Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 57
* WethankClaesBerg,JensIversen,GoranKatinic,JessicaRadeschnig,CarolineRichardsandAndersVredinfortheirvaluablediscussionsandinput.TheviewsexpressedinthispaperrepresentourownperceptionandcannotbeassumedtoexpresstheRiksbank’sview.
MacroeconomiceffectsofreducinghouseholddebtDariaFinocchiaro,MagnusJonsson,ChristianNilssonandIngvarStrid*TheauthorsworkintheMonetaryPolicyDepartmentoftheRiksbank
The Riksbank has for a long time emphasised that rising household debt is a concern for financial stability that needs to be addressed. Tighter macroprudential measures or tighter mortgage interest deduction are two alternative ways of tackling this problem. In this article, we study how these two approaches would affect different households using a macroeconomic model. We show that, contrary to what is often argued in the public debate, a tightening of the loan-to-value cap, the loan-to-income cap and the amortisation requirements would lead to a redistribution of resources from lenders to borrowers in the long-run. Moreover, tighter mortgage interest deduction affects households in different ways, depending on how the Government chooses to use the released budgetary resources. If borrowers are compensated, this policy could have positive effects for their consumption other than housing. We also analyse the implications for monetary policy of different measures to dampen household debt and show that the extent of mobility on the housing market plays an important role. In some cases, monetary policy might need to be more expansionary, and in other cases more contractionary. Finally, we study how household debt affects the transmission mechanism of monetary policy on inflation. The higher the indebtedness, the greater the effects of a rate hike on the interest expense and disposable income of borrowers. The effects of a rate hike on demand – and hence on inflation – are therefore greater today than when the inflation target was introduced in the mid-1990s.
58 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
1Whyisitimportanttodampenhouseholddebt?Inthepast20years,householddebthasrapidlyincreasedinSweden.Overtheperiod1995-2015,householddebtinrelationtodisposableincome–theloan-to-incomeratio–increasedfromaround90percenttocloseto180percent,seeFigure1a.Whenonlytakingintoaccounthouseholdswithmortgages,theloan-to-incomeratioiscurrentlyevenhigher,ataround320percent.Swedishhouseholddebtisnotjusthighinahistoricalperspective,butalsoinaninternationalcomparison.Theloan-to-incomeratioofSwedishhouseholdsisamongthehighestintheOECDcountries,seeFigure1b.
Theincreaseinindebtednesshasbeencoupledwithincreasinghouseprices,seeFigure2a.Inthepast20years,priceshavealmostquadrupled.Severalfactorscouldexplainsuchanincrease.Forexample,relativelyfewhomeshavebeenbuiltinthepastfewdecades;moreover,thehomesthatdoexistareperhapsnotbeingusedefficientlyduetorentalmarketregulations.1Butthepricesmightalsohavebeenpushedupbytheincreaseinhouseholds’disposableincome,whileinterestrateshavebeenlowatthesametime.Theinterest-to-incomeratioofhouseholds–theirinterestexpenseinrelationtotheirdisposableincome–iscurrentlyatitslowestlevelinaround40years,seeFigure2b.
60
80
100
120
140
160
180
70 75 80 85 90 95 00 05 10 15
Figure 1. Households’ loan-to-income ratio in Sweden and in an international perspectivePer cent of disposable income (a) Loan-to-income ratio in Sweden (b) Sweden’s loan-to-income ratio in an international perspective
0
50
100
150
200
250
300
350
Note. The loan-to-income ratios in figure (b) refer to 2014.Sources: Statistics Sweden and OECD
Hung
ary
Slov
enia
Chile
Pola
ndSl
ovak
iaCz
ech
Repu
blic
Esto
nia
Aust
riaIta
lyGe
rman
yFr
ance
Belg
ium US
Gree
ceFi
nlan
dSp
ain
Japa
nPo
rtug
al UK
Sout
h Ko
rea
Cana
daSw
eden
Switz
erla
ndAu
stra
liaIre
land
Nor
way
Net
herla
nds
Denm
ark
1 SeeEmanuelsson(2015).
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 59
0
50
100
150
200
250
300
70 75 80 85 90 95 00 05 10 15
Figure 2. House prices and the interest-to-income ratio of householdsIndex 2005 = 100 and percentage of disposable income
Tenant-owned apartmentsHouses
(a) House prices (b) Households’ interest-to-income ratio
0
2
4
6
8
10
12
70 75 80 85 90 95 00 05 10 15
Note. House prices refer to the house price index (private houses). Interest expense in the interest-to-income ratios is calculated after tax.Sources: Statistics Sweden, Valueguard and the Riksbank
Therearenumerouscross-countryexamplesshowingthatifhouseholddebtincreasesrapidlytogetherwithrisesinhouseprices,vulnerabilityinthefinancialsystem,i.e.theriskofafinancialcrisis,increases.Inmanyofthecountrieshithardestbytheglobalfinancialcrisisof2008-2009,thecrisiswasprecededbyrapidlyincreasingdebtsandhouseprices.Moreover,thepost-crisisslumpriskstobedeeperandmoreprotractedifdebtshavebeenrisingrapidlypriortothecrisis.2
Themostrecentglobalfinancialcrisishasalsotaughtusthattomonitorindividualfinancialinstitutions,i.e.microprudentialpolicy,isnotenough;thefunctionalcapacityofthefinancialsystemasawhole,i.e.itsexposuretosystemicrisk,mustalsobetakenintoaccount.Thisso-calledmacroprudentialpolicyisarelativelynewpolicyareathatcameintofocusafterthefinancialcrisistoaddresssystemicriskconcerns.Thepurposeofmacroprudentialpolicyisbothtomaintaintheresilienceofthefinancialsystem,andtocounteractrisks.
TheEuropeanSystemicRiskBoard,ESRB,issuesrecommendationsregardingmacroprudentialtargetsandspecificmeasuresinSwedenandotherEUcountries.AccordingtoESRB,anintermediatetargetofmacroprudentialpolicyistoavoidexcessivelyrapidcreditexpansionandhighloan-to-incomeratios.3Inline
2 SeeEmanuelssonetal.(2015)foranoverviewofempiricalstudiesoffinancialriskslinkedtohouseholdindebtedness.3 Theothertargetspertaintodifferencesinmaturitiesbetweenliabilitiesandassets,concentratedexposuresandmisguidedincentivesforlargeinstitutionslinkedtotheimplicitorexplicitpublicguarantees.
