[PDF] The financial transmission of housing bubbles: evidence from Spain

View - Download The financial transmission of housing bubbles: evidence from Spain

Previous PDF

Next PDF

Working Paper Series

The financial transmission

of housing bubbles: evidence from Spain

Alberto Martín, Enrique Moral-Benito,

Tom Schmitz

Disclaimer: This paper should not be reported as representing the views of the European Central Bank (ECB). The views expressed are those of the authors and do not necessarily reflect those of the ECB.

No 2245 / February 2019

Abstract

Howdo housingbubbles aectother economicsectors? We show thatin thepresence ofcollateral

constraints,a bubbleinitially raiseshousing creditdemand andcro wdsout creditto non-housing τrms.If

the bubblelasts, how ever,housingcreditrepayments raisebanks? netw orthandexpandcreditsupply,so that crowding-outeven tuallygivesway tocrowding-in.This isconsistentwithevidencefrom therecen t Spanish housingbubble. Initially, creditgrowth ofnon-housing τrmswaslower atbanks withhigher bubble exposure,andτrms relying onthese banksexhibitedlow ercre ditand outputgro wth.During the bubble?s lasty ears,theseeects reversed. Keywords:Housing bubble,Credit, Inv estment,FinancialFrictions,FinancialTransmission, Spain. JEL Codes:E32, E44,G21.ECB Working Paper Series No 2245 / February 20191

Non-technical summaryDuringthelastdecades,manycountriesexperiencedmassiveboomͲbustcyclesinhouseprices.

house enough-crowditin. crisis.Thesameresultsholdforvalueadded.

highlightingtheroleofthefinancialsystemasatransmissionmechanismbetweensectors.ECB Working Paper Series No 2245 / February 20192

1 Introduction

During thelast tw o decades,man yeconomiesexp eriencedmassive boom-bustcyclesinhouse prices.These housing bubblesη,whic hoccurredin theUnitedStates, theUnited Kingdom,Spain andIreland,andma y beongoing inChi na,are widelybelieved toha ve importanteectsnotonly forthe housingsector,butalso for thebroadereconom y(see Zhu,2014and Jordàet al.,2015a).Thus, understanding thec hannelsthrough whichthey aectother economicsectors hasb ecomea key concernfor economistsandp olicymakers. In thispap er,weanalyze thetransmissionof housingbubbles through thecreditmarket. Despiteits im- portance,the roleof thismark etis apriori unclear.Ontheone hand,it hasb eenargued thathousing

bubbles raisethedemand formortgages andcredit toreal estateand constructionτrms, reallocating credit

towardshousingat theexp enseof non-housingτrms (e.g.,Chakraborty etal., 2018).On theother hand,

housing bubbleshav ealsobeeniden tiτedas asourceofcreditb oomsextendingto allsectors ofthe economy ,

including non-housingones (e.g.Jimene zet al.,2014). Our papermakes twomain contributions.First,we constructa macroeconomicmodelofhousing bubblesand showthat theyha ve conictingcrowding-outandcro wding-ineectsthroughthe creditmarket. Crucially, these eectspla youtatdieren tmome nts intime.Whilea housingbubbleinitiallycrowdsout non-housing credit andin vestmentbyreallocatingcredit tothehousing sector,iteven tuallycro wdsthem inb yraising the netw orthofthebanking sectorand thus creditsupply .Second, we useadetailedbank andτrm-lev el

database tosho wthatthesetheoretic alpredictions arein linewiththeSpan ishexp erienceduring therecen t

housing boomandbust. Theseτndings implythatthecon trastingviews outlinedab ov eare notm utually exclusive,but instead describet wophasesofthesame phenomenon.

