MAKINGTHEHOUSEAHOME:THESTIMULATIVEEFFECTOF

HOMEPURCHASESONCONSUMPTIONANDINVESTMENT1

EfraimBenmelechKelloggSchoolofManagement,NorthwesternUniversityandNBER

AdamGuren

BostonUniversity

BrianT.MelzerTuckSchoolofBusiness,DartmouthCollege

April9,2019

Abstract

We introduce and quantify a new channel through which the housing market affectshousehold spending: the homepurchase channel. Using an event-study designwith datafromtheConsumerExpenditureSurvey,weshowthathouseholdsspendonaverage$3,700moreinthemonthsbeforeandthefirstyearfollowingahomepurchase.Thisspendingisconcentrated in the home-related durables and home improvements sectors, which arecomplementary to thepurchaseof thehouse.Expendituresonnondurablesanddurablesunrelatedtothehomeremainunchangedordecreasemodestly.WeestimatethatthehomepurchasechannelplayedasubstantialroleintheGreatRecession,accountingforone-thirdof the decline in home-related durables spending and a fifth of the decline in homemaintenanceand investmentspending from2005to2010, togethertotaling$14.3billionannually.

1WethankGadiBarlevy,SusantoBasu,MartyEichenbaum,SimonGilchrist,BarneyHartman-Glaser,Michael

Reher,JeremyStein,AmirSufi,andseminarandconferenceparticipantsattheFederalReserveBankof

Chicago,GreenLineMacroMeeting,HebrewUniversity,StanfordUniversity,TelAvivUniversity,Universityof

CaliforniaBerkeleyHaasSchoolofBusiness,UniversityofChicago’sConferenceonHousingand

Macroeconomics,UniversityofCopenhagen’sWorkshoponNewConsumptionData,UniversityofNotre

Dame’sRoundtableonHousingandMortgageMarkets,andtheUniversityofSouthernCaliforniaLuskCenter

forRealEstateSymposium.SashaIndarteandPaolinaMedinaprovidedgreatresearchassistance.Benmelech

isgratefulforfinancialsupportfromtheGuthrieCenterforRealEstateResearchattheKelloggSchoolof

Management.Allerrorsareourown.

1

I.INTRODUCTION

Whydohouseholdconsumptionandthehousingmarketmove in tandemthrough

periodsofbothprosperityanddecline?Thisquestionhasbeencentraltomacroeconomic

analysis and monetary policy-making in the United States since 2000, as the aggregate

economyexperiencedadramaticexpansionandcontraction thatmirrored theboomand

bust in the housingmarket. Previous studies of this pattern have focused on the role of

housingwealth inspurringhouseholdconsumption through itseffectsonoverallwealth,

creditconstraints,andemployment.

Inthispaperweproposeandprovideevidenceforafurtherlinkbetweenthehousing

marketandhouseholdconsumptionthatdoesnotoperatedirectlythroughhouseprices.We

argue thathomepurchases,whichexperiencedaboomandbust similar to thatofhome

pricessince2000,stimulatedurableconsumptionbyraisingdemandforgoodsandservices

complementary to thehome.Thisrelationship follows fromtwomainassumptions.First,

owingtosearchfrictions,householdscannotfindhomesthatmatchtheirspecifictastesand

stock of durable goods. Buyers therefore tailor their newly purchased home to their

preferencesbyalteringthephysicalstructureandbybuyingnewfurnishingsandappliances.

Second,thesealterationsandpurchasesareatleastinpartirreversible.Homerenovations

andadditions,forexample,cannotbemovedfromoneresidencetothenext.Manyfixtures,

appliances,andfurnishingsarealsopurchasedtocomplementaparticularphysicalspace

andsoarepurchasedanewafteramove.Giventheseassumptions,aggregateconsumption

willexpandandcontractwiththenumberoftransactionsduringhousingcycles.Thishome

purchase channel is particularly potent in housing downturns,when sales tend tomove

morestronglywith–andreactproportionatelymorethan–homeprices.

2

Ourprimaryanalysisusesmicrodataonhouseholdspendingandbuildingpermitsto

estimatetherelationshipbetweenhomepurchasesandhome-relatedspending.Weanalyze

monthlyexpendituresreportedbyhomeownersintheConsumerExpenditureSurvey(CE)

between2001and2013.Wealsoanalyzequarterlyhomeimprovementactivityusingthe

buildingpermithistoryofapproximatelyninemillionhomesthatsoldbetween2001and

2013.

Agreatdealofhouseholdspendingistiedtothehome.Homeownerssurveyedinthe

CEspendanaverageof$1,300peryearonhomedurablesand$2,500peryearonhome

improvement and maintenance. This home-related spending constitutes nearly 40% of

homeowners’totaldurableandimprovementspendingof$10,200peryear.

Weuseanevent-studymethodologytoestimateboththetimingandmagnitudeof

spending responsesafter thepurchaseof ahome.Withaggregatedata,homevaluesand

homepurchasesmove in tandem,which complicates the separate identification of home

value and home purchase effects. 2 Individual-level data, however, enable an analysis of

homebuyers’expenditurespreciselyaroundthedateoftheirhomepurchase.Thisfeature

allowsustoisolatespendingbynewhomebuyersfromspendingbyexistingownerswho

haveexperiencedachangeinhousingwealth.Thevariationinthetimingofhomepurchases

further allows for time fixed effects that absorb general business-cycle fluctuations in

spending. Lastly, our preferred specificationmakes use of the panel nature of the CE by

controllingforhouseholdfixedeffects.Thesefixedeffectsnarrowtheidentifyingvariation

2.Thecorrelationofhomesalesandhousepricesinlevelsisbetween0.75and0.85andinlogchangesis

between0.4and0.5.SeeAppendixBfordetails.

3

towithin-householddifferences in timeafterhomepurchaseandabsorbhousehold-level

spendingdifferencesthatrelate,forexample,tovariationinwealth,income,andstageoflife.

We estimate that homebuyers spend $5,740 (measured in 2009 dollars)more on

homedurablesandimprovementsfromthreemonthsbeforethepurchasethroughoneyear

after the purchase of a primary residence. This increase includes $2,340 of additional

spendingonhome-relateddurables,whichamountstoatriplingofspendingcomparedto

longer-tenured owners’ $1,130 annual spending on home-related durables. Recent

homebuyersalsospendsubstantiallymoreonhomeimprovementandmaintenance.These

investments in thehomemore thandouble in the firstyearafterpurchase, increasingby

$3,400relativetolonger-tenuredowners’averageannualinvestmentof$2,350.Bothhome

durables and investment spending spike particularly in the first quarter following home

purchaseandremainmodestlyhigher forsix toninemonthsafterhomepurchasebefore

levelingoffattheirlong-termaveragebytheendofthefirstyearofownership.Households

thatpurchaseadditionalproperties—vacationhomesorinvestmentproperties—alsoboost

spending,butbyasmalleramountthanowner-occupants.

Our analysis of building permits likewise shows a substantial increase in home

improvements:theestimatedcostofpermittedjobsmorethandoublesinthefirstyearof

homeownership,increasingby$2,500.Thebuildingpermitdataallowfurtherexamination

ofany intertemporal substitutionofhome improvements.We find thathomesellersalso

increasespendingonimprovementsbyanaverageof$730intheyearbeforethesale.We

conclude that first-year owners’ improvement spending is incremental, since it does not

merelyreplacesellers’deferredinvestments.

4

In contrast to the patterns we observe for home-related spending, we find that

spending on nondurables and durables unrelated to the home remain flat or decline

modestlyaroundhomepurchases.Onnet, the increases inhomedurables, improvement,

and maintenance spending are not offset by spending declines in other consumption

categories.Theabsenceofspendingincreasesunrelatedtothehomealsoreinforcesacausal

interpretationofthemainresults.

The event-study methodology is not immune to omitted variables critiques. For

instance,ashockthatcausesthehouseholdtobuyanewhome–suchasawindfallincrease

inwealth,ajobpromotionthatraisesincomeexpectations,orachangeinfamilystatus,such

asthebirthofachild–mayalsoincreasedurablespendingevenintheabsenceofahome

purchase. One would expect, however, that such omitted variables would cause

systematicallyhigherspending,evenamongcategoriesthatareunrelatedtothehome.Yet

purchasesofvehicles,jewelry,books,andhealthequipmentareverysimilarbetweenfirst-

year and long-tenured owners after accounting for household fixed effects. Indeed, only

audiovisual goods, which we view as home related but which the Bureau of Economic

Analysis’sNationalIncomeandProductAccounts(NIPA)classifyasrecreational,showan

increaseafterhomepurchase.Thisobservanceofspendingincreasessotightlyassociated

with the home suggests that household fixed effects are successful in absorbing omitted

variablesinourmainanalysis.

Therelationshipbetweenhomepurchasesandspendinghasprovenimportantinthe

aggregate,particularlyduringtheGreatRecession.FigureIshowsthetimeseriesforhome

sales (in blue and on the left axis) and for combined home durables, improvement, and

maintenancespending(inredandontherightaxis)throughtheGreatRecession.Homesales

5

plungedbynearly50%between2005and2010,from8.36millionunitsperyearto4.50

millionunitsperyear.Spendingonhomedurablesandhomeimprovementandmaintenance

alsodeclineddrastically,fallinginrealtermsby12%and28%,respectively,overthesame

period.TheseweresomeofthelargestdeclinesinspendingacrossallcategoriesintheGreat

Recession.

Drawing on our event-study estimates for spending after home purchase and the

declineinhomesalesfrom2005to2010,wecalculatethatthecollapseofhomepurchases

ledtoa$14.3billionannualdeclineinspendingandinvestmentduringtheGreatRecession.

This partial equilibrium aggregation excludes any spending by suppliers of real estate

services,suchasrealtorsandbuildingcontractors,whoearnincomefromhomepurchases

andimprovements.Nevertheless,thischannelexplainsmorethanone-thirdofthedeclinein

spendingonhomedurablesand15%to20%ofthedeclineinhomeimprovementsspending

followingthelate-2000shousingcrisis.

We provide a further point of comparison to the literature on home prices and

consumption (e.g.,Mian, Rao, and Sufi, 2013) by analyzing the aggregate sales of home-

relateddurablesbymetropolitanareaintheEconomicCensus.Usingvariationinhousing

cycles across metropolitan areas, we jointly estimate the elasticities of home-related

spending tohomepricesand tohomepurchases.Home-relatedspendingmovesstrongly

with home purchases during both the housing boom (2002 to 2007) and bust (2007 to

2012).Theelasticityofspendingtopurchasesis0.26duringtheboomand0.12duringthe

bust.Bycomparison,theelasticityofspendingtohomepricesis0.29intheboomand0.23

inthebust.Whiletheseresultsdonothaveasconvincingidentificationasourevent-study

results, they help to compare the home purchase channel to the housing wealth effect

6

channelandillustratethatthehomepurchasechannelisameaningfuldriverofspendingin

thehousingsector.