60 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
withthis,theRiksbankhaslongbeenemphasisingthatrisinghouseholddebtisproblematic,andthattheassociatedrisksneedtobeaddressed.Potentialmeasuresmentionedinthepolicydebatetodealwiththeserisksincludetighteningtheloan-to-valuecap,introducingaloan-to-incomecap,introducinganamortisationrequirement,ortighteningthemortgageinterestdeduction,i.e.reducingthetaxreliefonmortgageinterest.4
Thefirstthreemeasuresareexamplesofmacroprudentialtoolsthatmightbeexpectedtodampenhouseholddemandforcredit.Theloan-to-valuecaplimitsthesizeofthemortgagethatmaybegrantedtoahouseholdinrelationtohowmuchthehomeisworth.Itthereforehasadirectimpactonhowmuchdebtahouseholdmaytake.Aloan-to-incomecapalsolimitsthesizeoftheloan,butinrelationtotheborrower’sincomeratherthanthevalueofthehome.Theamortisationrequirementpotentiallyrestrictsthepossibilitiesofhouseholdstotakeoutloans.Inparticular,itposesalimitationforhouseholdsthatmeetthedownpaymentrequirements,butwhichmighthavedifficultyinsavingfurtheramountsinthefirstcoupleofyearsofthedurationoftheloan.
Thetaxreliefonmortgageinterestisafiscalpolicymeasure.Indebtedhouseholdscancurrentlydeduct30percentoftheirinterestexpensefromtheirtotaltax,uptoSEK100,000.Thismortgagesubsidyincentiviseshouseholdstotakeoutlargermortgagesthantheywouldhavedonewithoutthetaxrelief.Ifthetaxreliefonmortgageinterestwerereduced,itisthereforeprobablethathouseholds’demandformortgageswoulddecrease,anditwouldthusbeaneffectivemeasureforcurbingthebuild-upofdebt.
2SummaryofourfindingsInthisarticlewestudyhowdifferenthouseholdsareaffectedandwhatthemacroeconomicconsequenceswouldbefromtighteningtheloan-to-valuecap,theloan-to-incomecapandtheamortisationrequirements,andreducingthetaxreliefonmortgageinterest.Inourpolicyexperiments,weuseamacroeconomicmodelfittedtoSwedishdata.
Wefindthatallthemacroprudentialmeasuresconsideredinourstudycurtailindebtednessintheeconomyovertime;thisinturnbringsaboutareductioninboththeinterestexpenseofborrowersandtheinterestincomeoflenders.Thesemeasuresleadtoaredistributionofresourcesfromdebt-freetoindebtedhouseholds,i.e.fromlenderstoborrowers.Lenderswillreducetheirconsumptionofgoods,housingservicesandleisure,sincetheirinterestincomewillbelower.Ontheotherhand,borrowerscanincreaseconsumptionofgoodsandleisure.
4 Seee.g.SverigesRiksbank(2014,2015and2016).
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 61
Whetherornottheyalsoincreasetheirconsumptionofhousingservicesdependsonwhichmeasureisundertaken.Atighterloan-to-valuecapwillforceborrowerstocutbackontheirconsumptionofhousingservices.While,iftheamortisationrequirementandloan-to-incomecaparetightened,borrowerswillalsoconsumemorehousingservices.
Areductioninthetaxreliefonmortgageinterestleadsinthelong-runtoaredistributionoftheconsumptionofhousingservicesfromborrowerstolenders.However,iftheGovernmentusesthereleasedbudgetaryresourcestocompensateborrowers,theycanincreasetheirconsumptionofgoodsandleisure.
Thetighteningofvariousmacroprudentialmeasuresandthereductionofthetaxreliefonmortgageinterestcanalsohaveshort-termeffectsondemandandinflation.Thecentralbankmayneedtoconsiderthiswhensettingthepolicyrate.Howmonetarypolicyreactstolowerhouseholddebtdependsbothonthemeasureunderlyingthereduction,andonresidentialmobility.Ifmobilityinthehousingmarketisrelativelylow,householdswillcutbackonnon-durablesconsumption,therebydampeningdemandandhenceinflation.Insuchascenario,monetarypolicymayneedtobemoreexpansionary.If,however,themobilityinthehousingmarketisrelativelyhigh,monetarypolicymayinsteadneedtobetighter.
Wealsostudyhowthelevelofhouseholddebtaffectstheimpactofmonetarypolicyoninflationandhouseholdconsumption.Ahighdegreeofindebtednessmakesborrowersmoresensitivetochangesininterestrates.Thehighertheindebtedness,thegreatertheeffectof,forinstance,aratehikeonborrowers’interestexpenseanddisposableincome,whichinturnreducestheirscopeforconsumption.Thismeansthattheeffectondemand,andhenceoninflation,fromanincreasetothepolicyratewouldbehighertodaythanpreviously.Specifically,aratehikeatthecurrentlevelofindebtednessreducesborrowers’consumptionmuchmorethanitwouldhavedonehadindebtednessbeenatthesamelevelaswhentheinflationtargetwasintroducedinthemid-1990s.Theeffectoninflationfromthisratehikeisthusalsogreatertodaythaninthemid-1990s.
Thepaperisstructuredasfollows.Inthenextsection,wepresentthemacro-economicmodelusedinourpolicyexperiments,andinthesectionafterthatwedescribehowthemodelhasbeenfittedtoSwedishdata.Followingthatisasectionshowingthelong-termmacroeconomiceffectsofthealternativemeasures.Wethendiscusstheshort-termeffectsofdeleveragingoninflationandthepolicyrate,andhowthepotencyofmonetarypolicyisaffectedbythelevelofhouseholddebt.Finally,wecompareourfindingswithotherstudiesonhouseholddebt.
62 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
3TheeconomicenvironmentThissectionpresentsthemacroeconomicmodelweuseinourpolicyexperiments.5Figure3illustratesthestructureofthemodel.Theeconomyispopulatedbyhouseholds,corporations,banks,acentralbankandafiscalauthority.Forthesakeofsimplicity,wedisregardforeigntradeandcross-borderfinancialflowsintheanalysis.
Thehousingmarketisanimportantpartofthemodelsincethemeasureswestudyhaveadirecteffectonthehouseholds'demandforhousing.Asimplifyingassumptionisthathousingisinfixedsupply.Householdsbuyandsellhousingservicestoamarketdeterminedprice.6Furthermore,itisassumedthatthereareonlytwocategoriesofhouseholds–“lenders”and“borrowers”,whodifferinthewaytheydiscountfutureconsumption.Lenderssaveinbankaccountsandfinancetheirhousinginvestmentswithoutamortgage.Ontheotherhand,borrowersfinanceprimarilytheirhousinginvestmentswithbankloans.Inadditiontothesetwocategoriesofhouseholds,therearealso“entrepreneurs”whotakeoutbankloanstofinancetheirinvestments.
Bankshavetwosourcesoffunding,equityandhouseholds’depositandcanuseadepositfacilityatthecentralbank.Theinterbankrateisthusaffectedbythepolicyratesetbythecentralbank.Itisassumedthatdifferentbankscanofferslightlydifferentiatedfinancialproduct,i.e.theyoperateunder“monopolisticcompetition”.Thedegreeoftheirmonopolypowerdeterminesovertimethespreadsbetweenthecentralbank’spolicyrateandthebanks’ratesforhouseholdsandcorporations.Intheshort-term,thesespreadswillalsobeaffectedbythefactthatbanksaresubjecttoadjustmentcostswhenchangingtheinterestrates.Householdpreferencesaredescribedbyautilityfunction,whichdependsontheirconsumptionofgoods,housingservicesandleisure.Bothcategoriesofhouseholdsareforward-looking,buttheyattachdifferentweightstotheirexpectedfutureutility.Thelendersareassumedtobemore“patient”andcaremoreaboutfutureconsumptionthanborrowersdo;thisleadsthelenderstosavemore.