Our theoreticalanalysis isbased onan ov erlapping-generations, small- openeconomythatproduces twogoods,

housing andnon-hou sing.Theeconom yis populatedbyhousingen trepreneurs,non-housing entrepreneurs,

and bankers.Inorder toin vest incapital, entrepreneursfromboth sectorsb orrow frombankers,which inturn

borrowfrom aninternationalτnancial market. Crucially,we assumeth attheborro wingofentrepreneurs and

bankersis limitedb ya collateralconstraint,as theycannot crediblypromise torepaymore thana fraction of theirfuture income totheircreditors. Housing entrepreneursareendow edwith land,whichisused inhousin gproductionand tradedinacomp etitive market.A tanyp ointin time,thefundamentalvalueofland isthe present value ofitsfuturemargin al products.The market valueoflan d,howev er, mayexceedthis valueifitcontainsrationalbubbles.incertain periods,housingen trepreneursma ybewillin gtobuyland atapriceexceeding itsfundamentalv alueb ecause they expecttor esellit atahighprice inthe future.W erefe rto such episodes ash ou singbubbles.

ECB Working Paper Series No 2245 / February 20193

When ahousing bubblestarts, we findthat itat firstcro wdsout creditandinv estment in thenon-housing sector. Indeed,thebubble raises thecollateral ofhousingentrepreneurs andenables themto expandtheir

credit demand.As creditsupply isunaffected, thisincreases thedomestic inte restrate andreallo catescredit

(and investment)fromthenon-housingto thehousing sector.As timepasses and thebubble contin ues, however,crowding-outgiv esway tocrowding-in.The reasonisthattherepaymen tof housingcrediteven tually raises then etworthof bankers,which enables themtoborro wmorefrom theinternationalfinancial market and toexpand thedomestic credit supply .Thus,the riseintheinterestrate isreversed,and creditand investmentinthenon-housing sectorstart toincrease. Ifthe bubblelasts longenough, we show thatthis

crowding-ineffect outw eighstheinitialcrowding-out effect:thehousingbubble endsu praisingcreditto all

sectors, includingthe non-housingone. However,arationalbubble isonly sustained by theexp ectationthat landpricescontin uetobe highin the future. Whenthese expectations change,thebubble burstsandlandprices collapse. Thiswip esout the collateral ofhousing en trepreneursandlowerstheir creditdemand. Italsoreducesloanrepaymentsreceiv ed bybank ers,therebyreduc ingtheirnetw orthandcontracting theircre ditsupply .Jointlyconsidered, these

effects triggera suddenstop inb orrowing fromthe internationalfinancialmarket,an increase inthe domestic

interestrate, anda fallin credit andin vestmen tbothin thehousing andnon-housingsectors. The mainprediction ofou rmo delisthenon-monotonic patternofcredit innon-housing sectors:a housing bubble firstlo wersnon-housingcredit,but eventually raisesitagain.In orderto testthisprediction,we use datafrom themassiv eb oom-bustcyclein Spanishhousingpricesbet ween 1995an d2015.Thisc ycleis generally interpretedasthe resultof ahousing bubble,and thereforepro videsan ideallab oratoryfor our model.Ho wever,itisimportant tostress thatour maintheoreticalpredictionisnot specific tobubbles: the same patternof crowding-in andcrowding-outeffects couldarise inhousingcyclesdriv enb ypro ductivity

shocksorfinancial innov ations(e.g., changesintheextent towhic hincomecan be pledgedto bankers).Thus,

our empiricalanalysis isnot designedto establish whether thehousing cycleinSpain wasdriven by ab ubble,

but ratherto understand thefinancialtransmissionof thatcycle tothe non-housingsector. The boom-bustcycleinSpanish housing was spectacular. Between1995 and2008,Spainexp erienceda threefold increaseinhousing pricesand inthe num ber ofnew housesbuilt. In2008,thisboomgav ew ay to a prolongedbust: by 2015,housepriceshad fallenb ya thirdfrom the2008 peak, andtherewere essentially no newhouses being built.

1The housingbub blewasaccompanied byacredit andin vestmentb oom, anda

surge incapital inflows. Itsburstcoincided witha longand deeprecession(Baldwinet al.,2015). 1 These developmentsarediscussedin greaterdetailinSection 2.