Our work relates to a broad literature on the relationship between the housing

marketandconsumption.Severalpapershavequantifiedthisrelationshipandgenerallyfind

thathouseholdsconsumebetweenfourandelevencentsofeachdollarchangeinhousing

wealth.This literature includesworkwithregionalvariationbyCase,Quigley,andShiller

(2005,2012),MianandSufi(2011,2014a,2014b),Mian,Rao,andSufi(2013),andGurenet

al. (2018); workwith household-level data by Hurst and Stafford (2004), Campbell and

Cocco (2007),Attansioetal. (2009),andAttanasio,Leicester, andWakefield (2011);and

workwithaggregatedatabyCarroll,Otsuka,andSlacalek(2011).Intermsofunderstanding

themechanism,Cooper(2013)andDeFusco(2018)usenovelempiricalapproachestoshow

that the collateral channel is important. On the theoretical side, Chen, Michaux, and

Roussanov (forthcoming), Gorea andMidrigan (2018), Berger et al. (2018), and Kaplan,

Mitman, and Violante (2019) use calibrated models to inspect the mechanism, mainly

focusingonthecollateralchannel.MostdirectlyrelatedtothispaperisworkbyBestand

Kleven(2018),whoestimate thestimuluseffectsofchanges inhousing transactiontaxes

usingidentificationfromtaxnotchesintheUnitedKingdom.Inanextension,theyuseannual

U.K.microdatatoestimatethatahousingpurchasetriggersauxiliaryspendingequalto5%

ofthehouse’spriceinthefirstyearand1%thefollowingyear.Theiranalysisusesannual

repeatedcrosssectionsandcannotcapturethefinenessinconsumptionandheterogeneity

thatisafocusofourpaper.

Ourpaperalsorelatestotheliteratureontheconsumptionofdurables.Mostofthe

theoreticalliteraturefocuseson(S,s)adjustmentmodelswithfixedcosts(e.g.,Eberly1994).

7

Our findings highlight that many durable goods purchases are complementary to other

goodsorlifechanges,suchaspurchasinganewhouse.

Therestofthepaperisorganizedasfollows.SectionIIdescribesthedata,summary

statistics,andempiricalmethodology.SectionIIIpresentstheempiricalresultsofhousehold

spendingpatternsaroundhomepurchaseforhomedurablesandhomeimprovementsand

maintenance. In Section IV we discuss our empirical identification strategy and present

placebo and robustness tests. Section V explores heterogeneity in the home purchase

channel.SectionsVIandVIIprovideestimatesoftheaggregateeffectofthehomepurchase

channelandacomparisontotheeffectofhomeprices.SectionVIIIconcludes.

II.DATAANDMETHODOLOGY

We use survey data on household spending from the Consumer Expenditure

InterviewSurvey(CE),administrativedataonbuildingpermitsfromthedataanalyticsfirm

BuildFax,anddataonretailsalesfromtheEconomicCensus.

II.A.ConsumerExpenditureSurvey

OurprimarydatasourceistheCE,whichprovidesmonthlypaneldataonhousehold

spendingforarandomsampleofnearlythirtythousandhouseholdsperyear.Thedataare

ideal for our study because they combine measures of household expenditures with

information on home purchase timing, home characteristics, and household mortgage

borrowing. Survey participants remain in the sample for one year and report their

expendituresretrospectivelythroughfourquarterlyinterviews.

8

WeusetheCE’sdetaileddataonthetiming,value,andcategoryofexpendituresto

construct monthly data on purchases of durable goods and spending on home-related

maintenanceandimprovementprojects.Thesurveyincludesanextensivesetofquestions

aboutspendingondurables,withseparatesectionsdevotedtohomefurnishings,appliances

andhouseholdequipment,vehicles,andhomemaintenanceandimprovementprojects.For

durable goodspurchasedover theprevious threemonths, households report the typeof

product(e.g.,diningtableorrefrigerator),thepricepaid,andthemonthofpurchase.For

newandusedvehicles,householdsreportthemonthofpurchase,thenetpricepaid,andthe

value of any trade-in. Finally, for homemaintenance and improvement projects, survey

participantsreportthetypeofproject(e.g.,addinginsulationorreplacingsiding)andtheir

monthlyspendingonsuppliesandcontractorlabor.Theseprojectsincludehomeadditions

and remodelingprojectsaswell as repairsandmaintenance throughout the interiorand

exteriorofthehome.Wedeflateallexpendituresto2009dollarsusingtheConsumerPrice

Index (CPI) for each spending category for durables and using NIPA deflators for home

maintenanceandimprovement.

Inadditiontotrackinghouseholdspending,thesurveygathersinformationonhome

ownershipthatallowsustomeasurethetimeelapsedsincehomepurchase.Foreachhome,

includingthehousehold’smainresidenceaswellasanyvacationandinvestmentproperties,

thesurveycollectsthemonthandyearthepropertywasacquiredandtheestimatedcurrent

value.Forowner-occupiedproperties, the surveyalsocollects theageandbasicphysical

characteristics(e.g.,numberofbedroomsandbathrooms)ofthehome.Wethereforefocus

ourmainanalysison thepurchaseofprimary residences, forwhichwe canobserveand

control for theageand characteristicsof thehome.For eachhouseholdwe calculate the

9

numberofmonthssinceacquiringtheirprimaryresidence.Infurtheranalysisweestimate

the aggregate effects of all home purchases by considering purchases of non-primary

residencesandinvestmentpropertiesaswell.Inthatcasewemeasuretime-sincepurchase

astheminimumnumberofmonthssinceacquisitionacrossthehousehold’sproperties.

AnimportantfeatureoftheCEsurveydesignisthat itdoesnotfollowhouseholds

thatmoveafterenteringthesample;thesehouseholdsexitthesampleuponmoving.Asa

result,we observe atmost threemonths of expenditures before amove,with pre-move

spendingreportedretrospectivelybythefewhouseholdsthatentertheCEsamplejustafter

movingintotheirnewresidence.Ourmainanalysisthusprimarilyexploitsthevariationin

time-since-purchase for survey participants while they remain at a given residence. In

additionaltests,weusethepre-movingexpenditurestoexamineintertemporalsubstitution

in themonthspreceding themove, although confidence intervalswiden in thesemonths

becauseofsmallsamplesizes.

II.B.BuildingPermitData

Our second data source is a database of residential building permits compiled by

BuildFax. City or county agencies typically require homeowners or their contractors to

obtainbuildingpermitsbeforemakingsignificanthomeadditionsandalterations.BuildFax

collects these records from permitting agencies and organizes them into property-level

historiesofpermittingactivity.BuildFaxreportsthenumberofpermits,brokendownbythe

typeofwork:electrical,mechanical,plumbing,orstructural.Insomejurisdictions,BuildFax

observesanestimateofthebuilding,electrical,mechanical,plumbing,andtotaljobcost.The

10

dataspan1994to2013,withBuildFax’scoverageofpermittingjurisdictionsgrowingover

thistimeperiodfromroughly25%to50%ofU.S.homes.

Werestrictoursampletohomeswithasalestransaction.Usingpropertydeeddata

fromDataQuick,weobtainthepropertyaddressesofallsingle-familyhomeswithasales

transactionbetween2001and2013.BuildFaxthenmatchestheseaddressestothepermit

recordsinitsdatabaseandreturnsapaneldatasetwiththequarterlypermittingactivityfor

eachpropertyinajurisdictionwithdatacoverage.ThematchedBuildFax–DataQuickdataset

includesninemillionpropertiesforwhichthereisatleastonehomepurchasetransaction

duringoursampleperiod.

ThepermitdataprovidetwoadvantagesovertheCEdataforthestudyofresidential

improvements. First, the permit data afford more statistical power because of the

substantiallylargersampleofhometransactions.Second,thepermitdatacoverawidertime

window around home purchases.Whereas the CE data are available only threemonths

beforeahomepurchase,thepermitdataenablethestudyofimprovementsintheyearsand

monthsleadinguptoahomesale.Totheextentthatsellersarealsobuyersofanewproperty,

thiswider timewindowallowsus toevaluate theextentof intertemporal substitution in

improvements.

II.B.EconomicCensusData

OurthirddatasourceistheU.S.CensusBureau’sEconomicCensus,whichmeasures

theannualsalesandemploymentofallbusinesseseveryfiveyears.Themostrecentthree

surveysprovidedatafor2002,2007,and2012,whichcorrespondroughlytothebeginning,

peak,andtroughofthehousingcycle.Weexamineaggregatesalesbymetropolitanareafor

11

retailersofhome-relatedgoods,includingfurniture,electronicsandappliances,andbuilding

supplies.

II.C.SampleDescriptionandSummaryStatistics

Table I presents summary statistics for the CE and BuildFax samples. The table

reports household characteristics (Panel A), property characteristics (Panel B), and

householdspending(PanelC)fortheCEdata,anditreportspermittingactivity(PanelD)for

theBuildFaxdata.

TheCEsample includesallhomeownerssurveyedbetweenApril2001andMarch

2013.Thesampleincludes665,802monthlyobservationsfor70,529homeownerswithnon-

missinginformationonthedateofhomepurchase.

As the sample statistics in Panel A show, average household annual income and

financialassetholdings,measuredin2009dollars,are$73,516and$57,444,respectively.In

termsofeducation,11%ofhouseholdheadsinthesamplelackahighschooldiploma,26%

have only a high school diploma, 29%have some college education, 21%have a college

degree,and13%haveagraduatedegree.Ofthesampledhouseholdheads,73%arewhite,

14%areHispanic,8%areblack,and3%areAsian.Moving tomaritalstatus,65%of the

householdheadsaremarried,13%aredivorced,11%arewidowed,and10%havenever

beenmarried.Themeanfamilysizeis2.6.Theaverageagefortheheadofhouseholdis53

years,and24%ofhouseholdshaveaheadofhouseholdorspousewhoisretired.

PanelBreportssummarystatisticsonpropertycharacteristics.Thetypicalproperty

inoursamplewaspurchasedaroundthirteenyearsbeforethesurveydate.Morethan60%

ofthehouseholdshaveamortgage,andabout7%ofthehouseholdshaverefinancedtheir

12

mortgageinthetwelvemonthsbeforethesurvey.Theaveragehomeis37yearsoldandhas

3bedroomsand2bathrooms.