5 ThemodelisbasedonGeralietal.(2010),whichprovidesamoredetaileddescription.However,wehavemodifiedtheirmodelonsomekeypoints,inordertostudytheeffectsofamortisationrequirements,aloan-to-incomecapandreducedtaxreliefonmortgageinterest.SeeFinocchiaroetal.(2016).6 Therearethereforenorentedhomesinthismodel.“Goods”referstoanaggregateofnon-durableconsumergoodsandservicesotherthanhousingservices.
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 63
Housingmarket
Corporate sector
Publicsector
House-holds Banking sector
Borrowers Loans
Deposits
Dividends
Repos/certificates
Loans
Houses
Lenders
Central bank
Taxes/transfersCo
nsum
er g
oods
Labo
ur
Figure 3. An overview of the model structure
Source: Own illustration
Thelenderschoosehowmuchtheywillwork,spendonconsumergoodsandhousinginvestments,andhowmuchtheywilldepositinthebanktomaximiseutility.Theirdisposableincomeconsistsofwageincome,interestontheirsavingsandtransfersortaxesfromthepublicsector,aswellasdividendsfromcompanies.Borrowersaresubjecttosimilarbudgetrestrictionsbuttheycanalsofinancetheirhousingwithmortgages.Inouranalysis,weareparticularlyinterestedinwhichtermsandconditions,besidesthelendingratelevel,applyforhouseholdstobegrantedtheseloans.
Householdscanincreaseorreducetheirhousingholdingsbybuyingorsellingonthehousingmarket.ThehousesdepreciateovertimeatarateofδH so the household’shousingholdingatacertainpointintimet, Ht ,isdeterminedbyundepreciatedhousing(1−δH) Ht−1andnewhousinginvestmentsIt ,
(1) Ht = (1−δH) Ht−1+ It.
Notethatwecanbreakdownthehousinginvestmentsofthehouseholdintodepreciationsoftheexistinghomesandchangesinthehousingstock.Inthelong-run,totalhousinginvestmentsconsistofreplacementinvestments,δH H̅ ,whereH̅istheconstantvolumeofhouses.7Thetotalvolumeofhousesisinfixedsupply.Thisimpliesthatif,forexample,theborrowersbuymorehousingservicesthelendersmustconsumeless.
Themodelcanbeusedtostudytheeffectsofchangesinboththeloan-to-valuecapandintheloan-to-incomecap.Theloanrestrictionfortheborrowinghouseholdsdiffersinthesetwocases.Whenborrowingisconstrainedbyaloan-
7 Adashoveravariabledenotesthatitisalong-termvalue.
64 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
to-valuecap,newloansofhouseholds,N,maynotbegreaterthantheshareμofthenominalvalueofhousinginvestments,
(2) Nt≤μQtIt,
whereQisthenominalhouseprice.8Aggregatehouseholddebt,B,isaffectedbyhouseholds’amortisationoftheexistingdebtρBwheretheparameterρ istherateofamortisation.9Hence,indebtednessprogressesaccordingtothefollowingrelationshipovertime,
(3) Bt =(1−ρ)Bt−1+ Nt =(1−ρ)Bt−1+ μQtIt.
Inthesecondcase,borrowingislimitedbyaloan-to-incomecap.Inthatcase,newloansmayatmostbeashareσofwageincome,
(4) Nt≤σWt Lt,
whereWisthenominalhourlywagerateandLarethehoursworkedbytheborrowinghousehold.10Inthiscasetoo,thehouseholdsareassumedtoamortisetheloansattherateofρ.Householddebtthusprogressesaccordingto
(5) Bt =(1−ρ)Bt−1+ Nt =(1−ρ)Bt−1+ σWt Lt.
Thebudgetconstraintoftheborrowersisasfollows:
(6) Pt Ct + Qt It+(ρ + RHt−1)Bt−1 + Tt + At = Wt Lt + Nt + ωRH RH
t−1Bt−1.
Thelefthandsideshowsexpenditure,whichbesidespurchasesofconsumergoodsCatthepricePandhousinginvestmentsQIalsoconsistofamortisationρB andinterestpayments,RHBonthemortgage.11Thehouseholdsalsopaylump-sum
8 Theloanrestrictioninequation(2)thuslimitsthehousehold’snewloansinrelationtothevalueofhousinginvestments,whichisanapproachproposedbyKydlandetal.(2016).Analternativeformulationoftheloanrestrictionisthattheloanstockofhouseholdsislimitedby(theexpected)valueofthehousingstockasine.g.Iacoviello(2005).Thisentailshoweverthatallloansareshort(oneperiod)andthattheaggregateloanstockofhouseholdsisdirectlyaffectedbythedevelopmentofhouseprices.9 The“impatience”oftheborrowersisconsideredtobesogreatthattheyalwaysborrowuptothelimit,i.e.therestrictioninequation(2)isalwaysbinding.Thesameassumptionismadefortherestrictioninequation(4).10 ThefactthatthenewloansNarelimitedbyincomefromsalaryalsoimpliesthattotalloans,B,arelimited.Inthemodelcalibration,theparameterσissetsothatthetotaldebtsoftheborrowinghouseholdsinrelationtoannualincome,σ / 4ρ,hasareasonablevalue,seemorebelow.11 Allinterestratesinthemodelarefloating(oneperiod),sothehouseholdspaythesameinterestrateirrespectiveofwhentheloanwasgranted.
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 65
taxesTandadjustmentcostsAthatarisewhenhouseholdschangetheirhousinginvestments.12Thesecostsaimatcapturing,inasimplifiedway,theeffectsofvariousfactorsthataffectmobilityonthehousingmarket.Therighthandsideofthebudgetconstraintshowstheresourcesavailabletofinanceexpenditures.BesideslabourincomeWLtheresourcesconsistofnewbankloansN(whicharelimitedbyequation(2)orequation(4))andofgovernmentsubsidiesofthehouseholds’mortgageexpensesωRH RH B.ThedegreeofthesubsidyisgivenbyωRH.Anotherwayofreducinghouseholddebt–besideslimitingtheloan-to-valueratioμand/ortheloan-to-incomeratioσorimposingdemandsforahigheramortisationrateρ–isthustoreducethissubsidy.
Inordertofinancepublicconsumptionandmortgageinterestsubsidies,thepublicsectorbalancesitsbudgetbylevyinglump-sumtaxesfromthehouseholds,andataxonthedividendsthatthelendersreceivefromthecorporatesector.GovernmentconsumptionisassumedtomakeupaconstantshareofGDP,whilethelump-sumtaxisdistributedbetweenthelendersandtheborrowersinrelationtotheirsharesofthetotalwagesum.