ECB Working Paper Series No 2245 / February 20194

While ourmo delisconsisten twith mostof theseaggregate developmen ts,w eaimtotestitmore directly, byconfron tingitsmainpredic tionswith micro-level datafromtheSpanish Credit Registry(whichcon tains

information onvirtually allloans tocommercial firms).Our empiricalstrategy relieson thefact thatnot all

banks wereequallyexp osedto thebubble,astheir business models didnot assign thesame importanceto housing. Usinga simpleextension ofou rmo del withheterogeneousbanks,w eshowthatthecro wding-out

and crowding-ineffectsshould thenb eobs erved atthebanklev el.Credittonon-housingfirmsshould initially

growless athighly exposed banksrelativ etolessexp osedones, butthispatternshould reverse inlater years

and creditto non-housingfirms shouldactu allygro wmore atmoreexpos edbanks.Figure1 shows thatthis

prediction isin linewith theevidence, by plottingthe evolution oftotalcreditto non-housingfirms forthe

Spanish banksin thehighest andlo wes tdeci lesofourbaselinemeasureofexp osuretothehousin gbub ble. 2

In thefirst years afterthestartof thehousing bubblein thelate 1990s,credit tonon-housing firmsgrew less

in themost expos edbanks.However, thispattern eventuallyreversedandto wards theend ofthebubble, credit tonon-housing firms hadactuallygrown morein themost exposedbanks. Figure 1:Credi ttonon-housingfirms indiffere nt banks 0 200
400
600
800

1000199619982000200220042006200820102012

High exposed banksLow exposed banksSource: CIRandauthors'calculations (seeSection 5for details).High (low) exposed banksareabo ve(belo w)the 90th(10th)

percentileofthe exposure measure.Both seriesarenormalizedto 100 in1995 .Dashed linesare HPtrends ofthe originalseries.

While thepattern shown inFigure1is suggestive, itma yb edue tosystematicdifferencesb etw eentheclients

of highand low- exposurebanks.Tocontrolforthese factors,w eregressannual creditgrowthof non-housing

firms atthe loanl evel (thatis,foranyban k-fi rmpair) onbank exposuretothe housingbubbleandfirm-time2

Throughout, wedenenon-housing rmsas rmswhichdo notb elo ngtotheconstructionorrealestatesectors. Ourbasel ine

measure ofexp osuretothehousing bubbleis thebank's ratioof mortgage-back edcredit tototal creditb etw een1992 and1995,

beforetheb eginningof thebubble (datedb ymost observers,suchasfor instanceF ernández-Villaverdeet al.(201 3),inthe

second halfof the1990s). Thismeasure capturesa bank'sabilit yto assessreal estatecollateral. We considertwo alternative

measures ofexp osureinourempi ri calanalysis, andresultsareunchanged.ECB Working Paper Series No 2245 / February 20195

fixed effects,follo wingKhwa jaand Mian(2008). Firm-timefixed effectscon trol forfirm-lev elshocksto credit

demand. Coefficientsarethus identified bydifferences inthecreditgro wthofthe samefirmwithmore orless

exposedbanks, andonly reflectc hanges atthe bank-level.Theseregressionsconfirmthe model"s predictions:

for thesame firm,ann ualcre ditgrowthis significantlylo weratmore expose dbanksduringthefirsty earsof

the bubble,but thenb ecomessignifican tlyhigheratthese banks.We alsopro videevidence thatthe eventual

crowding-inis driven bychanges inbanknetw orth,aspredictedb ythe model. Finally ,duringthecrisis ,w e

showthat creditgro wthat moreexposedbanks againb ecomessignifican tlylower.

Toconclude, we extendourempiricalanalysis tofirm-lev elou tcomes.If non-housing firmscoul dfreelyswitch

across banks,cro wding-outandcrowding-in shouldb eaggregatephenomenaaffe ctingall non-housingfirms

in thesame wa y.Iffrictionspreventfirms fromswitc hing bankseasily,howev er,the creditgro wthofaspecific

firm islik elytodepend onits pre-existingbankrelationships. To test whetherthesefrictionsare empirically

relevant,we regressannual creditgrowthat thefirm-levelon aw eightedav erageof housing bubbleexposure of allbanks fromwhic hthe firmborrows. We findthat indeed,firmsthatborrow morefrommoreexp osed banks experiencedlow ercreditgrowthduring thefirstyears ofthebubble, higher credit growth initslast years,and low ercreditgrowthduringthe crisis.These resultsareconfirmedwhen we considerv alueadded growthinstead ofcredit growth, sho wingthatthedifferences increditgrowthhad real effects.