PanelCpresents informationonhouseholdspending thatweuse toconstructour

dependentvariables in theregressionanalysis.TheCEexpendituredataaredetailedand

comprehensive.Weclassifyspendingas(1)homeimprovementandmaintenance,(2)home

durables,(3)non-homedurables,and(4)food.Theimprovementsandmaintenancevariable

includesallspendingformaterials,tools,andlaborbutexcludesthecostofhomeappliances,

whichweallocatetospendingonhomedurables.Thevastmajorityofspendingonnon-home

durables,morethan75%,goestowardvehicles.

As Panel C demonstrates, the average spending on home improvement and

maintenanceis$209permonth,amountingto$2,508peryear,whiletheaveragespending

onhomedurablesis$108permonth,or$1,296peryear.Theotherspendingcategoriesof

non-homedurablesandfoodaverage$532and$612permonth,respectively.Households

makehome improvementexpenditures inabout17%of themonths inoursample.They

purchasehomedurablesmorefrequently,in28%ofthehousehold-monthobservationsin

theCEsample.

Thepermittingsampleincludesallhomeswithasalestransactionbetween2001and

2013. Our sample includes 9,081,284 properties with at least one purchase transaction

duringour sampleperiod,with44.1quartersofBuildFaxdata coverageperpropertyon

average,withatotalof400,060,883property-quarterobservations.AsPanelDillustrates,

theaveragepropertyhas1.85permitsduringoursampleperiod,withameancountof0.31

permits forelectricalwork,0.21permits formechanicalwork,0.22permits forplumbing

work,and1.11permitsforstructuralwork.Themeantotaljobcostforallproperty-quarter

13

observations in thedata is $634perquarter.However, conditional onproperty-quarters

with positive permit-related expenditure, the mean total job cost is $42,990. This high

averagejobcostrelativetotheCEreflectsthefactthatbuildingpermitsarenecessaryfor

largehome improvementsbutnot forsmaller improvements.Becauseof this,wesee the

analysisoftheBuildFaxdataassupportiveofthemorerepresentativefiguresintheCE.

III.ESTIMATINGHOME-RELATEDSPENDINGPATTERNSFOLLOWINGHOMEPURCHASE

AshighlightedintheIntroduction,homepurchasesarehighlycorrelatedwithhome

values.Time-seriesregressionsusingaggregatedatafailtoidentifyhomepurchaseeffects

separately from home value effects because of this collinearity. Cross-sectional analyses

using instruments thatprojectontobothpriceandsalesvolume,suchas theSaiz (2010)

instrument, also cannot separate the home purchase channel from housing wealth or

collateraleffects.

Inthispaperweapplyamethodologythatisinessencesimilartoaneventstudy.We

use spending microdata at the household level, which provides additional variation to

identifyhomepurchaseeffects:cross-sectionalvariationinspendingcategoriesaswellas

time-seriesvariationmeasuredastimesincehomepurchase.

III.A.AverageConsumptionandPermittingActivitybyTimeSincePurchase

Table II provides a first pass at exploiting this variation by showing households’

averagemonthlyspendingrelativetothedatetheypurchasedtheirprimaryresidence.The

rawdifferencesinspendingaroundhomepurchaseeventsarenotnecessarilycausedbythe

homepurchase,buttheyprovideausefulpointofdepartureforthesubsequentanalysis.

14

Panel A of Table II reports averagemonthly spending by category in the quarter

beforeahomepurchaseandeachofthefirstfourquarterafterthepurchase.Thefinalrow

ofthetableshowsaveragemonthlyspendingforhouseholdsthatpurchasedfiveormore

quartersago.Allspendingfiguresareinrealterms(2009dollars)andadjustedforinflation

usingtheConsumerPriceIndexforAllUrbanConsumers(CPI-U)pricedeflator.

AsPanelAofTableIIshows,theaveragemonthlyhomeimprovementspendingin

thefirstquarterafterthehomepurchaseis$598comparedtothemeanof$196permonth

afterthefirstyearofownership.Thatis,householdsspendthreetimesmoreineachofthe

first three months after the home purchase. Moreover, home improvement spending

remainselevated inthesecondquarterafter thepurchase—spending is$382permonth,

whichisnearlytwiceaslargeastheaveragespendingbeyondthefirstyearofownership.

Spendingonhomeimprovementalsoremainshighinthethirdandfourthquartersafterthe

homepurchase,at$325and$280permonth,respectively.

Likewise,householdspendingonhomedurablesincreasesdramaticallyinthefirst

quarterafterthepurchaseofaproperty.AsTableIIshows,theaveragemonthlyspending

onhomedurables in the first threemonths after thehomepurchase is $750permonth,

comparedtoamean$94permonthafteroneyearofownership,representinganalmost

eightfoldincreaseinspendingonhomedurables.Thelevelofspendingonhomedurables

remainshigherthanthemeanthroughoutthefirstyearafterthehomepurchase.Wealso

findanincreaseinspendingonnon-homedurablesintheyearafterthehomepurchase,but

thisincreaseissmallerthanthoseforhomeimprovementorhomedurables.Forexample,

spending on non-home durables is $787 per month in the first quarter after the home

purchase, representing an increase of 51% over the $521 per month spent among

15

households beyond the first year of ownership. Finally, we do not find any significant

changes in household spending on food following a home purchase, suggesting that the

purchaseitselfisunlikelytoleadtoadramaticchangeinthehousehold’slifestyle.

In Panel B of Table IIwe summarize the building permit activity following home

purchases.Thefractionofpropertieswithat leastonepermit is7.7%inthefirstquarter

followingapurchase.The incidenceofpermits then falls steadilyduring the firstyearof

ownershipandreaches2.6%amonghomesthataremorethanoneyearbeyondthepurchase

date.Theestimatedtotal jobcostdisplaysasimilarpattern,decliningfromanaverageof

$1,292inthefirstquarterfollowingthepurchaseto$506perquarterbeyondthefirstyear

ofownership.

Theserawdifferencesinspendingofcoursemayreflectselectionintowhopurchases

rather than spending caused by the home purchase itself. For example, if wealthier

households both spend more on food and move more often, we would observe similar

spendingpatternseveniffoodspendingwereunaffectedbyahomepurchase.Ourempirical

strategyaimstoaddressthisissueandprovideacausalestimateofthespendingresponses

tohomepurchase.

III.B.AnEmpiricalModelofSpendingPatterns

Thesummarystatisticspresentedabovesuggestthathouseholdsspenddramatically

moreonhomeimprovementandhomedurablesduringthefirstyearofownership.Wenow

turntoamultivariateanalysisofhouseholdspendingandpermittingactivityfollowinghome

purchases.

16

Weuseanevent-studymethodology,estimatingthe followingregressionmodel in

theCEsample:

1 𝑆𝑝𝑒𝑛𝑑𝑖𝑛𝑔*+ = 𝛼 + 𝛽01 𝑀𝑜𝑛𝑡ℎ𝑠𝑠𝑖𝑛𝑐𝑒𝑃𝑢𝑟𝑐ℎ𝑎𝑠𝑒 = 𝑚<<0=>? + 𝛿+ + 𝜸′𝑿*+ + 𝜀*+ ,

where the dependent variable is spending by household i in month t. As measures of

spending, we consider alternately the level of spending (dollars per month), the log of

spending (natural logarithm of 1 + dollar spending), and the incidence of spending (an

indicator for spending>0).Aside fromdifferent scaling, the levels specification tends to

focusonfittinglargeexpenditures,whereasthelogspecificationputsmoreweightonsmall

purchases.

Thecoefficientsofinterestarefixedeffectsforeachmonthrelativetothetimeofthe

homepurchase(𝛽0,𝑚 ∈ −3, 11 ).Thesefixedeffectsmeasurethehousehold’sincremental

spendingineachofthethreemonthsbeforethehousepurchaseaswellasthefirsttwelve

monthsofhomeownershiprelativetoanexcludedcategoryofthirteenormoremonthsafter

purchase. Time-since purchase pertains to the household’s purchase of its primary

residence,forwhichwecanobserveandcontrolfortheageandphysicalcharacteristicsof

thehome.We include theseproperty characteristics in thevectorX,whichalso includes

controlsforhouseholdincomeandwealth,aswellasdemographics:householdsize,theage

of the head of household, and indicators for the head of household’s marital status,

retirementstatus,race,andeducation.Themodelalsoincludesmonth-by-yearfixedeffects

(𝛿+)tocontrolforcommonvariationinspendingovertime,suchasfluctuationsthroughthe

businesscycle.

Mostimportant,ourpreferredspecificationmakesuseofthepanelnatureoftheCE

by controlling for household fixed effects in addition to the control variables discussed

17

above.Thesefixedeffectsnarrowtheidentifyingvariationtowithin-householddifferences

intimeafterpurchaseandabsorbhousehold-levelspendingdifferencesthataredrivenby

suchfactorsaswealth,income,orstageoflife.

Weestimatethemodelwithordinaryleastsquares,andinrobustnesstestsweshow

similar findings forProbitandTobitmodelsdesigned fordiscreteandcensoredoutcome

variables. We calculate Huber-White standard errors with observations clustered by

household.Mostoftheanalysisispresentedusingfiguresthatplotthepatterninhousehold

spendingfollowinghomepurchaseormortgagerefinancing.Thefiguresareplottedbased

on the coefficients of the fixed effects for each month relative to the date of the home

purchase (𝛽0,𝑚 ∈ −3, 11 ). These fixed effects measure the household’s incremental

spendingineachofthethreemonthsbeforethepurchaseaswellasthefirsttwelvemonths

ofhomeownershiprelativetoanexcludedcategoryofthirteenormoremonthsfollowing

thepurchaseafterwecontrolforpropertyandhouseholdcharacteristics.

Weestimateasimilarmodelinthepermittinganalysis:

2 𝑃𝑒𝑟𝑚𝑖𝑡𝑠J+ = 𝛼 + 𝛽K1 𝑄𝑢𝑎𝑟𝑡𝑒𝑟𝑠𝑆𝑖𝑛𝑐𝑒𝑃𝑢𝑟𝑐ℎ𝑎𝑠𝑒 = 𝑞NK=>N + 𝜹𝒕 + 𝜽𝒑 + 𝜀J+ .

Theunitofobservationinthisanalysisisthepropertypinaquartert.Animportantfeature

ofthepermittingdataisthatwecanobservepermittingintheyearsbeforeatransaction,so

weareabletoexpandthetime-relative-to-purchasedummiestoincludeeachoftheeight

quarters before and after the home purchase. Even though the quarters before the sale

concern a different homeowner, which means that we cannot measure intertemporal

substitutionforasinglehouseholdwiththepermitdata,wecanneverthelesstestwhether

onaveragesellersdefermaintenanceorupgradetheirhousebeforeselling.Becausemost

sellers repurchase, this helps us assess intertemporal substitution for permitted home

18

improvements.Bycontrast,becausetheCEdoesnottrackmovers,wecanobserveonlyup

tothreemonthsbeforeapurchase.