TheobjectiveofmonetarypolicyinSwedenistomaintainpricestability,whileatthesametimemonetarypolicyshallcontributetosustainablegrowthandahighlevelofemployment.Inthemodel,itisassumedthatthecentralbankcanadjustitspolicyrateRtostabilisebothinflationπandGDPgrowthΔYintheeconomy. InflationisdefinedastheannualpercentagechangeinthepriceoftheconsumergoodsP.Morespecifically,thepolicyrateevolvesaccordingto,
(7) Rt=φRRt−1+(1−φR)[R̅+φπ(πt−π̅ )+φY ∆Yt ] + εt,
whereR̅isthelong-termlevelofthepolicyrate,π̅isthelong-termleveloftheinflationrate(whichcoincideswiththecentralbank’sinflationtarget)andε adisturbancetermthatweusetocapturethenon-systematic(unexpected)componentofmonetarypolicy.TheparameterφRisameasureoftheinertiainsettingtheinterestrate,whileφπandφYmeasurehowmuchthecentralbank’schoiceofpolicyratelevelisaffectedbyinflationandgrowthfluctuations,respectively.Inthelong-run,monetarypolicydoesnotaffecttherestoftheeconomy.Intheshortterm,however,themacroprudentialandfiscalmeasuresinourstudymayhaveanimpactondemandandinflation,towhichmonetarypolicymayneedtoreact.
Lowerloan-to-valueorloan-to-incomecaps,stricteramortisationrequirementsandreducedtaxreliefonmortgageinterestareallexamplesof
12 Lump-sumtaxesmeansthatthehouseholdspayafixedamountintax.Thehouseholds’behaviourthereforedoesnotaffectthesizeofthetaxpayment.
66 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
alternativemeasuresthathaveemergedintheSwedishdebateaspotentialsolutionsforcounteractingtherisksassociatedtohighindebtedness.Inthemodel,changestotheparametersμ(loan-to-valuecap),σ(loan-to-incomecap),ρ(amortisationrequirement)andωRH(taxreliefonmortgageinterest)correspondtothesemeasures.
4FittingthemodeltoSwedishdataInordertostudythemacroeconomiceffectsofreducingSwedishhouseholddebt,themodelmustbefittedtoSwedishdata.Inotherwords,themodelparametersmuchbeassignedvaluessothatthecharacteristicsofthemodelreflecttheSwedisheconomy.Theparameterscanbedividedintotwocategories.
Thefirstcategoryconsistsofparametersthatprimarilyaffectthemodel’slong-termcharacteristics.Forthoseparameters,thereisoftenanobservableequivalentindata.Anexampleistheparameterthatdeterminesthedegreeofcompetitiononthemortgagemarket,whichinturndeterminesthebanks’averagemarginsonmortgages.Thisparametercanbedeterminedbasedontheaveragedifferencebetweenmortgageratesandthereporate.InTable1wedescribelong-termlevelsforcertainvariablesinthemodelthatareofparticularimportance.Theaimisnottoexactlymatchthelong-termvalueswithdata,butrathertoverifythattheconsistencyissufficientlygoodforthemodeltobeusedtostudytheeffectsofreducinghouseholddebtinSweden.
Theothercategoryconsistsofparametersthatprimarilyaffecttheshort-termdynamics.Anexampleofonesuchparameteristhatwhichdetermineswhatimpactachangetothereporatewillhaveonthemortgagerate.Fortheseparameters,thereistypicallynoobservableequivalentindata,soothermethodsareneededtoestimatethem.WehaveusedBayesianeconometricstoestimatethem,entailingthatwehavecombinedinformationindatawithjudgementsbasedoneconomictheoryandempirics.
4.1Householddebt,loan-to-valueratioandamortisationAswenotedinthebeginningofthisarticle,thedebtratiooftheSwedishhouseholdshasincreasedfromaround90percentin1995tocloseto180percentin2015.Around75percentofthosedebtsconsistofloanscollateralisedbyhomes.Mortgagesthusmadeuparound130percentofhouseholds’disposableincomein2015.Thefactthatthisratiohasincreasedoverthepast20yearsmakesitdifficulttocalibratethedebtratio,becauseinthemodelitisconstantinthelongterm.Wehavesetthelong-termratioofmortgagestohouseholds’disposableincometo124percent,i.e.somewhatlowerthanthecurrentlevel.
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 67
Thisratioismainlydeterminedbyhowthehouseholdsfinancepurchasingtheirhomes,andhowfasttheypaydowntheloans.Whenthesizeoftheloanislimitedbyhowmuchthehomeisworth,thefollowingrelationshipgivestheloan-to-valueratioforthemortgagestockinthelongterm:
(8) μB = μδHρ ,
where μB is the loan-to-value ratio in the mortgage stock, μ is the loan-to-valueratiofornewloans,δHistherateofdepreciationofthehousingstock,andρistheamortisationrate.Inthemodel,theloan-to-valueratiofornewloanscoincideswiththeloan-to-valuecap,becausetheloanrestrictionisassumedtobebindingandtheborrowinghouseholdsareidentical.Inreality,theloan-to-valueratiovariesbetweenhouseholds, andmost households borrow below the loan-to-value cap.With thepresentleveloftheloan-to-valuecap,thesizeofaloancollateralisedbythehomemayamounttonomorethan85percentofthevalueofthehome.Theaverageloan-to-valueratiofornewloanswas72percent in2014,whilethe loan-to-valueratiofortheentiremortgagestock inthesameyearwas63percent.13Usingmicrodataregardinghouseholddebtsfor2013-2014,wecancalculatetheannualamortisationrateataround2percentannually,moreorlessequallinganamortisationperiodof50years.Basedthereon,wesetμB =65percent,μ=75percent,andρ=0.7percent(quarterly).14TherelationshipabovethengivesarateofdepreciationofδH=0.6percent(quarterly).
Inorderfortheloan-to-incomeratiointhemodeltobeconsistentwithdata,theborrowinghouseholds’shareofthetotalwagesumissetto40percent.Borrowers’mortgagesinrelationtodisposableincomewillthenbe456percent,whichisinlinewithdata.Theaverageloan-to-incomeratiofornewloanswas387percentin2014and406percentin2015.15Thevolume-weightedaverageloan-to-incomeratiowascloseto500percentin2014.
Theloan-to-incomecaplimitshowmuchhouseholdsmayborrowinrelationtotheirwageincome(seeequation(4)).16Inthemodel,σ / 4 ρcorrespondstotheratioofmortgagedebttotheborrowinghouseholds’wageincome,andissetat251percent.Whichevermeasureofincomeweuse,atighteningoftheloan-to-incomecapwillaffectdebtsasashareofboththewageincomeandthedisposableincome.Inourcalculationsbelow,wewilldescribehowtheyareaffectedinrelationtowageincome.