Our paperisr elatedto severalstrandsof literature.First, itrelatestoa growi ngb ody ofw orkstudying the

role ofhousing formacro economicfluctuations (seeIacoviello(2010) andPiazzesi andSc hneider(2016)for tw o

recentsurv eys).Mostofthis literatureanalyzes consumptiondynamic s.F orinstance, Kaplanet al.(2017) recentlysho wedthatbelief-driven changes inhousepricesaccount foralargepart ofaggregate fluctuations in theUnited States,mainly throughw ealtheffects inconsumption (theempiricalimportance ofwhic his

also underlinedb yMianandSufi, 2011).Guerrieri andUhlig (2016)analyze thege neralrelations hipb etw een

housing andcredit bo oms.Wecomplementth isworkb yfocusingon thetransmissionofhousing boomsto the restof theeconom ythrough thecreditmarket. Severalempirical papers havestudied theeffectofhousepriceson credit andin vestmen t.Some provide

evidence fora positiv eeffectthroughacollateralc hannel,with higherhouse pricesincr easingth evalueof

corporateheadquarters forlisted firms(Chaney etal. ,2012) andof private homesforen trepreneurs(A delino

et al.,2015; Bahaj etal.,2017).Jimenez etal. (2014)argue fora positiv ee ffectof housing onbanks" credit supplyduringthe Spanish bo om,drivenb ymortgagesecuritization.Ontheother hand,Chakrabort y et al.(2018) show thatbankswhic hw ere moreexposedtotheUS housingboomreducedtheir loansto

firms, asmortgages crowde doutcorporatecredit.Fin ally, HernandoandVillanuev a(2014) andCuñatetal.

ECB Working Paper Series No 2245 / February 20196

(2016) showthatb anksthat were exposedtohousingreduc edth eirl endingacrosstheb oardwhenhousing prices fell,b othintheUnite dStates andi nSpain.Ourpap ersho wsthat thesefindingsarenot mutually

incompatible, butcapture different phasesofthetransmission ofa housingbubble throughthe creditmark et.

Most importantly,weshow thatthecrowding-out ofnon-housingcr editdo cumented byChakrabort yet al. (2018) eventuallygivesw aytoc rowding-in.Whilethelatterfinding isin linewithJimenezetal. (2014),w e argue thatcro wding-inisdriven by anincreaseinbank networthrather thanb yaccess tosecuritization, and providesomesuggestiv eevidence inthisdirection. More generally,ourpap errelates toasmallbut growing literatureemphasizing therole ofthe financial system forthe transmissionof sectoralsho cks. For instance,BigioandLa"O(2016) useanetw orkmodelto study howsectoralfinancial shoc ksprop agatethroughtheeconomy .Bustosetal. (2017)sho wthatBrazilian banks thatare moreexp osed toregionsexperiencinganincrease inagricultural profitse xpandtheirlending to non-agriculturalfirms elsewherein thecoun try. Thisfinding issomewhatreminiscentofour crowd ing- in effect,b ywhichbanks thataremoreexp osedto the housingbubble even tuallyexpandtheir lendingto non-housing firms. The theoreticalmo delinthispap erbuilds onMartín andVentura (2012),who develop aframew orkfor

analyzing thein teractionbetw eenrationalbubblesandcreditwhenthe formerpro videcollateral.Martínand

Ventura(2015)exten dthis modeltoan open-econom ysetting,anduse itto studythe relationshipbetween bubbles, creditand capitalflo ws.With respecttotheir work, ourmodeladds financialin termediaries, multiplesectors andbank heterogeneity ,enabling ustostudytheroleof banknet worth inthe propagation of sectoral(bubble) shoc ks.