Wemeasurepermittingactivityalternatelyastheincidenceofapermit(perquarter),

thenumberofpermits(perquarter),thetotalestimatedjobcost(perquarter),andthelog

oftheestimatedjobcost(natural logarithmof1+ jobcost).Ascontrolvariables insome

specifications,weincludequarter-by-yearfixedeffects(𝛿+)andpropertyfixedeffects(𝜃J).

Weestimatethemodelwithordinaryleastsquaresandclusterobservationsbyproperty.

III.C.HomeDurableSpendingFollowingHomePurchase

TableIIIpresentsresultsfromestimatingRegression(1)usingameasureofhome

durablesspendingasadependentvariable.Asthetableillustrates,thereisanincreasein

spendingduringthetwelvemonthsfollowingahomepurchase.Whenwedonotcontrolfor

householdorpropertycharacteristics,logspendingincreasesby1.87logpointsinthefirst

month,2.02logpointsinthesecondmonth,and1.09logpointsinthethirdmonth.Theselog

differencesequatetoproportionalincreasesof549%,654%,and197%,respectively.This

patternofincreasedconsumptionthatpeaksinthesecondmonthremainssimilarwhenwe

controlforpropertyandhouseholdcharacteristicsandyearfixedeffectsinColumn(2)as

wellaswhenwecontrolforhouseholdfixedeffectsinColumn(3).

Controlling forhouseholdcharacteristicsand thenhousehold fixedeffectsreduces

the estimated spending responses modestly for the first few months following a home

purchaseandmoresubstantiallyformonths6to12followingthepurchase.Theelevated

spendinglateinthefirstyearofownership—reflectedincoefficientsof0.2to0.3formonths

9 to 11 of the first specification—falls modestly after controlling for household

19

characteristicsandthenbymorethanhalfaftercontrollingforhouseholdfixedeffects.Some

oftheelevatedspendinginthefirstyearofownershipevidentintherawdata,therefore,is

causedbygenerallyhigherspendingbyhouseholdsthattendtobuyhomesmorefrequently

ratherthanbyahomepurchaseperse.Thevariationacrossspecificationsisshownvisually

inAppendixA.Nevertheless,thereisasubstantialhomedurablesspendingresponseinthe

first nine months after purchase even for the most stringent specification that includes

householdfixedeffects.

Intermsofdollarspending,homedurablesspendingincreasesby$851inthefirst

monthand$744 in thesecondmonthafterahomepurchase. In the thirdand the fourth

monthsafterthepurchase,spendingincreasesby$303and$184,respectively.Finally, in

Column(5)ofTableIIIweusetheincidenceofspendingasourdependentvariableandfind

thatthepropensitytospendonhomedurablesincreasesby20%inthefirstmonth,24%in

thesecondmonth,and12%inthethirdmonthafterahomepurchase.Theincidenceofhome

durablesspendingis5%–8%higherinmonths4,5,and6and3%–4%higherininmonths

7,8,and9.

Table III alsopresentshousehold spending for the threemonthsbefore thehome

purchase.Asthetabledemonstrates,householdsreducetheirspendingonhomedurables

inthethreemonthsbeforemakingahomepurchase.Forexample,thepropensitytospend

is 18%, 19%, and 13% lower in months −3, −2, and −1 relative to the home purchase,

respectively. The spending patterns in these three months are consistent with some

intertemporalsubstitutionofconsumption inwhichbothbuyersandsellers inapending

transaction may delay durable goods purchases and home improvements until the

transactioniscompleted.However,theanalysisofdollarspendingshowsnodifferencein

20

spendingbeforeahomepurchaseandtheanalysisoflogspendingindicatesthatspending

afterthepurchasedwarfsthedeclinebeforethepurchase.Intheanalysisoflogspending,

forexample,thecoefficientsimplyacumulativespendingdifferenceof+6.97logpointsin

thetwelvemonthsafterapurchase,whichisthreetimesaslargeasthecumulative−2.13log

pointdeclineinthethreemonthsbeforethepurchase.

ThepatternofspendingonhomedurablesisillustratedinFiguresII.A,II.B,andII.C.

Eachfigureisplottedbasedonthecoefficientsofthefixedeffectsforeachmonthrelativeto

thedateofthehomepurchase,usingasdependentvariablesthelogarithmofhomedurables

spending(Figure II.A), thedollaramountofhomedurablesspending(Figure II.B),or the

incidenceofhomedurablesspending(FigureII.C).

III.D.HomeImprovementandMaintenanceFollowingHomePurchase

We also analyze household spending patterns on home improvement and

maintenanceandreporttheresultsinTableIVandFiguresIII.A,III.B,andIII.C.AsTableIV

demonstrates, we find a significant increase in home improvement and maintenance

spendingfollowingahomepurchase.Theincidenceofhomeimprovementspendingis11%

inthemonthofthehomepurchase,whileitisnotstatisticallydifferentfromzerointhethree

months before the purchase. The likelihood that a household will conduct home

improvementandmaintenanceincreasesto15%inthefirstmonthafterthepurchaseandis

10% and 7% in the second and third months, respectively, after the purchase. The

magnitudesofthesespendingarealsosignificant,amountingtoanaverageof$485,$479,

and$301ineachofthethreemonthsafterthehomepurchase.

21

Togaugethetotaleffectofahomepurchaseonspending,weanalyzethecombined

spendingonalldurablegoods(homeandnon-home)aswellasonhomeimprovementand

maintenance.TheresultsarepresentedinTableVandinFiguresIV.A,IV.B,andIV.C.AsTable

V, Column (4) shows, the combined spending on durables and home improvement and

maintenanceis$1,462inthemonthofthehomepurchase.Spendingremainshighinthenext

threemonths,withatotalof$1,323inthefirstmonthafterthepurchaseand$659and$452

inthesecondandthirdmonthsafterthepurchase,respectively.

III.E.PermittingActivityAroundHomePurchases

Results from estimating Regression (2) for our analysis of building permits are

presentedinTableVI.ThedependentvariableinColumns(1)to(3)isthenaturallogarithm

of1+totaljobcost,whileinColumns(4)and(5),thedependentvariableistheactualjob

cost indollars.Column(1)showsan increaseof0.241 logpoints inpermit total jobcost

duringthequarterinwhichahomeispurchased,oranincreaseof27.3%inpermitcost.The

estimatedjobcostisalso0.181and0.094logpointshigherinthefirstandsecondquarters

afterthepurchase,representingproportionalincreasesof19.8%and9.9%,respectively.The

patternofincreasedjobcostduringthequarterofthehomepurchaseandthefollowingtwo

quarters also remains similarwhenwe control for year fixed effects in Column (2) and

propertyandyearfixedeffectsinColumn(3).

Intermsofactualjobcost,theaverageincreasesby$763duringthequarterofthe

homepurchaseandby$835inthefirstquarterafterthepurchase.Inthesecondandthe

third quarters after the purchase, the average job cost increases by $561 and $365,

respectively,anditincreasesbyanadditional$223inthefourthquarterafterthepurchase,

22

asshowninColumn(4).Interestingly,permitjobcostsalsoincreasebeforethepurchase,

suggesting that home sellers invest in their homes before selling. This reveals that

householdsthatarebothbuyersandsellersdonotintertemporallysubstitutebydeferring

maintenanceontheiroldhomebeforepurchasinganewhomeandinvesting in it. Italso

suggeststhatifanythingourestimateofhomeimprovementandmaintenanceintheCEis

anunderestimatebecause itdoesnot fullyaccount forupgradesmadebeforea sale that

occurmorethanthreemonthsbeforethepurchase.

Finally,inColumn(5)ofTableVIweusetheincidenceofhavinganypermitasour

dependentvariableandfindthatthelikelihoodofhavingapermitincreasesby5.5%during

thequarterofthehomepurchase,followedbya3.5%increaseinthefirstquarterafterthe

purchaseand1.9%and1.3%increasesrespectivelyinthesecondandthirdquartersafter

thepurchase.

ThepermitdataoverallcorroboratetheresultsfromtheCE.Althoughwedonotwish

to takea strong standonaggregatedollarvalues, given thatnot allhome improvements

require a permit, the general time pattern is highly consistent withwhat we see in the

consumer survey, and a two-year look-back produces no evidence consistent with

intertemporalsubstitution.

IV.CAUSALITYANDROBUSTNESS

Theresultssofarsuggestthathouseholdsincreasetheirspendingonhomedurables,

improvement,andmaintenancewhentheypurchaseahouse.Theevidenceshedslighton

thelinkbetweenthehousingmarketandprivateconsumptionthroughatransactionchannel

ratherthanapricechannel.However,itisimportanttoexploremoredeeplywhetherthese

23

effectsarecausalorwhetherunobservedheterogeneityatthehouseholdlevelisdrivingthe

results. Our regressions control for household income and wealth and various

demographics:householdsizeandheadofhouseholdage,maritalstatus,retirementstatus,

race,andeducation.Ourresultsarerobusttotheinclusionofhouseholdfixedeffects,which

absorbfixeddifferencesinexpendituresduringtheyearthatthehouseholdappearsinthe

data.

Byincludinghouseholdfixedeffects,weexploittheexacttimingofspendingrather

thanrelyingonacoarsecomparisonbetweenhouseholdsthatrecentlypurchasedahome

andthosethatdidnot.Toclarifythispoint,itishelpfultoconsidertheexampleofelderly

homeowners, who tend to stay in the same home and to keep their current household

appliances.Amodelwithoutcontrolsorhouseholdfixedeffectswouldfindhigherspending

ondurablesineachmonthduringthefirstyearofownership,sincetheelderlyspendata

low rate and are disproportionately represented in the excluded category: owners who

purchasedmorethanoneyearago.Omittingagefromthemodelcausesanupwardbiasin

thespendingcoefficientsacrosstheboardineachofthefirsttwelvemonthsafterpurchase.

Controllingforageor,moreflexibly,absorbingunobservedheterogeneitywithahousehold

fixedeffectresolvesthisproblem.

It is still possible, however, that unobserved shocks that coincidewith (andmay

potentially drive) home purchase lead to increased spending precisely in the first few

monthsafterthepurchase.Thesepotentialomittedvariablesmayinclude:(1)unobserved

financial or housingwealth that permits higher spending; (2) an increase in permanent

incomethatmayresultinanincreasedpropensitytoconsume;and(3)anincreaseinfamily

sizethatrequiresgreaterspendingondurablesandmayalsocausethehouseholdtomove.