13 Fornewloans,thevolume-weightedaverageinFinansinspektionen’ssampleisused.SeeFinansinspektionen(2015).14 Withthisamortisationrate,aroundhalfthevalueofahomewillbeamortisedafter25years.15 SeeFinansinspektionen(2016).16 Theborrowinghouseholds’disposableincomeconsistsofwageincome,lessinterestexpensesandtaxes.
68 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
4.2InterestratesThebanksoperate,toacertaindegree,asmonopolists.Thiscreatesapositivespreadbetweentheirlendingratesandtherisk-freerate,whiledepositratesarelower.Thevariousspreadsinthemodelarecalibratedbasedonaveragedeviationsfromthereporatefortheperiod1995-2015.Theaveragereporateforthatperiodwas2.7percent.Theshortmortgageratewas4.0percentonaverage,theshortcorporatelendingratewas4.7percent,andthedepositrate1.4percent.
4.3ThebankingsectorThemodeldescribesaclosedeconomy.Thebanks’assetsandliabilitiesthusrelatetodomesticmortgagesandcorporatelendinginthebanks’balancesheets.Swedishmortgagesmakeuparound15percentofthebanks’assets,andlendingtonon-financialcorporationsmakesuparound10percentoftheassets.Hence,aroundonequarteroftheSwedishbanks’assetsarecoveredbythemodel.TheassetsofthefourlargeSwedishbanksasashareofGDParearound400percent,andtheyaccountforaroundthreequartersofalldepositsandlendinginSweden.Wehavethereforecalibratedthebanks’assetsasashareofGDPtoavaluesomewhatabove100percent.Inthemodel,thebanks’lendingisonlyfundedbyhouseholddepositsandequity.Inthedata,depositsandborrowingsfromthegeneralpublicmakeuparound35percentofthefourlargebanks’funding.
Thefourlargebanks'equityinrelationtotheirassets(leverage)variesintherangeof4to6percent.Inthemodel,theleverageratioofthebankingsectoriscalibratedat4.6percent.Wealsoassumethattheresourcesneededtorunthebanks,suchassalariesforemployees,dependonthesizeofthebanks’capital.Thebankingsector’sshareofGDPissetat3.6percent.
4.4TaxreliefonmortgageinterestThe30percenttaxreliefonmortgageinterestequals2.5percentofpublicexpenditureinthemodel.Inthemodel,theindebtedhouseholdsspendaround7.4percentoftheirgrossincomeoninterestpayments.Thisisarelativelyhighpercentagecomparedwithrecentdataonthehosueholds'interest-to-incomeratioandthedifferencemay,atleastpartially,beexplainedbythefactthattheinterestrateinrecentyearshasbeenmuchlowerthanitsnormallevel.
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 69
Table 1. Calibration of long-term values in the model Percent
GDP components and tax relief on mortgage interest Value
Consumption,percentageofGDP 49
Investments,percentageofGDP 22
Publicexpenditure,percentageofGDP 29
Taxreliefonmortgageinterest,percentageofpublicexpenditure 2.5
Interest rates
Reporate 2.7
Mortgage rate, short 4.0
Corporatelendingrate,short 4.7
Depositrate,short 1.4
Debts, loan-to-value ratios and amortisation
Mortgagedebt,percentageofGDP 55
Mortgagedebt,percentageofdisposableincome,allhouseholds 124
Mortgagedebt,percentageofdisposableincome,indebtedhouseholds 456
Mortgagedebt,percentageofwageincome,indebtedhouseholds 251
Loan-to-valueratio,newloans 75
Loan-to-valueratio,mortgagestock 65
Amortisation,percentageofloanstock,byquarter 0.7
5Long-termeffectsofreducinghouseholddebtAsdiscussedabove,thereareseveraldifferentmeasureswhichSwedishauthoritiesmayundertaketodampentheincreasinghouseholddebt.Inthissectionweexamine,fromamacroeconomicperspective,whatthelong-termeffectswouldbeoffourdifferentmeasures:
• Tighteningtheloan-to-valuecap
• Tighteningtheamortisationrequirement
• Introducingaloan-to-incomecap
• Reducingthetaxreliefonmortgageinterest
70 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
Wecalculatehowthedifferentmeasureswouldaffecthouseholdsbyapplyingeachmeasuresuchthattheborrowinghouseholds’loan-to-incomeratio(thepercentageofmortgagedebtofincomefromsalary)declinesby10percent.Thisimpliesthatmortgagedebtasashareofdisposableincomedeclinesbyaround12percentagepoints.Suchastandardisationwillfacilitatethecomparisonoftheeffectsofthedifferentmeasures.17Wethencalculatewhatthemacroeconomiceffectswouldbeforthehouseholds.Theleveloftheloan-to-incomeratiocanbeseenasameasureofpotentialrisksinhouseholddebt,becausemanyhouseholdsusetheircurrentincometopaytheirloanexpenses.
5.1Effectsoftighteningtheloan-to-valuecapAloan-to-valuecapmeansthatthesizeofamortgageislimitedwithrespecttohowmuchthehomeisworth.Ifthecapistightened,householdscanthusfinanceasmallerproportionoftheirhomepurchaseswithloans.InSwedishdataregardingloan-to-valueratiosfornewloans,thereisawidespreadacrosshouseholds.Between20and30percentofhouseholdsarelimitedbytheloan-to-valuecap.Inthemodelweassumethatallborrowershavethesameloan-to-valueratiofornewloans μ.Wealsoassumethattheloan-to-valueratioofthehouseholdsdecreasesiftheloan-to-valuecapistightened.Theloanrestrictionfortheborrowinghouse-holds(equation2)can,inlong-termequilibrium,bewrittenas,
(9) N̅ Q̅ I ̅
= μ.
Similarly,thedebtequation(equation3)canbewrittenas
(10) N̅ = ρ B̅,
wheretheparameterρdenotestheborrowers’amortisationrateandBtheirindebtedness.Hence,thenewborrowingofhouseholdsNequalstheirloanamortisationinthelongrun.
Inthemodel,themortgagerateovertimeisdeterminedbytwofactors:thestateofcompetitiononthemortgagemarket,andhouseholds’preferencesforconsumptiontodayinrelationtoconsumptioninthefuture.Changestotheloan-to-valuecapthereforedonotaffectthemortgagerate,andneitherdoanyoftheothermeasures.Thatmeans,forexample,thatiftheloan-to-valuecapistightened,theinterestexpenseofborrowerswilldecrease–indebtednesswilldeclinewhiletheinterestrateremainsthesame.
17 Themeasurethusleadstotheborrowinghouseholds’mortgagedebtsasapercentageofwageincomedecreasingfrom251to226percent.
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 71
Figure4showshownewborrowing,amortisation(ρB)andindebtednesschangeovertimeiftheloan-to-valuecapistightened.Theloan-to-valuecapisreducedatthepointintimet.Wecanseethatnewborrowingsdropimmediately,whileamortisationandindebtednessslowlyadapttonew,lowerlevels.Intime,indebtednessdecreasesinthesameproportionasnewborrowing,becauseatightenedloan-to-valuecapdoesnotchangetheamortisationrate.
