-mail],PhDAbbaLernerProfessorofEconomicsDepartmentofEconomicsFloridaStateUniversityTallahassee,FL32306-2180(850)644-3586Phone(850)[e-mail]oredbyTheNATIONALASSOCIATIONOFREALTORS®December2003iEXECUTIVESUMMARYTHECOMPOSITIONOFHEDONICPRICINGMODELS:AREVIEWOFTHELITERATUREAhouseismadeupofmanycharacteristics,allofwhichmayaffectitsvalue.Ahousehasbeencomparedtoabundleofgroceriesinthatthebundlescomeindifferentsizesandmaycontaindifferentitems.Unlikegrocerieshowever,thepriceofindividualhousingcharacter-isticscannotbedirectlyobserved.Hedonicregressionanalysisistypicallyusedtoestimatethemarginalcontributionofindividualcharacteristicstothetotalvalueoftheproperty.Understandingthemarginaleffectofindividualcharacteristicscanbevaluabletoanumberofmarketparticipantsincludinghomebuyers,realestateagents,developers,andrealestateappraisers.Thisstudyprovidesareviewofrecentstudiesthathaveusedhedonicmodelingtoestimatehouseprices.Thepaperismadeupofseveraldiscussions:theearlyhistoryofhedonicmodeling,therelationshipbetweensellingpricesandtimeonthemarket,andrecentstudiesusinghedonicmodeling.EventhoughCourt(1939)isoftenviewedasthefatherofhedonicmodeling,earlierhedonicstudiesthatexaminedthevalueoffarmlanddatebacktoHaas(1922)andWallace(1926).Laterstudiesdevelopedthemicroeconomicfoundationforestimatingthevalueofutility-generatingcharacteristics(Lancaster,1966)andfornonlinearhedonicpricing(Rosen,1974).Sellingpriceandtimeonthemarkettendtobeinteractive,thusspecifyingmodelsforthesevariablesinasimultaneousframeworkisoftendifficult.Timeonthemarketisgenerallynegativewhenestimatedinasellingpriceequation.Thisimpliesthatalongersellingtimeresultsinalowersellingprice.Whensellingpriceisincludedinatimeontheiimarketequation,theresultsarelessclear.Somemodelsshowthathouseswithhighersellingpricessellfasterwhileotherstudiesshowthathouseswithhighersellingpriceshavelongersellingtimes.Listingpriceisgenerallythoughttobeamajorfactorintimeonthemarketandstudiesshowthatahigherlistingpriceresultsinalongertimeonthemarket,thathousingliquiditydependsonmarketparticipants'searcheffortthatispartiallydeterminedbylistingprice,andthatitisexpensivetooverpriceahouseinitially.Usingtherecentliterature,thecharacteristicsthataremostfrequentlyincludedinhedonicpricingmodelsareidentified.Theseincludelotsize,squarefeet,age,thenumberofstories,thenumberofbathrooms,thenumberofrooms,thenumberofbedrooms,fireplace,centralairconditioning,basement,garage,deck,pool,brickexterior,distancetoCBD,timeonthemarket,andatimetrend.Thesevariablesgenerallyhavetheexpectedsignalthoughinsomeinstancestheyarenotsignificant.Duetothelargenumberofvariables,categoriesarecreatedandthetopcharacteristicsfromeachcategoryareidentified.Thecategoriesandcharacteristicsare:structuralfeatures:lotsize,squarefeet,age,numberofbathrooms,andnumberofbedrooms;internalfeatures:fullbaths,halfbaths,fireplace,airconditioning,hardwoodfloors,andbasement;externalfeatures:garagespaces,deck,pool,porch,carport,andgarage;naturalenvironmentalfeatures:lakeview,lakefront,oceanview,andgoodview;neighborhoodandlocation:location,crime,distance,golfcourse,andtrees;publicservices:schooldistrict,percentofschooldistrictminority,publicsewer;marketing,occupancy,andsellingfactors:assessor'squality,assessedcondition,vacant,owner-occupied,timeonthemarket,andtimetrend;andfinancing:FHAfinancing,VAfinancing,foreclosure,favorablefinancing,andpropertytaxes.Mostofthecharacteristicsiiihaveapositiveeffectonsellingprice.Thosecharacteristicsthathavehadanegativeeffectonpriceincludeage,crime,percentofschooldistrictminority,andifapropertyisvacant.Followingaresomeotherinterestingvariablesthatareseentoaffectsellingprice.Thosethathaveapositiveeffectincludeaslantedversusflatroof,asprinklersystem,agardenbath,aseparateshowerstall,adoubleoven,andagatedcommunity.Othercharacteristicsthathaveanegativeeffectonsellingpriceincludenothavingatticspace,livinginanearthquakezone,proximitytoahogfarm,proximitytoalandfill,proximitytohighvoltagelines,corporateownedproperties,andpropertiesthatrequirefloodinsurance.Thestudycomparesestimatedcoefficientsacrossgeographicalregionsforselectedcharacteristics.Somemajorconclusionsare:•theeffectofsquarefootageonsellingpricedoesnothaveagreatdealofvaria-tionacrossregions.ThegreatesteffectwasintheSouthwestandthelowestaver-ageeffectwasintheMidwest;•theeffectoflotsizewasgenerallyconsistentacrossregions;•agewasconsistentlynegativeandtheeffectonpriceseemstobeconsistentacrossregions;•forstudiesprimarilyfromtheNortheastandSouthwest,eachadditionalbath-roomincreasedsellingpriceinthe10to12percentrange;•forstudieslimitedtotheNortheastandSouthwest,theeffec