3In thisregard, ourmo delis thefirsttostudy thetransmission ofsectoral

bubbles throughfinancial intermediaries. Finally,our paper addstothelarge literatureon creditb ooms andbusts (includingJordà etal., 2015b; Mendoza andT errones,2008,2012;Reinhart andRogoff, 2009,2014). Thesestudies do cument thatcredit boomstendto be accompaniedb ycapitalinflo wsand risinghouseprices, andincrease theriskoffinancial crises. Ourpap erisconsistent withthese stylizedfacts,and provides additionaldetails onSpain.The

Spanish experienceitselfhas alsob eenthe focus ofextensive researc h(see,forinstance, Fernández-Villa verde

et al.,2013; Akinet al.,2014; Santos, 2017a,2017b), inv estigatingtheorigins ofthe housingbubbl e,the

driversof capitalinflo wsand theflawsof the Spanishbankin gsystem.Whilew ebuild onsomeofthe insigh ts

of thesestudies, we donotaimto provide aunified narrative forSpain"s economicdevelopment duringthe 3

Other relatedmodelsincludeArce andLóp ez-Salido(2011),whostudy thein terac tionof housingbubbles withcollateral

constraints,andBasco (2014),w hostudies therelationship ofbubbles withτnancial liberalization. Ven tura(201 2)studiesthe

interactionb etweenbubblesandcapi tal ows, butinhissetting,bubblesaectthecostofcapital andnotthe stoc kof credit.

Den Haanet al .(2003)proposeamo delof macroeconomicuctuationsin whic hlendersareτnancially constrained.ECB Working Paper Series No 2245 / February 20197

period.For instance,wedo notin vestigatewhetherthehousing bubblew ascausedby thefall inSp anishreal

interestrates afterthe creationof theEuro, ordecomp oseaggregate dynamics tosee which partisexplained

bymo vementsintherealinterestrate andb ythe housingbubb le.

4Instead, wetake thehousingbubble as

givenand focus onitstransmissionto the restof theeconom ythrough thecreditmarket. The remainderof thep aper isstructuredasfollows. Section2 providessomebac kgroundinformation about the Spanishb oomandbust.Section3 setsout ourmo delof housing bubbles, andS ection4 illustratesits

results andpr edictions.Section5tests thetheoretical predictions withmicro data,and Section 6concludes.

2 TheSpanish bo omandbust

2.1 Thehousing bubble

In themiddle ofthe 1990s,according toJimeno andSan tos(2014, P. 128)," the Spanishec onomy[had] developedsomechar acteristicsthat madeitespe cially prone fora housingbubble": thebanking sectorw as

able toattract capitalinflo ws, constructionfirmshadbuiltuplarge capacitiesduring earlierinfrastructure

projects,and theSpanish population was youngandgro wingfast.Risinghouse pricesweresustained by

changesin zoningand landuse regulationsin 1997and 1998(whic hdecen tralizedand liberalized thegran ting

of housingp ermits),andweak lendingstandards, especiallyin regionalbanks subject tocapturebylocal

politicalelites (Fernánde z-Villaverdeetal.,2013;Akinetal.,2014). Asa result,b othnominal house prices

and theconstruction ofnew housestripled bet ween 1995and 2008,asshowninFigure2.

Figure 2:Hous ePricesandHousing Construction

1995199 82001200 42007201 02013201 605

00100015002000

1995200 02005201 002

00000400000600000

€ per square meter

House Prices, 1995-2015Construction of new houses, 1991-2014Source: Ministryof Construction.See Appendix Bfor furtherdetails. 4

Wealso abstractfrom therising misallocation ofcapital duringthe perio d,do cumentedbyGarcía-San tanaetal.(2016)

and Gopinathet al.(2017), andp otentially responsible forSpain'slowaggregatepr oductivitygro wth.Basco etal. (2017)argue

that thehousing bubblew aspartly responsibleforthe increase incapitalmisallocation, asto oman yresourceswere channeled

to unproductivermswithhigh realestate collateral,esp eciallyin municipalit ieswith fast-growing housingprices.ECB Working Paper Series No 2245 / February 20198

The boomwasfollo wed byasp ectacularcollapse. Pricesfellsixy earsin aro w,and inthe yearsb etw een

2010 and2014, yearly housingconstructionrepresented only6% ofthe pre-crisispeak.In thenext section,

wep rovidesomefurtherdetailson themacro economic context ofthis housing cycle.