24

Wenextconsideraseriesofplaceboteststoalleviateconcernsthatourresultsare

drivenbyomittedfactorsunrelatedtothehomepurchase.Totestwhetheromittedvariables

aredrivingourresults,weinvestigatewhetherhouseholdspendingincategoriesunrelated

to home purchase display the same patterns as spending on home durables and home

improvementandmaintenance.UsinginformationprovidedintheCE,westudyspendingon

non-homedurables,food,andentertainment.

FiguresV.A,V.B,andV.Cshowspendingpatternsfornon-homedurables(Figure5.A),

food (Figure V.B), and entertainment (Figure V.C) corresponding to fixed effects in

Regression(1).3Asthefiguresdemonstrate,thereislittlechangeinconsumptionpatterns

in any of these categories. We find a very modest response of spending on non-home

durablesandnoresponseof spendingon foodorentertainment in themonths following

home purchase. If any wealth, income, or household size shocks coincided with home

purchasesandcausedshort-runincreasesinspending,onewouldexpectthemalsotoboost

non-housingspending.Instead,wefindmuchalargerresponseofhome-relatedspending.

Weshouldalsonoteherethatnon-homedurablesandfoodspendingarenotinsensitiveto

income andwealth – coefficients on those controls show the expected relationship. The

failuretofindanon-housingandnon-durablespendingresponsethusdoesnotresultfrom

anygeneralinelasticityofthesespendingcategoriestoincomeandwealth.

3.Thefiguresareconstructedfromregressionsthatusethelogarithmoftheexpenditureasadependent

variable.FiguresthatusetheactualdollarcostlooksimilarandcanbefoundinAppendixA.

25

V.HETEROGENEITYINTHEHOMEPURCHASECHANNEL

Therichnessofourdataallowsus toexploreheterogeneity in thehomepurchase

channeltoprovidemoretextureonhowitoperates.TableVIIassessesseveraldimensions

ofheterogeneityinthehomepurchasechannelbyreportingthecumulativechangeindollar

spending from three months before a home purchase through twelve months after the

purchase.

Wefirstsplitthesampleintoaboomperiodwithhighsalesfrom2001to2006anda

bustperiodwithlowsalesfrom2007to2012.Wefindthatmarginalpurchasesintheboom

periodinducemorehomedurablespendingbutlesshomeimprovementspendingthando

purchases in thebustperiod.Thenext foursplitsarebycharacteristicsof thepurchased

home.Wefindthattherearelargerresponsesforlargerhomesintermsofsize(numberof

bedrooms)andhomevalue.Intermsofhomeage,theincreaseindurablesandimprovement

expendituresislargestforhomesunderthreeyearsold,suggestingthatthereisstillsome

customization of newer homes. The purchase of homes serving as a primary residence

inducesmorehomedurablespendingandsubstantiallymoreimprovementspendingthan

doesthepurchaseofavacationorinvestmentproperty.Thefinaltwosamplesplitsareby

homeowner characteristics.We find that older and higher-income homeowners increase

their spending on durables and improvements by a larger amount following a home

purchase.

V.A.SummaryofCumulativeEffectofHomePurchaseonSpending

Figure VI summarizes our overall results by showing the cumulative impulse

responseofspendingtoahomepurchasebycategory.Asdiscussedabove,theincreasesin

26

spendingonhomedurablesandhomeimprovementandmaintenancefollowingpurchase

dwarf the declines in these series and on non-homedurables spending before purchase.

Cumulatively, the increase in spending from threemonths before purchase until twelve

monthsafterisabout$3,400forhomeimprovementand$2,340forhomedurables,withan

$850declineinnon-homedurablesspending.Theaggregateeffectisthusnearly$5,000for

the purchase of primary residences. These effects are somewhat smaller for secondary

residences, so theaggregate cumulativeeffectunconditionalonwhether the residence is

primaryorsecondaryis$3,700.

VI.THEAGGREGATEEFFECTOFTHEHOMEPURCHASECHANNELFROM2000TO2011

In this section, we assess how much the home purchase channel contributed to

changesinconsumptioninthehousingboomfrom2000to2005andtheensuingbustfrom

2005to2011.Todoso,wecomparethechangeinconsumptionthatonewouldpredictby

multiplying the change inhome salesbyourpreferred estimates of the effect of a home

purchase on consumption in each category to the actual change in consumption in that

categoryintheCE.4,5Thisprovidesasimple,partialequilibriumaccounting,similartothe

5.WedonotcomparetoNIPAtablesbecauseNIPAusesdifferentdefinitionsthatmakeitdifficulttocompare

levelschanges.Thisproblemisparticularlyacuteforhomemaintenance,repair,andimprovement,because

theseCEcategoriesaresplitacrossresidentialinvestment,“otherservices”inpersonalconsumption,and

“imputedrent”inpersonalconsumption.

6.Onemayworrythatoureffectsoverstatetheaggregateconsumptionchangebecausehouseholdssellhome

durablesaroundasale,andsosomeofthedurablesconsumptioniscanceledoutintheaggregatebysalesof

durables.However,wefindnoevidenceofincreasedhouseholdincomefromsalesofgoodsaroundahome

saleintheCE.Wethusconcludethatourestimatesdonotoverstatetheaggregatechangeinconsumptionfor

thisreason.

27

literature on housingwealth effects, that assesses howmuch of the observed change in

consumptioncanbeaccountedforbythehomepurchasechannel.

To create the national time series for consumption and home improvement and

maintenance, we aggregate the CE using the provided sample weights. We deflate

subcategoriesofdurableconsumptionbytheirsubcategoryCPIdeflatorandsubcategories

of improvement and maintenance by their NIPA deflator. For home sales, we use non-

seasonallyadjusteddataonexistinghomesalesfortheentireUnitedStatesfromtheNational

AssociationofRealtorstogetherwithmonthly,non-seasonallyadjusteddataonsalesofnew

homes fromtheCensusBureau.Thesedataprovideamoreaccurate timeseriesofhome

salesthanhomesalesaggregatedusingtheCEweights,althoughreassuringlythetwoseries

arequiteconsistent.Thisisthecasebecausethesecategorieshadlowerinflationthanthe

aggregateCPIintheboombuthigherinflationthantheaggregateinthebust.Consequently,

deflatingbycategorypriceindicesleadstoalargerrun-upinconsumptionintheboomand

asmallerdeclineinconsumptioninthebust.Wemultiplythemonthlytimeseriesbyour

preferredestimatesof averagedollar spending in the threemonthsbeforepurchase, the

monthofpurchase,andeachofthefollowingtwelvemonthsandthenaggregatetheimplied

timeseriestotheannuallevel.Wenexttakedifferencesbetween2000–2002and2005for

theboomaswellas2005to2008–2011forthebust.Wethendividethischangebytheactual

changeintheaggregateconsumptiontimeseriesforthesamecategorycreatedusingtheCE

microdataandweights,whichgivesusthepercentageoftheaggregatechangeexplainedby

thehomepurchasechannel.Weuse2005asourbaseyearbecausehomesalespeakedin

2005.

28

OurresultsareshowninTableVIII,withtheboomabovethehorizontallineandthe

bustbelowthehorizontalline.Ourresultsaregenerallystrongerforthebustthantheboom

fortworeasons.First,homesalesfellmoreinthebustthantheyroseintheboom.Second,

homedurablesconsumption–andtoalesserextenttheothercategories–grewbymorein

theboomthanitshrankinthebust,sothedenominatorisbiggerintheboomthanthebust.

Column (1) of Table VIII shows that in the boom the home purchase channel

accountedforabout8.5%to9.0%ofthegrowthinspendingonhomedurables,whileinthe

bustthehomepurchasechannelcontributed23%to39%percentofthedeclineinspending

onhomedurables.6Thehomepurchasechannelhasasmallnegativeeffectonnon-home

durables spending, mostly because households cut back slightly on non-home durables

spending in the months preceding a move. Column (4) shows the effect on home

improvementsspending,whichwas7.4%to9.3%intheboomand13%to19%percentin

thebust.Maintenanceisamuchmoreerratictimeseries,particularlyintheboom,butwe

can see a 10% to40%effect in thebust, as shown inColumn (5). Column (6) combines

improvements and maintenance spending, finding slightly higher numbers when

maintenanceisaddedthanforimprovementsalone.

Thecategoriesofspendingaffectedbythehomepurchasechannelareonlypartof

overalldurablesandhomemaintenancespending.Toshowtheeffectofthehomepurchase

channelontheselargercategories,TableVIIIshowstheaggregateeffectofthesedeclineson

total durables in Column (3) and on total durables plus home improvements and

7.Thedifferencesbetween2005–2008,2009,2010,and2011canlargelybeascribedtothetimepathofthe

denominator.Thisisbecausehomesalesfallfrom2005to2008andthenstayatroughlythesamelowlevel,

whilehomedurablesconsumptioncontinuestodeclineslightlyuntil2009,afterwhichitbeginstorebound.

Similarpatternscanbeseenfortheothervariables.

29

maintenanceinColumn(7).Intheboom,theaggregateeffectontotaldurablesspendingwas

1.5%to2.0%,risingto3.7%to4.8%whenspendingonmaintenanceandimprovementsis

added.Inthebust,theeffectontotaldurablesspendingwas2.7%to3.6%,risingto6.6%to

9.1%whenspendingonmaintenanceandimprovementsisadded.Notethatthesefigures

are likely underestimates for the impact of home transactions on maintenance and

improvements spending, because the CE design precludes us from accounting for

improvementsmadebysellersinordertomarketandselltheirhouse.

Toprovideasenseofthemagnitudeofthiseffect,theaveragehomepurchasetriggers

a net spending of $3,700 on durables, home improvement, andmaintenance from three

monthsbeforepurchasetooneyearafterpurchase,7andfrom2005to2010salesfellby3.86

millionunits.Thisimpliesanannualdeclineinspendingofapproximately$14.3billion,or

approximately0.1%ofGDP.Asayardstickforcomparison,Mian,Rao,andSufi(2013)find

thathomeequityfellby$5.6trillionfrom2006to2009andfindamarginalpropensityto

consume (MPC) metric out of housing wealth of 5.4%, implying a total decline in

consumptionof$302.4billion,with$128.8billionaccountedforbyautos,$89.6billionby

non-durables,and$61.6billionbynon-autodurables.Ourannualeffectof$14.3billionover

three years is thus equivalent to roughly three-quarters ofMian,Rao, and Sufi’s average

annualestimateddeclineinnon-autodurablesassociatedwiththedeclineinhousingprices

from 2006 to 2009 in the Great Recession, or roughly 15% of their overall decline in

consumption.

8.Thisfigurediffersfromtheanalysispresentedabovebecauseitcombinesbothprimaryresidencesand

secondhomes,whereasthemainanalysisaboveisforprimaryresidencesonly.