New borrowings Amortisation Indebtedness
t t
Figure 4. Adjustments of new borrowings, amortisation and indebtedness following a tightened loan-to-value capSEK
Source: Own illustration
InTable2thelong-termeffectsforhouseholdsfromtighteningtheloan-to-valuecapsuchthattheloan-to-incomeratiodecreasesby10percentareshown.Theaverageloan-to-valueratiofornewloansthendeclinesfrom75to70percent.Theborrowersusethefundsreleasedbythelowerinterestexpensetoincreaseconsumptionofgoods(around2percent)andleisure(around1percent).Butthetighterloan-to-valuecapalsobringsaboutareductionintheconsumptionofhousingservicesofaround1.5percent.
Inthemodel,thesupplyofhomesisassumedtobeconstant,soonlyaredistributionofthehousingstockbetweenborrowersandlendersoccurs.Becausetheborrowerscutdownontheirconsumptionofhousingservices,thelenderswillincreasetheirconsumptiontothesameextent.Inthiscase,thelenders’consumptionofhousingservicesincreasesbyaround1percent.
Forthelendersthelowerloan-to-valuecapleadstoanincreaseintheconsumptionofhousingservices,whileatthesametimetheirsavingsdecline
72 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
andhencetheirinterestincometoo.Consequently,theyincreaselaboursupplysomewhatandreducetheirconsumptionofgoods.Thelowerindebtednessintheeconomythusbringsaboutashiftinresourcesfromlenderstoborrowers.
5.2EffectsofincreasingtheamortisationrateThepurposeofanamortisationrequirementistomakehouseholdsamortisetheirdebtsfaster.Anincreasedamortisationratewill,likealowerloan-to-valuecap,leadtodebtsdecreasingandbeinglowerinthenewlong-termequilibrium.However,themeasuresaffectnewborrowingsindifferentways,sotheeffectsonhouseholds’consumptionofgoodsandhousingserviceswillalsobedifferentwithanincreasedamortisationrateρcomparedtowithalowerloan-to-valuecap.Assumethattheamortisationrequirementistightened,leadingtoanincreaseintheamortisationrate.Inthatcase,newborrowingsandindebtednesswillnotchangeproportionallytoeachother,seeequation(10).Inthemodel,afasteramortisationrateleadstoadeclineinindebtedness,whilenewborrowingsarelargelyunchangedbothintheshortandlongterm.Theeffectthusdiffersfromalowerloan-to-valuecap,whichleadstonewborrowingsdecliningrapidlyandbeinglowerovertime,becausetheloan-to-valuecapdirectlylimitsnewborrowings.Anamortisationrequirementdoesnothavethesamedirecteffectonnewborrowings,andwillthereforehavelessofanimpactonthehousingmarket.Figure5illustratestheeffectsafteratighteningoftheamortisationrequirementatthepointintimet.Theincreasedamortisationrateleadstoamortisationsrisinginitiallyandbeinggreaterthannewborrowings,tothenfallbackasindebtednessdeclinestoalowerlong-termlevel.Inthelongrun,thesizeofamortisationsρBismoreorlessunchanged.TheshareofdebtthatisamortisedeachyearρhasincreasedbutatthesametimethedebtBhasdecreased.Becausenewborrowingsintimeareaffectedinthesamewayasamortisations,thisalsomeansthatnewborrowingsarelargelyunaffectedovertime.
Thisillustratesanimportantdifferencebetweentighteningtheamortisationrequirement,andtighteningtheloan-to-valuecap.Bothmeasuresleadtolowerindebtednessovertime,butahigheramortisationratehaslittleimpactonnewborrowings,whilealowerloan-to-valuecapleadstoadropinnewborrowingsinproportiontoindebtedness.Thisisimportantbecauseitisnewborrowings,andnotindebtedness,thataffectthepossibilitiesofborrowerstoconsumehousingservices.
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 73
t t
Figure 5. Adjustments of new borrowings, amortisation and indebtedness following a tightened amortisation requirement SEK
New borrowings Amortisation Indebtedness
Source: Own illustration
Table2showsthequantitativeeffectsofatighteningoftheamortisationrequirement.Inorderfortheloan-to-incomeratiotofall10percent,theamortisationperiodmustshortenfrom50to45years.Newborrowingswillbeslightlylowerinthelong-termwiththismeasure.Theborrowinghouseholdsusethemoneytheyhaveleftoverfromreducedinterestexpensetoconsumegoodsandleisure,besidesconsumingslightlymorehousingservices.Theinterestincomeofthelendersdeclines,becausesavingintheeconomyfalls.Therefore,theycutdownontheconsumptionofgoods,housingservicesandleisure.
5.3Effectsofintroducingaloan-to-incomecapTheloan-to-incomecapisamacroprudentialmeasurethatresemblestheloan-to-valuecap.Thedifferenceisthathouseholds’loansarenotlimitedbythevalueofthehome,butratherbythehouseholds’incomefromsalary.Hence,theintroductionofaloan-to-incomecapleadstoachangeinhouseholds’newborrowingsinrelationtotheirindebtedness,justlikeafteratighteningoftheloan-to-valuecap.However,unlikeatighteningoftheloan-to-valuecap,areductioninnewborrowingsduetoaloan-to-incomecapdoesnotnecessarilyreduceborrowers’possibilitiestoconsumemorehousingservices,becausenewborrowingsinthiscasearenotlinkedtothevalueofthehome.18
18 Formally,thismeansthattherestrictioninequation(2)isnotbinding.Evenifnewborrowingsdecline,theborrowinghouseholds’consumptionofhousingservicescanbeunchangedorevenincreasebecausetheirinterestexpensedecreases,aslongastheymeetthebudgetrestriction(6).
74 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
Inordertoreducetheloan-to-incomeratioby10percent,aloan-to-incomecapmustbeintroducedthatreducesborrowers’loan-to-incomeratio,i.e.theirdebtsinrelationtotheirwageincome,by25percentagepointsfrom251to226percent.ThecolumnatthefarrightinTable2showsthelong-termeffectsthiswouldhave.Wheretheborrowersareconcerned,thismeansthatconsumptionofbothgoodsandleisureincreases.Thefactthatconsumptionofhousingservicesincreasesisalsonoteworthy.Thisshowsthatatighteningoftheloan-to-incomecapneednotinitselfdampenborrowers’housingconsumption.Thelenders,fortheirpart,cutdownontheirconsumptionofgoods,housingservicesandleisureduetolowerinterestincome.Inpurelyqualitativeterms,atighteningoftheloan-to-incomecap,oratighteningoftheamortisationrequirement,affectshouseholdsinasimilarway,althoughthequantitativeeffectsdiffer.