2.2 GDP,credit,and capitalino ws

Between1995and 2008,Spain experien cedan economicb oom,withthereal GDPof thebusiness economy increasing ona verageby3.8%p eryear(see theleft panelofFigure3). Thiswasfollo wed by adeep crisis during whichrealGDP fellfiv ey earsin arow.The expansionsaw acredit bo om,bothin mortgagecredit to householdsand incredit tofi rms.

5The rightpanelof Figure3 illustratesthe latterp oint by plottingth e

ratio offirm creditto business-econom yGDP ,showingthatthislev erageratiodoubledb etween1995 and

2008. Leveragecontin uedtoriseuntil2010 (ascredit fellmore slowlythanGDP), be foredelev eragingset

in. Thecredit bo omwasfinancedb ybanks,which channeledcapitalinflo wstofirmsand households.As a consequence, theexternal debtof Spanishbanks almosttripled bet we en2002 and2007. 6 Figure 3:Real GDPand leve rageof theSpanishbusinesseconomy, 1995-2014

2000200 52010201 580009

000100001100012000

19952000200520100.500.751.001.25Real GDP, 1996-2016Firm Credit-to-GDP ratio, 1995-2014Source. INEandBankof Spain.The rightpanelplots theratio bet ween creditto productiveactivitiesandbusiness economy

GDP (includingall sectorsexcept publicadministration, defense,so cialsecurit y, health,educ ation, artsandentertainment).

A simpleaccoun tingdecompositionsh owsthathousing sectordynamicscontributedsubstan tiallyto these aggregate developments,mostofall forcredit. Figure4 shows thatthe shareof housing (constructionand real estatefi rms)inbusinesseconom yGDP increasedsubstan tiallyduringtheb oom, from18% in1997to

25% in2007. Thecomp ositionc hangeforcredit,ho wever,w asm uch moreextreme:whilehousingmade

up 22%of firmcr editin 1995,thatsharehad increasedto 48%in 2007.Had the GDPshare ofhousing 5

This factdieren tiatesSpainfromthe contemporaneousexp erienceoftheUnitedStates,whereτrm leverage didnot increase

during thehousing bo om(MianandSuτ,2011).

6Statistical bulletinof theBank ofSpain, Series17.31 (https://www.bde.es/webbde/es/estadis/infoest/bolest17.html).ECB Working Paper Series No 2245 / February 20199

remained constantbet ween1995and 2007,and had leverageincreas edbythesamerate thanintherest of the economy,theov erallincrease intheSpanishfirmleverage ratiosho wnin Figure3 would onlyha vebeen

half aslarge (37instead of71 percen tagep oints). Furthermore,thelargeincreasein householdcreditduring

the boomwasalmost entirelydriven by mortgagelending.

The driversofSpain"s extraordinaryb oom-bust cycleha vebeenwidelydebated. Clearly, productivitywasnot

one ofthem, asS painactually experiencednegativ egro wthintotalfactorprodu ctivity(TFP) throughout, particularly soin thehousing sector.

7Instead, thefall inreal interest ratesafter thecreationof theEuro

and thehousing bubble itselfaregenerallyregarded asthe key drivers ofaggregate dynamics. Figure 4:Comp ositionofbusinessGDP andfirm credit, 1995-2014

19952000200520100.150

.180.210.240.270.30

19952000200520100.20

.30.40.5

Housing share of business GDPHousing share of firm creditSource: INEand Bankof Spain.See Appendix Bfor details.

These explanationsare ofcourse notm utuallyexc lusive .Ouraimin thispaperisnottojudgetheir relative importancein accounting forSpain"sbo om-bustcycle, orto provideanexhaustivepicture ofall channels through whichtheyma yha veaffectedthe restoftheeconomy. Instead,w estart fromthe premisethat Spain experiencedahousin gbub ble,andthenstudy itsspillover effectsto therest ofthe economy throughthe credit market.Inp articular,w eareinte restedinanalyzingwhether themassiv ecreditgrowthfor housing firms showninFigure 4slo wed down creditandinvestment growthin othersectors, orwhethe ritactually stimulatedthem. Inthe nextsecti on ,w epresentasimplemodel toaddressthese questionsinasystematic and rigorousmanner. 7