30

VII.COMPARINGTHEEFFECTSOFHOMEPURCHASESANDHOMEVALUESONCONSUMPTIONUSINGCITY-

LEVELSPENDING

To benchmark the home purchase channel relative to the more-widely-studied

housingwealtheffectchannel,wecarryoutananalysisofaggregate,city-levelspendingand

home purchases and compare the response of spending to house prices and house

purchases.WeusedataonretailsalesfromtheEconomicCensus,whichcollectstheannual

salesofallbusinesseseveryfiveyears.Themostrecentthreesurveysprovidedatafor2002,

2007,and2012.Thesedatescorrespondroughlytothebeginning,peak,andtroughofthe

housingcycle.Wefocusonsales inhome-relatedgoods,whichincludesalesbyfurniture,

electronicsandappliances,andbuildingsuppliesstores.

Wemeasure theelasticityof spending tohomepurchasesandhomevaluesat the

CBSAleveloverthesefive-yearhorizons.Toparsetheseparateeffectsofhomepurchases

and home values,we estimatemultivariate regressions that rely on variation in housing

cyclesacrossCBSAstoseparatelyidentifytheeffectsofhomepurchasesandhomevalues.

OurmethodologyissimilartothatofMian,RaoandSufi(2013),whichmeasuresthe

elasticityofspendingtohousingwealthatthecountyleveloverathree-yearhorizon(2006

to2009).Weestimateelasticitiesoverfive-yearhorizonsusingthefollowingmodel:

3 ∆(𝐿𝑜𝑔𝑆𝑝𝑒𝑛𝑑𝑖𝑛𝑔)W+ = 𝛼 + 𝛽∆𝐿𝑜𝑔(𝐻𝑜𝑚𝑒𝑃𝑢𝑟𝑐ℎ𝑎𝑠𝑒𝑠)W+>< +

𝛾∆𝐿𝑜𝑔(𝐻𝑜𝑚𝑒𝑃𝑟𝑖𝑐𝑒𝐼𝑛𝑑𝑒𝑥)W+>< + 𝜹𝒕 + 𝜀W+ ,

wherethedependentvariableisthefive-yearchangeinlogannualretailsalesinCBSAcas

ofyeart.WeusedatafromCoreLogictomeasuretheannualvolumeofhomepurchasesand

theyear-endCoreLogichomepriceindexforprices.Wecomputethesemeasuresfortheyear

t-1,inordertoexaminespendingthatoccurssubsequenttothechangesinhomepurchases

31

andhomeprices.Similartothedependentvariable,wetakefive-yearchangesinloghome

purchases and the log home price index. The coefficients𝛽 and𝛾 thereforemeasure the

elasticitiesofspendingtohomepurchasesandhomeprices,respectively,andareseparately

identifiedbydifferencesinthetimeseriespatternsofpurchasesandprices.Weincludeyear

fixedeffects(𝛿+)toabsorbchangesinspendingcommontoallmetropolitanareasovereach

five-year period. We estimate the model using ordinary least squares with population

weights and clustering by metropolitan area for the calculation of standard errors. Our

estimationsampleincludesanobservationfor246CBSAswiththerequisitespendingand

housingdatain2007and267CBSAswiththerequisitedatain2012.

Themodel estimates, reported inTable IX, showsignificantpositiveelasticitiesof

home-related spending to both home purchases and home prices. In the housing boom,

between2002and2007,theelasticitiesofspendingtohomepurchasesandhomepricesare

0.26(p<0.01)and0.29(p<0.01),respectively.Inthesubsequenthousingbust,between

2007 and 2012, the elasticities of spending to home purchases and home prices decline

somewhatto0.12(p<0.01)and0.23(p<0.01),respectively.Whiletheseresultsarenotas

well-identifiedasourevent-studyresults,theyprovidecomplementaryevidencethathelps

to benchmark the strength of the home purchase channel relative to the more-studied

housingwealtheffectandunderscoreourmainconclusionthatthehomepurchasechannel

isimportantandcomplementarytohousingwealtheffects.

VIII.CONCLUSION

This paper describes and quantifies a new channel for the causal relationship

betweenhousingmarketsandspending:thehomepurchasechannel.Inthemonthsbefore

andintheyearfollowingahomepurchase,householdsspend$3,700,primarilyonhome-

32

related durable goods and homemaintenance and repair. To address concerns that this

relationship is driven by unobserved heterogeneity, we use an event-study design that

includeshousehold-levelfixedeffects,sothatallidentificationcomesfromavariationwithin

households before and after a homepurchase. To address concerns that the spending is

causedbyaneventthattriggersahomepurchaseratherthanthehomepurchaseitself,we

show that there is no related effect for food or entertainment spending, and a modest

offsettingeffectfornon-homedurablegoodsspending(includedinthe$3,700figureabove).

ThischannelplayedaquantitativelysizableroleinintheGreatRecessionandamore

modestroleinthepriorhousingboom.Itaccountedforathirdofthedeclineinhomedurable

goodsspendingduringtheGreatRecessionandafifthofthedeclineinhomeimprovement

and maintenance spending. A back-of-the-envelope calculation implies that the home

purchasechannelaccountedfora$14.3billion–or0.1%ofGDP–declineinspendingper

yearintheGreatRecession.

Thehomepurchasechannelcomplementstheresponseofconsumptiontochangesin

housingwealthdrivenbyhousepricesthathasbeenthefocusofrecentliterature(e.g.,Mian,

Rao,andSufi2013).AppendixAshowsthatchangesinhousingpricesandsalesvolumeare

highly correlated, both in levels and in log changes. The home purchase channel is

quantitatively smaller than the housingwealth channel because it only affects non-auto

durablesspending,butforthesecategoriesitissubstantial:themagnitudeofoureffectis

equivalenttothree-quartersofthecausaleffectofthe2006–2009changeinhousingwealth

onnon-autodurablesconsumptionestimatedbyMian,Rao,andSufi(2013).

Beyondunderstandingthemechanismsconnectingthehousingmarketandspending

intheGreatRecession,ourestimatesareofrelevancetopolicymakers.Totheextentthat

33

monetarypolicyispassedthroughintomortgageinterestrates,monetarypolicycanhavea

substantial impact on housing transaction volume. Our estimates are a crucial input for

monetarypolicymakerswhowishtounderstandtheeffectofthesesalesondurablesand

homeimprovementspending.Inaddition,ourfiguresareacrucialinputintothecost-benefit

analysisforfiscalpolicymakersinterestedinpursuingpoliciesdesignedtostimulatehome

sales, such as the new homebuyer tax credit in the Great Recession (Berger et al.,

forthcoming).

34

References

Attanasio,Orazio,AndrewLeicester,andMatthewWakefield,“DoHousePricesDrive

ConsumptionGrowth?TheCoincidentCyclesofHousePricesandConsumptionin

theUK,”JournaloftheEuropeanEconomicsAssociation,9(2011),399–435.

Attanasio,Orazio,LauraBlow,RobertHamilton,andAndrewLeicester,“BoomsandBusts:

Consumption,HousePricesandExpectations,”Economica,76(2009),20–50.

Berger,David,VeronicaGuerrieri,GuidoLorenzoni,andJoeVavra,“HousePricesand

ConsumerSpending,”ReviewofEconomicStudies855(2018),1502-1542.

Berger,David,NicholasTurner,andEricZwick,“StimulatingHousingMarkets,”Journalof

Finance(forthcoming).

Best,Michael,andHenrikKleven,“HousingMarketResponsestoTransactionTaxes:

EvidencefromNotchesandStimulusintheUK,”ReviewofEconomicStudies85

(2018),157-193.

Campbell,JohnY.,andJoaoF.Cocco,“HowDoHousePricesAffectConsumption?,”Journal

ofMonetaryEconomics,54(2007),591–621.

Carroll,ChristopherD.,MisuzuOtsuka,andJiriSlacalek,“HowLargeAreHousingand

FinancialWealthEffects?ANewApproach,”JournalofMoney,Credit,andBanking,1

(2011),55–79.

Case,KarlE.,JohnM.Quigley,andRobertJ.Shiller,“ComparingWealthEffects:TheStock

MarketVersustheHousingMarket,”AdvancesinMacroeconomics,5(2005),1–30.

———,“WealthEffectsRevisited:1975–2012,”CriticalFinanceReview,2(2013),101–128.

Chen,Hui,MichaelMichaux,andNikolaiRoussanov,“HousesasATMs?Mortgage

RefinancingandMacroeconomicUncertainty,”JournalofFinance(forthcoming).

35

Cooper,Daniel,“HousePriceFluctuations:TheRoleofHousingWealthasBorrowing

Collateral,”ReviewofEconomicsandStatistics,95(2013),1183–1197.

DeFusco,Anthony,“HomeownerBorrowingandHousingCollateral:NewEvidencefrom

ExpiringPriceControls,”JournalofFinance73(2018),523-573.

Eberly,JaniceC.,“AdjustmentofConsumers’DurableStocks:EvidencefromAutomobile

Purchases,”JournalofPoliticalEconomy,102(1994),403–436.

Gorea,Denis,andVirgiliuMidrigan,“LiquidityConstraintsintheU.S.HousingMarket,”

WorkingPaper,2018.

Guren,AdamM.,AlisdairMcKay,EmiNakamura,andJonSteinsson,“HousingWealth

Effects:TheLongView,”WorkingPaper,2018.

Hurst,Erik,andFrankStafford,“HomeIsWheretheEquityIs:MortgageRefinancingand

HouseholdConsumption,”JournalofMoney,Credit,andBanking,36(2004),985–

1014.

Kaplan,Greg,KurtMitman,andGianlucaViolante,“ConsumptionandHousePricesinthe

GreatRecession,”WorkingPaper,2019.

Mian,Atif,andAmirSufi,“HousePrices,HomeEquity-BasedBorrowing,andtheU.S.

HouseholdLeverageCrisis,”AmericanEconomicReview,101(2011),2132–2156.

———,“WhatExplainsthe2007–2009DropinEmployment?,”Econometrica,82(2014),

2197–2223.

———,“HousePriceGainsandU.S.HouseholdSpendingfrom2002to2006,”Working

Paper,2014.

Mian,Atif,KamaleshRao,andAmirSufi,“HouseholdBalanceSheets,Consumption,andthe

EconomicSlump,”QuarterlyJournalofEconomics,128(2013),1687–1726.

36

Saiz,Albert,“TheGeographicDeterminantsofHousingSupply,”QuarterlyJournalof

Economics,125(2010),1253–1296.