Table 2. Long-term effects of tightened macroprudential measuresPercent,excepttheamortisationrequirement,whichisinyears
Loan-to-value cap
Amortisation requirement
Loan-to-income cap
Initialvalue 75.0 50.0years 251.2
Endvalue 69.5 44.9years 226.1
Loan-to-incomeratio,change −10.0 −10.0 −10.0
Borrowers
Consumptiongoods,change 2.0 1.5 1.2
Consumptionhousingservices,change −1.7 0.2 1.2
Leisure,change 1.2 0.9 1.0
Lenders
Consumptiongoods,change −0.6 −0.5 −0.5
Consumptionhousingservices,change 1.0 −0.1 −0.5
Leisure,change −0.1 −0.1 −0.1
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 75
5.4EffectsofreducingthetaxreliefonmortgageinterestPeoplewhoborrowmoneytopurchaseahomecan,intheirtaxreturns,deductfromtheirtaxedamount30percentofpaidinterestexpense.Thiscanbeviewedasagovernmentsubsidyofloanfinancingtopurchaseahome.19Ifthetaxreliefonmortgageinterestdecreases,thesubsidywillbeless,andhencesotootheGovernment’sexpenditureforit.TheeffectofreducingtaxreliefonmortgageinterestthereforedependsonwhattheGovernmentusesthemoneyforinstead.Inpractice,itcouldbeusedtoincreasetransfers,reducetaxes,increasepublicexpenditureortopaybackthesovereigndebt.Inourmodel,whichhasarelativelysimplisticdescriptionofthepublicsector,thefundsreleasedcanonlybeusedtoincreasetransferstolendersandborrowers,ortoincreasepublicconsumption.20,21
5.4.1 Decreasing the tax relief on mortgage interest lowers the borrower's consumption of housing services Wewillstudythreedifferentscenariostoillustratewhattheeffectswillbefromreducingthetaxreliefonmortgageinterest,andhowthoseeffectsdependonwhattheGovernmentusesthereleasedbudgetfundsfor.Inthefirstscenario,theGovernmenttransfersthemoneytoborrowersandlendersinproportiontothewagesumforeachcategoryofhousehold.Inthesecondscenario,theGovernmenttransfersthemoneyentirelybacktotheborrowers.Thismeansthatthedistortingeffectsofthetaxreliefonmortgageinterestareweededout.Thisscenariothusgivesanindicationoftheextenttowhichtheloansubsidyaddstoexcessiveindebtednessandhenceloanfinancingofhomepurchasing.Inthethirdscenario,theGovernmentinsteadusesthemoneyentirelytoboostpublicconsumption.
Inallthreescenarios,thetaxreliefonmortgageinterestdeclinestothelevelrequiredtobringtheloan-to-incomeratiodownby10percent.Inthefirstcase,whenthereleasedbudgetfundsaretransferredtoallhouseholds,thetaxreliefonmortgageinterestmustbereducedfromthecurrentlevelof30percenttoaround3percent,asshowninTable3.Theborrowersareaffectedmainlythroughtwochannels.First,thedistortingincentivestotakeondebtwillbelower.Second,theyaresubjecttoanegativeeffectonincomebecauselessthanhalfofthereleasedbudgetfundsfromthetaxreliefonmortgageinterestaretransferredbacktothem.Onthewhole,thisleadstoborrowers’consumptionofhousing
19 ReferstointerestexpensesuptoSEK100,000annually.20 Inthemodel,therearenodistortingeffectsfromtaxes,withtheexceptionofthetaxreliefonmortgageinterest,sothereisnodifferencebetweenincreasingtransfersandreducingtaxes.21 SeealsoEnglund(2016)forastudyoftheeffectsofreducingtaxreliefonmortgageinterest.
76 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
servicesdecliningbyaround7percent,whileatthesametimetheirconsumptionofgoodsandleisureismoreorlessunchanged.Wherethelendersareconcerned,thereductioninthetaxreliefonmortgageinterestgivesapositiveeffectonincome,becausemorethanhalfofthemoneyistransferredtothem.Therefore,consumptionofbothgoodsandhousingservicesincreases.
Inthesecondscenario,themoneyfromthetaxreliefonmortgageinterestistransferredentirelybacktotheborrowers.Inorderfortheloan-to-incomeratiotodecline10percent,thetaxreliefonmortgageinterestmustbereducedtoaround−6percentinthiscase.Thetaxreliefonmortgageinterestbeingnegativemeansthatitisataxonloanfinancingforhomes.So,a6percenttaxonloansforhomesisneededfortheloan-to-incomeratiotodecreaseby10percent.Inthatcase,borrowers’consumptionofhousingservicesdeclinesbymorethan6percent,whileconsumptionofgoodsincreasesbyover3percent,andleisurebyaround2percent.Thesecalculationsillustratethedistortingeffectsofthetaxreliefonmortgageinterest,becausetheGovernmentfullytransfersbackthemoneyfromthetaxreliefonmortgageinteresttotheborrowers.Thetaxreliefonmortgageinterestthusleadstotheborrowersconsuming“toomuch”ofhousingservices,butalsotothemconsuming“toolittle”ofothergoods.Thelenders,fortheirpart,cutdownontheirconsumptionofgoodsandleisureduetolowerinterestincome.However,theirconsumptionofhousingservicesincreasesbecausethetotalhousingstockisunchanged.
Inthethirdscenario,theGovernmentusesthemoneyfromthetaxreliefonmortgageinterestentirelytoboostpublicconsumption.Inorderfortheloan-to-incomeratiotodecline10percent,inthiscaseitsufficestoreducethetaxreliefonmortgageinteresttoaround6percent.Itisthereforethemostappropriatealternativeiftheaimofreducingthetaxreliefonmortgageinterestissolelytoreducehouseholddebt.Wheretheborrowersareconcerned,however,itisanexpensivealternative–despiteinterestexpensebeinglower–becausetheymissoutontheentireloansubsidy.Theborrowers’consumptionofhousingservicesdeclinesbyover6percentandtheconsumptionofgoodsandleisurebyjustshyof1percenteach.Theconsequencesforlendersaremorebeneficial.Consumptionofhousingservicesincreasesbyjustshyof4percent,whileatthesametimetheconsumptionofgoodsandleisureismoreorlessunchanged.However,inordertoperformanaccuratemacroeconomicevaluationofthisalternative,theutilitytohouseholdsprovidedbyhigherpublicconsumptionmustalsobeweighedin.
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 77
Table 3. Long-term effects of reducing tax relief on mortgage interest Percent
Transfers to both borrowers and
lenders in proportion to total salary
Transfers only to borrowers
Public consumption
Initialvalueoftaxreliefonmortgageinterest 30.0 30.0 30.0
Endvalueoftaxreliefonmortgageinterest 2.8 −6.5 6.2
Loan-to-incomeratio,change −10.0 −10.0 −10.0
Borrowers
Consumptionofgoods,change −0.1 3.1 −0.8
Consumptionhousingservices,change −7.0 −6.1 −6.4
Leisure,change 0.1 1.8 −0.5
Lenders
Consumptionofgoods,change 0.5 −1.1 0.1
Consumptionhousingservices,change 4.2 3.7 3.8
Leisure,change 0.4 −0.3 −0.1
5.5Whichmeasureisthemosteffectiveinreducinghouseholddebt?Therearedifferentwaysofevaluatingwhichmeasureisthemosteffectiveinreducinghouseholddebtfromamacroeconomicperspective.Inthemodel,householdutilityisanappropriatemeasureforsuchananalysis.Theutilitydependsontheconsumptionofgoods,housingservicesandleisure.Thefactthattheutilityisaffectedbyleisureisanimportantassumptionbecauseitmeansthatanincreaseinconsumptionduetoworkingmoredoesnotnecessarilyincreasehouseholdutility,becauseahigherlaboursupplymeanslessleisuretime.