AccordingtotheEUKLEMS database( http://www.euklems.net/), constructionsector TFPde clinedb y24%b etween1995

and 2007.Ho wever,thepopularconceptionthatmisallocation ofcapital inowstothe low-pro ductivity constructionsector

caused theaggregate TFPdecline isinconsisten twith thedata. Indeed,housingaccounts onlyfor asmall partof theagg regate

decline, whichwas observedinvirtually allsectors(Fernández-Villa verde etal., 2013; García-Santa naetal.,2016).Gopinath

et al.(2017) argueinstead thatcapital inows were misallocated withinmanufacturingindustries, helpingonlynancially

unconstrained rms(rather thanthe mostpro ductiv eones) toexpand.ECB Working Paper Series No 2245 / February 201910

3 At wo-sectormodelofbubbles andfinancial intermediation

This sectiondev elopsamodel ofa smallopeneconom ywith tw osectors, housingandnon-housing.In both

sectors, entrepreneursborro wfromdomesticbankstoτnance capital accumulation. Banks,in turn,b orrow

from internationalτnancialmark ets.Crucially ,alllendingrelations hipsaresub jectto collateralconstrain ts.

Wemo delthehousing bubblebyin trod ucingan additionalasset,land. Landisusedinhousingproduction and heldb yhousingentrepreneurs, whocan usetheincomeit generatesas collateral.Imp ortantly ,land prices areprone touctuations driven by rationalbubbles.Whenland pricesincrease,sodo esthe value of housing entrepreneurs?collateral andthustheir creditdemand. However, theloan repaymen tssustained by this collateralev entuallyalsoraisethenet worth ofbanks, allowing them toincreasecredit supply. This interplaybet weencreditdemandandcreditsupply isat theh eartof ourtheoretical results. In themo del,boom-bust cyclesarethereforedriven byrationalbubbles. landprices may risebecause agentsexp ectthemtorise even morein thefuture, andmaycollapse because expe ctationsc hange.Thi s

modelingc hoiceprovidesa simpleway ofin troducin gpriceuctuationsthatarefully consistent withrational

expectations.How ever,ourresultswouldnotc hangeif landprices weredriven by otherfactors(forinstance, byb eliefandpreference shoc ks,asinKaplan etal.,2017).Note,moreov er,that housingb ooms inthe modelare driv enbythepriceof landandnotof structures.This isin linewith theempirical evidence. Indeed, Piazzesiand Schneider (2016,P.1560) notethat intheUnitedStates, movements inthe valueof the residentialhousingsto ckar emostlydueto movementsinthevalue ofland η.

3.1 Agents,preferencesand technologies

Agentsand preferencesTime isdiscrete (t2N). Weconsidera smallop eneconom yp opulatedby

generations ofagen tsthatlive fort woperio ds.Agen tsarerisk-neutralandderiv eutilityfromtheirold-age

consumption ofthe econom y?sτnalgood.Th us,for agentibornin perio dt, utilityisgiv enb y U i;t=Et(Ci;t+1),(1) whereCi;t+1denotes theconsumption ofagen tiin periodt+ 1. Eachgenerationof agents consistsof three types.housingen trepreneurs,non-housing entrepreneurs,and bankers. Weconsiderthroughoutsymmetric

equilibria inwhic hallagents ofa certaintype areid entical .This allowsustofocus, withoutloss ofgenerality,

on there presentativeagentforeach typeand generation. Agentsderiv etheirincomeeither fromtheir participationin thepro duction process orfrom theirrole in ECB Working Paper Series No 2245 / February 201911 credit intermediation.Therefore,w enext describe thepro ductionstructure. ProductionThe finalgo odisassembledb ycomp etitivefirmsf romtwo intermediategoods,housing (H) and non-housing( N), accordingto theCobb-Douglas production function Y t= (YN,t)τ(YH,t)1-τ;with2(0;1):(2) The finalgo odistradable,and we normalizeitspriceto 1. Intermediategoo ds,ontheotherhand,ar en ot tradable. LettingPN,tandPH,tdenote theprices ofthe intermediate goo dsinperio dt, costminimization byfi nalgoods producersimplies Y N,tY