37

TableI.SummaryStatistics

PANELA:HouseholdCharacteristics Mean Std.Dev

PANELB:PropertyCharacteristics Mean Std.Dev

Income/Wealth Monthssincepurchase 160 156

Annualincome 73,516 66,335 Purchasedinprevious

12months?(%) 6.37 24.42Financialassets 57,444 291,913 Mortgagor?(%) 63.96 48.01Assetsinformation

missing? 0.11 0.32 Monthssince

refinancing# 59 70

Education Refinancedinprevious

12months?(%) 6.82 25.21Nohighschooldiploma 0.11 0.32 Ageofhome(years) 37 29Highschooldiploma

only 0.26 0.44 Ageofhomemissing? 0.09 0.29Somecollege 0.29 0.46 Rooms 6.63 2.03Collegedegree 0.21 0.41 Bedrooms 3.11 0.89Graduatedegree 0.13 0.33 Bathrooms 1.80 0.74

Race/Ethnicity Centralair? 0.68 0.47White 0.73 0.44 Swimmingpool? 0.11 0.31Black 0.08 0.27 Porch? 0.82 0.38Hispanic 0.14 0.35 Off-streetparking? 0.83 0.37Asian 0.03 0.18

Other 0.01 0.12 PANELC:HouseholdSpending

MaritalStatus Married 0.65 0.48 Spendingpermonth($)

Widowed 0.11 0.31 Homeimprovement

andmaintenance 209 1,960Divorced 0.13 0.33 Homedurables 108 698Separated 0.02 0.12 Non-homedurables 532 3,623Nevermarried 0.10 0.30 Food 612 500

Other Likelihoodofpurchase(permonth)

Ageofhouseholdhead 52.86 15.82 Anyhome

improvement? 0.17 0.38Familysize 2.63 1.45 Anyhomedurables? 0.28 0.45

Retired?(%) 23.69 42.52 Anynon-home

durables? 0.52 0.50

Anydurablesor

improvement? 0.38 0.49 Notes:Sampleincludes665,802monthlyobservationson70,529homeownerswithnon-missinginformationondateofhomepurchaseinCEsamplebetween2001and2013.

38

TableI.SummaryStatistics(Continued)

PANEL D: Building Permits Estimated Total Job Cost ($)

Unconditional, per quarter 634 13,149 Conditional on permit 42,990 99,502

Number of permits, by type (per property) Electrical 0.313 1.989 Mechanical 0.208 1.420 Plumbing 0.224 1.525 Structural 1.106 5.936 All 1.850 8.999

Notes:Buildingpermitsdataincludes400,060,883quarterlyobservationson9,081,284propertiesbetween2001and2013.

39

Table II. Household Spending Following Home Purchase or Mortgage Refinancing

Panel A: Spending After Home Purchase

Average Monthly Spending ($)

Quarters Since Purchase Home Improvement Home

Durables Non-home Durables Food

1 598 750 787 596 2 382 251 717 607 3 325 175 627 608 4 280 151 657 606

5+ 196 94 521 613

Panel B: Building Permits After Home Purchase

Building Permit Activity

Quarters Since Purchase Any Permit? Number of Permits Estimated Job

Cost ($)

1 0.077 0.122 1,292 2 0.057 0.094 1,344 3 0.040 0.063 1,032 4 0.034 0.052 813

5+ 0.026 0.039 506

Notes: Real spending per month measured in January 2009 dollars (CPI-U as price deflator).

40

TableIII.SpendingonHomeDurablesSurroundingaHomePurchase

DependentVariable:HomeDurables

MonthssincepurchaseLog

SpendingLog

SpendingLog

Spending Spending($) Spending>0?

-3 -0.45*** -0.48*** -0.86*** -53 -0.18*** (0.15) (0.15) (0.22) (226) (0.04)-2 -0.66*** -0.61*** -0.79*** 4 -0.19*** (0.06) (0.06) (0.10) (78) (0.02)-1 -0.38*** -0.36*** -0.48*** 121 -0.13*** (0.05) (0.05) (0.09) (78) (0.02)0 1.87*** 1.90*** 1.84*** 851*** 0.20*** (0.07) (0.07) (0.09) (63) (0.02)1 2.02*** 2.02*** 1.93*** 744*** 0.24*** (0.06) (0.06) (0.08) (54) (0.01)2 1.09*** 1.07*** 0.95*** 303*** 0.12*** (0.06) (0.06) (0.07) (39) (0.01)3 0.77*** 0.73*** 0.58*** 184*** 0.08*** (0.05) (0.05) (0.07) (41) (0.01)4 0.54*** 0.49*** 0.36*** 51** 0.05*** (0.05) (0.05) (0.06) (23) (0.01)5 0.52*** 0.47*** 0.35*** 35 0.06*** (0.05) (0.05) (0.06) (22) (0.01)6 0.40*** 0.36*** 0.23*** 44 0.03*** (0.05) (0.05) (0.06) (27) (0.01)7 0.40*** 0.36*** 0.24*** 13 0.04*** (0.04) (0.04) (0.05) (18) (0.01)8 0.38*** 0.34*** 0.21*** 18 0.03*** (0.04) (0.04) (0.05) (22) (0.01)9 0.22*** 0.18*** 0.07 8 0.01 (0.04) (0.04) (0.05) (19) (0.01)10 0.26*** 0.23*** 0.10** -4 0.02* (0.04) (0.04) (0.05) (19) (0.01)11 0.26*** 0.23*** 0.11** 19 0.02** (0.04) (0.04) (0.05) (23) (0.01)N 665,802 665,802 665,802 665,802 665,802R2 0.01 0.06 0.29 0.22 0.30

Controlvariables: Propcharacteristics? N Y Y Y YHHcharacteristics? N Y Y Y Y

HHfixedeffects? N N Y Y YYearfixedeffects? N Y Y Y Y

41

TableIV.

SpendingonImprovementandMaintenanceSurroundingaHomePurchase DependentVariable:HomeImprovementandMaintenance

MonthssincepurchaseLog

SpendingLog

SpendingLog

Spending Spending($) Spending>0?

-3 0.03 0.07 -0.02 1182 -0.02 (0.22) (0.22) (0.25) (1063) (0.03)-2 -0.09 -0.03 -0.08 40 -0.01 (0.07) (0.07) (0.10) (148) (0.02)-1 0 0.05 -0.01 149 0.00 (0.05) (0.05) (0.08) (98) (0.01)0 0.75*** 0.81*** 0.77*** 485*** 0.11*** (0.06) (0.06) (0.08) (95) (0.01)1 1.06*** 1.10*** 1.01*** 479*** 0.15*** (0.06) (0.06) (0.08) (80) (0.01)2 0.72*** 0.75*** 0.63*** 301*** 0.10*** (0.05) (0.05) (0.07) (74) (0.01)3 0.57*** 0.60*** 0.41*** 197** 0.07*** (0.05) (0.05) (0.07) (82) (0.01)4 0.40*** 0.42*** 0.25*** 123* 0.04*** (0.05) (0.05) (0.06) (67) (0.01)5 0.32*** 0.34*** 0.18*** 97 0.03*** (0.04) (0.04) (0.06) (65) (0.01)6 0.23*** 0.25*** 0.08 163 0.01 (0.04) (0.04) (0.06) (136) (0.01)7 0.24*** 0.26*** 0.11** 43 0.02** (0.04) (0.04) (0.05) (52) (0.01)8 0.16*** 0.19*** 0.03 19 0.01 (0.04) (0.04) (0.05) (39) (0.01)9 0.19*** 0.21*** 0.07 67* 0.01 (0.04) (0.04) (0.05) (38) (0.01)10 0.18*** 0.20*** 0.05 -13 0.01 (0.04) (0.04) (0.05) (29) (0.01)11 0.17*** 0.19*** 0.04 70 0.00 (0.04) (0.04) (0.05) (48) (0.01)N 665,802 665,802 665,802 665,802 665,802R2 0.00 0.03 0.27 0.23 0.29

Controlvariables: Propcharacteristics? N Y Y Y YHHcharacteristics? N Y Y Y YHHfixedeffects? N N Y Y YYearfixedeffects? N Y Y Y Y

42

TableV.

TotalSpendingonDurables,Improvement&MaintenanceSurroundingaPurchase DependentVariable:AllDurables,HomeImprovement&HomeMaintenance

MonthssincepurchaseLog

SpendingLog

SpendingLog

Spending Spending($) Spending>0?

-3 -0.05 -0.24 -0.69** 818 -0.08** (0.27) (0.26) (0.31) (1105) (0.04)-2 -0.82*** -0.85*** -1.00*** -350 -0.14*** (0.10) (0.10) (0.13) (269) (0.02)-1 -0.48*** -0.55*** -0.64*** 19.48 -0.10*** (0.07) (0.07) (0.11) (201) (0.02)0 1.49*** 1.42*** 1.38*** 1,462*** 0.10*** (0.07) (0.06) (0.10) (188) (0.01)1 1.70*** 1.58*** 1.48*** 1,323*** 0.11*** (0.06) (0.06) (0.09) (161) (0.01)2 1.11*** 0.96*** 0.82*** 659*** 0.07*** (0.06) (0.06) (0.08) (143) (0.01)3 0.91*** 0.72*** 0.52*** 452*** 0.05*** (0.06) (0.06) (0.08) (154) (0.01)4 0.66*** 0.46*** 0.28*** 164 0.04*** (0.06) (0.05) (0.07) (150) (0.01)5 0.59*** 0.40*** 0.22*** 52 0.03*** (0.05) (0.05) (0.07) (117) (0.01)6 0.44*** 0.26*** 0.09 61 0.01 (0.05) (0.05) (0.07) (166) (0.01)7 0.43*** 0.25*** 0.11* 118 0.02 (0.05) (0.05) (0.06) (119) (0.01)8 0.41*** 0.23*** 0.08 -50 0.01 (0.05) (0.05) (0.06) (93) (0.01)9 0.37*** 0.19*** 0.07 66 0.01 (0.05) (0.05) (0.06) (99) (0.01)10 0.35*** 0.17*** 0.03 -112 0.01 (0.05) (0.05) (0.06) (93) (0.01)11 0.35*** 0.18*** 0.04 73 0.01 (0.05) (0.05) (0.05) (105) (0.01)N 665,802 665,802 665,802 665,802 665,802R2 0.01 0.13 0.42 0.18 0.46

Controlvariables: Propcharacteristics? N Y Y Y YHHcharacteristics? N Y Y Y YHHfixedeffects? N N Y Y YYearfixedeffects? N Y Y Y Y

43

TableVI.BuildingPermittingActivitySurroundingaHomePurchase

Quarterssincepurchase DependentVariable:BuildingPermits

-8(Omitted)Log

JobCostLog

JobCostLog

JobCost JobCost($) Anypermit?