H,t=1P

H,tP

N,t:(3)

Furthermore,perfect competitionimplies thatthe priceof thefinal goo dis equalto itsmarginalcost,sothat

PN,t

τPH,t1

1 = 1 :(4) Intermediatego odsarealso producedby perfectlycompetitiv efirms.These firmsuseaCobb-Douglas productionfunction combining capital,laborand land,giv enby Y j,t= (Lj,t)1-αj-βj(Kj,t)αj(Tj,t)βjforj2 fN;Hg;(5)

whereLj,tstands forthe labor employedb ysectorjin periodt,Kj,tfor itscapital stoc kandTj,tfor itsland

use.jandjare twopositive parameterssatisfyingj+j<1. Forsimplicity ,weassumethatN= 0, implying thatland isonly usedin housing production. Factorsupply All threepro ductionfactorsaresupplied by entre preneurs.Eac hgenerationof j-sector entrepreneursinelastically suppliesone unitof sector-spec ificlab orwhen young.Furthermore,young en- trepreneurs haveaccesstoa sector-specific inv estment technology ,whichallowsthemtocon vertoneunitof the finalgo odinperiod tintoon eunitoftheir sector"sc apitalin perio dt+ 1. Finally,y ounghousingentrepreneurs arealso endowedwith oneunit ofland, whichcanbeused inpro duction when theyare old.This impliesthat theaggregate stoc kof landgr ows overtime,asa newland "vintage" is added inev eryperiod. Weinterpretthisgro wthinthesto ckofland ascapturing thegrantingof construction

permitsto housing entrepreneurs.Ass hownby Fernández-Villa verdeetal.(2013),th iswasa keyfeatureof

ECB Working Paper Series No 2245 / February 201912 the Spanishhousing bo om,andit playsanimp ortant roleinourmodelas well.

Our assumptionsen tailthatallpro ductionfactors aresector-sp ecific.Thisiscon venien tb ecauseiteliminates

all directspillov ersofahousingbubblethrough factormark ets,and thus enablesus toisolate itstransmission

through thecredit market. However, factorspecificityisnotnecessary forour results. 8 Weassume throughoutthat capitaldepreciates fully, andthat landis productive for justone perio d.This last assumptionsimplifies themo del byensuringthatthe stockof productiveland atan ygiv enpointis constantand equal toone.Asw esho win Appendix A.1,noneofour mainresultsrelies onthis assumption. Factormark etsandequilibriumpro ductionof intermediate andnalgoo dsAs landandlab or

are specificfactors,Equation (5)pins down theoutput ofeac hsectorfora given level ofthe capitalstocks.

Factormark etsbeingcomp etitive,thew ageforeachtyp eof laborj2 fN,Hgequals itsmarginal product, w j;t= (∞αjβj)Pj;t(Kj;t)j,(6) where weha vealreadyusedthefactthat inequi librium,LN;t=LH;t=TH;t= ∞. Likewise,theren talrates of capitaland landare alsoequal totheir marginalpro ducts, r j;t=αjPj;t(Kj;t)j1,(7) whererj;tdenotes theren talrateofcapital insector j, and m t=βHPH;t(KH;t)H.(8)

wheremtdenotes theren talrateofland. Thus, summingup, fora givenlevel ofcapital stoc ksin both sectors,

Equations (2)to (8)join tlydetermine theproductionan dprice ofeac hintermediate good,the return tothe

quotesdbs_dbs14.pdfusesText_20

[PDF] indeed suisse

[PDF] indeed traduction

[PDF] indeed uk

[PDF] indemnisation catastrophe naturelle inondation

[PDF] indemnisation degat des eaux maif

[PDF] indemnisation maif accident

[PDF] indemnisation maladie professionnelle fonction publique

[PDF] indemnité d'éloignement nouvelle calédonie

[PDF] indemnité de caisse agent comptable eple

[PDF] indemnite de deplacement a l'interieur du france

[PDF] indemnité de licenciement ile maurice

[PDF] indemnité de licenciement france imposable

[PDF] indemnité de résidence nouvelle calédonie

[PDF] indemnité de stage controleur des finances publiques

[PDF] indemnité dégressive et ppcr