-7 -0.006 -0.004 -0.005 -7 -0.001 (0.000) (0.000) (0.000) (5.283) (0.000)-6 -0.010 -0.006 -0.007 2 -0.001 (0.000) (0.000) (0.000) (5.456) (0.000)-5 -0.007 -0.004 -0.005 48 -0.001 (0.000) (0.000) (0.000) (5.625) (0.000)-4 0.002 0.003 0.002 146 0.000 (0.000) (0.000) (0.000) (5.841) (0.000)-3 0.011 0.015 0.014 268 0.002 (0.000) (0.000) (0.000) (6.057) (0.000)-2 0.022 0.027 0.026 288 0.006 (0.000) (0.000) (0.000) (6.029) (0.000)-1 0.030 0.035 0.034 29 0.012 (0.000) (0.000) (0.000) (5.513) (0.000)0 0.241 0.245 0.244 763 0.055 (0.001) (0.001) (0.001) (7.057) (0.000)1 0.181 0.188 0.187 835 0.035 (0.001) (0.001) (0.001) (7.255) (0.000)2 0.094 0.104 0.103 561 0.019 (0.000) (0.001) (0.001) (6.828) (0.000)3 0.063 0.073 0.072 365 0.013 (0.000) (0.000) (0.001) (6.458) (0.000)4 0.043 0.054 0.053 223 0.009 (0.000) (0.000) (0.001) (6.137) (0.000)5 0.023 0.038 0.037 135 0.006 (0.000) (0.000) (0.000) (5.895) (0.000)6 0.010 0.029 0.028 95 0.005 (0.000) (0.000) (0.000) (5.814) (0.000)7 0.007 0.026 0.026 78 0.004 (0.000) (0.000) (0.001) (5.770) (0.000)8 0.003 0.023 0.023 63 0.003 (0.000) (0.000) (0.001) (5.708) (0.000)N 194,786,763 194,786,763 194,786,763 194,786,763 197,457,069R2 0.00 0.00 0.05 0.03 0.06

Yearfixedeffects? N Y Y Y YPropertyfixedeffects? N N Y Y Y

44

TableVII.HeterogeneityintheSpendingResponsetoHomePurchaseCumulativeDollarSpendingAroundHomePurchaseinMontht

(Sumofmarginalspendingresponses,t-3andt+12)

TimePeriodorSampleRestriction HomeDurables

HomeImprovement

andMaintenanceTimePeriod Housingboom(2001-2006) 2,544 3,069Housingbust(2007-2013) 1,923 4,699

HomeSize TwoorfewerBedrooms 1,596 2,447ThreeBedrooms 2,499 2,075FourormoreBedrooms 2,566 5,902

HomeValue BelowMedian(≤$175,000) 1,583 408AboveMedian(>$175,000) 3,114 6,125

HomeAge Lessthanthreeyearsold 6,139 7,198Threetotenyearsold 2,853 2,658Tenormoreyearsold 1,317 3,127

Residencyatpurchasedproperty? Yes;primaryresidence 2,340 3,400No;vacationorinvestmentproperty 1,486 801

HomeownerAge Thirty-fiveoryounger 1,462 1,901Thirty-sixorolder 2,964 4,517

HouseholdIncome BelowMedian(≤$57,000) 1,799 1,806AboveMedian(>$57,000) 2,740 4,645

45

TableVIII.AggregateEffectofHomePurchaseChannelinGreatRecession

Home Home HomeTotal

Durables,

HomeNon-Home Total Improve- Mainten- Improvements Improvements

Durables Durables Durables ments anceand

Maintenanceand

Maintenancet2000–2005 8.9% -0.6% 1.4% 8.7% 9.9% 8.8% 3.7%2001–2005 8.5% -0.6% 1.6% 7.4% 61.9% 8.4% 3.8%2002–2005 8.8% -0.8% 1.9% 9.3% -139.9% 10.9% 4.8%2005–2008 35.0% -1.3% 3.6% 19.9% 32.3% 21.0% 9.1%2005–2009 23.1% -1.4% 3.4% 13.3% 37.4% 14.5% 7.9%2005–2010 33.5% -1.0% 2.7% 13.3% 16.7% 13.7% 6.6%2005–2011 39.3% -1.3% 3.3% 15.7% 9.7% 14.5% 7.8%

Note: Each cell reflects the fraction of the total change in a consumption category for a given time periodexplainedbythehomepurchasechannel,computedbymultiplyingthechangeinhomesalesforthetimeperiodby our preferred estimate of the dollar amount of consumption associatedwith a homepurchase. For thepreferred estimate and the consumption time seriesused in thedenominator, theCEdata is deflatedby acategory-leveldeflatorfromtheCPI(fordurables),NIPA(forhomeimprovements,andthissamedeflatorisalsousedformaintenance,whichdoesnothaveitsowndeflator),oracombinationofCPIandNIPA(forthelast column). This is done at themonthly level and aggregated to the annual level. Each column reflects aconsumptioncategory,whileeachrowreflectsthetimeperiodoverwhichchangesarecomputed.ThedataonthetotalchangeinaconsumptioncategoryforagiventimeperiodareaggregatescomputedusingCEweights.ThesalesseriesiscreatedbycombiningNationalAssociationofRealtors(NAR)dataonexistinghomesaleswithCensusdataonsalesofnewsingle-familyhomes,bothtakenfromFRED(notethattheNARdataisnolongeronFREDbutcanbeobtainedfromtheNAR).

46

TableIX.HeterogeneityintheSpendingResponsetoHomePurchase

Dependentvariable:

∆Log(Home-relatedretailspending)

TimePeriod: 2002-2007 2007-2012

∆Log(HomePurchases) 0.26*** 0.12*** (0.06) (0.02)

∆Log(HomePriceIndex) 0.29*** 0.23*** (0.03) (0.04)

Yearfixedeffects? Y Y N 246 267R2 0.44 0.28

Note:Wereportestimatesoftheelasticitiesofretailspendingtohomepurchasesandhomeprices.Thedataonannualspending,whichareatthemetropolitanarealevel,comefromthe2002,2007and2012EconomicCensus.Thedataonhomepurchasesandhomeprices,whicharealsoatthemetropolitanarealevel,comefromCoreLogic.Wemeasurerealspendingandrealhomepricesin2012dollars,usingtheCPI-Uasthepricedeflator.Weestimatethemodelwithordinaryleastsquares,weightedbypopulation,andclusterobservationsbymetropolitanareawhencalculatingstandarderrors,whicharereportedinparentheses.

47

FigureI.HomeSalesandHome-RelatedSpendingintheGreatRecession

Note:Bothseriesarenormalizedbytheirmaximumvalue.Homesales,inblueontheleftaxis,arethesumoftheNationalAssociationofRealtors’existinghomesalesseriesandtheCensus’sseriesofnewhomesales.A12-monthmovingaveragecenteredattheindicateddateisshowntosmoothoutseasonality.Homedurables,improvement,andmaintenance,inredontherightaxis,isthesumofthesecategoriesfromtheConsumerExpenditureSurveyaggregatedupbythesurveyweightsandnormalizedto2009dollarsusingthecategorypriceindex.

.7.8

.91

Hom

e D

ur, I

mpr

, and

Mai

nt, P

eak

= 1

.4.6

.81

Sale

s, P

eak

= 1

2000 2005 2010 2015Date

Home Sales Home Dur, Impr, and Maint

48

FigureII.HomeDurablesImpulseResponsetoHomePurchase

49

50

FigureIII.HomeImprovementandMaintenanceImpulseResponsetoHomePurchase

51

52

FigureIV.TotalDurable,Improvement,andMaintenanceSpendingImpulseResponse

53

54

FigureV.Non-HomeDurables,Food,andEntertainmentResponsestoPurchase

55

56

FigureVI.SummaryofCumulativeEffectofHomePurchasebyCategory

57

AppendixA:TheRelationshipBetweenHousePricesandSalesVolume

Theliteraturehasfocusedlargelyonhowchangesinhousingwealthdrivenbyprice

movementsaffectsconsumption.Thehomepurchasechannelwedescribeiscomplementary

tothischannelbecausechangeinhousepricesarehighlycorrelatedwithchangesinsales

volume.

Figure A.1 shows the relationship between severalmeasures of home prices and

homesalesfrom1968to2013.Thepriceseries,whichincludetheNationalAssociationof

Realtors(NAR)mediansalesprice(1968–2013),theCoreLogicnationalhousepriceindex

(1976–2013),andtheFederalHousingFinanceAuthorityalltransactionspriceindex(1975–

2013), are all aggregated to quarterly, seasonally adjusted (by the data producer), and

adjustedforinflationusingtheCPIandnormalizedtoonein1990.Thesalesseriesisthe

NARsingle-familyhomesalesseries(1968–2013).Weuseonlysingle-familysalesrather

thanallsalesbecausesingle-familysalesareavailableforamuchlongerpanelthanallsales.

Tominimizeseasonalityinthisseries,whichisnotseasonallyadjustedbythedataprovider,

weshowaone-yearaveragearoundtheplotteddate.PanelAofFigureA.1showstheseries

inlevels,whilepanelBshowstheseriesinannuallogchanges.

Thepriceseriesarehighlycorrelatedwith thesalesvolumeseries,althoughsales

volumeissomewhatmorevolatile.Indeed,thecorrelationofthepriceserieswiththesales

seriesrangesbetween0.75and0.86,dependingontheindex.Beyondthiscorrelation,the

figurerevealsthatinboththeearly1980sandthelate1980s,whennationwidehouseprices

fellmodestly, and in theGreatRecession, salesvolume fellmuchmore thanprices. Sales

volumeappearstoleadpricechanges,butonlybyafewquarters.Inlogchanges,thereisstill

a fairly strong correlation between sales and prices, although sales volume is far more

58

volatile.Nonetheless,inannuallogchanges,thecorrelationsbetweenthesalesseriesand

pricesare0.47fortheNARmedianpriceseries.0.39fortheCoreLogicPriceIndex,and0.43

fortheFHFApriceindex.

FigureA.1.RelationshipBetweenPriceandSalesVolume

A.PriceVersusSalesVolumeinLevels

.51

1.5

2Sa

les

Volu

me

/ Pric

e, 1

990

= 1

1970 1980 1990 2000 2010Quarter

NAR SF Sales, 1yr Ave NAR Real SF Median PriceCL Real Price Index FHFA All Trans Real Price Index

59

B.PriceVersusSalesVolumeinLogChanges

−.4

−.2

0.2

.4Lo

g Ch

ange

in S

ales

Vol

ume

/ Pric

e

1970 1980 1990 2000 2010Quarter

NAR SF Sales, 1yr Ave NAR Real SF Median PriceCL Real Price Index FHFA All Trans Real Price Index