GDPpercapitaisanothermeasurethatiscommonlyusedtoquantifymacroeconomicbenefitorwelfareinasociety.Itisasimplemeasureandisthereforeoftenusedinpractice.However,adrawbackwithGDPpercapitaisthatwelfaredoesnotonlyconsistofmaterialwellbeing,butalsootherfactorsthataremoredifficulttomeasure,suchasleisureandhealth.
Table4showstheeffectsofreducingtheloan-to-incomeratioby10percentthroughthefourdifferentmeasuresthatweanalysedabove:loan-to-valuecap,
78 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
amortisationrequirement,loan-to-incomecapandtaxreliefonmortgageinterest.Forthetaxreliefonmortgageinterest,weshowthesamethreecasesasbefore.Inthefirsttwo,themoneyreleasedfromthetaxreliefonmortgageinterestisusedfortransferstohouseholds,andonlytoborrowersinthesecond.Inthethirdcase,themoneyisusedforpublicconsumption.Asameasureofhouseholdwelfarefromamacroeconomicperspective,householdconsumptionofgoods,housingservicesandleisureisshownontheonehand,andGDPpercapitaontheother.22
Thehouseholds’aggregateconsumptionofhousingservicesisnotchangedbythevariousmeasures,becausethetotalsupplyisconstant.Iftheborrowersandlendersvaluehousingservicesconsumptioninthesameway,theeffectsonhouseholds’housingconsumptionwouldthushavenobearingonjudgingwhichmeasureismosteffectiveforreducingindebtedness.23Thechangetotheconsumptionofgoodsisalsoclosetozeroforthesemeasures,althoughitincreasesbyaround0.2percentforthetaxreliefoninterestincomewhentheGovernmentusesthemoneyreleasedfortransferstobothcategoriesofhousehold.Thechangetohouseholds’leisureisofgreatersignificance,decliningby0.3percentwhenthegovernmentusesthemoneyreleasedfromreducedtaxreliefonmortgageinterestforpublicconsumption,butincreasesby0.6percentwhenthemoneyisinsteadusedfortransferstoborrowinghouseholdsonly.Thechangeinpublicconsumptionfromthesemeasuresalsoappearstobesignificant,particularlywhentheGovernmentusesthemoneyfromthereducedtaxreliefonmortgageinterestpaymentsforpublicconsumption.Inthatcase,publicconsumptionincreasesby2percent.
Theseresultsshowthat,fromamacroeconomicperspective,itisdifficulttodrawanyclear-cutconclusionsaboutwhichmeasurewouldbemosteffectiveinreducinghouseholddebt.Thethreemacroprudentialmeasureshaveslighteffectsonconsumptionofgoods,whileleisureincreasesbyaround0.3-0.4percent,butpublicconsumptiondropsbyaroundasmuch.Intermsofthetaxreliefonmortgageinterest,theeffectsalsodivergeslightly.Ifthebudgetfundsreleasedfromreducingtaxreliefonmortgageinterestareusedfortransferstoallhouseholds,theimprovementwouldhoweverbeclearbecauseconsumptionofbothgoodsandleisureincreaseswhileatthesametimeconsumptionofhousingservicesandpublicconsumptionareunchanged.24
22 Thepopulationisnormalisedtoone,andhencethereisnopopulationgrowthinthemodel,soGDPandGDPpercapitaarethesame.23 Inthemodel,theborrowersgainslightlymoreutilitythanthelendersfromconsuminghousingservices,sothedifferenceintermsofutilityisnotzero.24 ThefactthathouseholdconsumptionofgoodscanincreaseatthesametimeasGDPpercapitadecreasesisduetoareductioninbothcapitalformationandtheconsumptionofentrepreneurs.
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 79
IfthemacroeconomiceffectsareinsteadevaluatedusingGDPpercapita,areductionintaxreliefonmortgageinterest,withthereductioninGovernmentexpenditurebeingusedforpublicconsumption,ismosteffective.ThatmeasurewouldincreaseGDPbyaround0.3percent.ItisworthnotingthatallothermeasuresleadtoareductioninGDPpercapita,becausethosemeasuresleadtohouseholdsoptingformoreleisuretime,sothenumberofhoursworkedwillbelower.
Table 4. Long-term effects of reducing the loan-to-income ratio by 10 per cent from a macroeconomic perspectivePercent
Mortgage cap
Amortisation requirement
Loan-to-in-come cap
Tax relief on interest
expense transfers to
both bor-rowers and
lenders
Tax relief on interest
expense transfers to
borrowers only
Tax relief on interest expense
public consumption
Initialvalue 75.0 50 251.2 30.0 30.0 30.0
Endvalue 69.5 44.9 226.1 2.8 −6.5 6.2
Loan-to-incomeratio,change −10.0 −10.0 −10.0 −10.0 −10.0 −10.0
Consumption
Goods,change 0.0 0.0 −0.1 0.2 −0.1 −0.1
Housing services,change 0.0 0.0 0.0 0.0 0.0 0.0
Leisure,change 0.4 0.3 0.4 0.3 0.6 −0.3
Publicconsumption,change −0.4 −0.3 −0.4 0.0 0.0 2.0
GDP per capita, change −0.4 −0.3 −0.4 −0.3 −0.6 0.3
80 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
6Whataretheshort-termimplicationsformonetarypolicyofreducinghouseholddebt?Changestothevariousmacroprudentialmeasuresandtothetaxreliefonmortgageinteresthavelong-termmacroeconomiceffects,butcanalsohaveshort-termeffectson,forinstance,demandandinflation.Thetaxreliefonmortgageinterestaffectsdemandintheeconomythroughchangesinpublicsectorincomeandexpenditure,whilevariousmacroprudentialmeasuresaffectdemandthroughhouseholddept.Thecentralbankmayneedtoconsiderthiswhensettingthepolicyrate.
Inordertoevaluatetheimplicationsformonetarypolicyfromreducinghouseholddept,wehavestudiedtheeffectsofintroducingthefourdifferentmeasures(loan-to-valuecap,amortisationrequirement,loan-to-incomecapandtaxreliefonmortgageinterest).Eachmeasureisintroducedatacertainpointintime,andistheninplaceforalongtime,butgraduallyrevertstoitsoriginallevel.Inallcases,thelevelsofthevariousmeasuresaresetsuchthathouseholddeptafter10yearsisaround10percentlowerthanitsinitiallevel.
Weshowthatanimportantfactoraffectingthemeasures’short-termeffectsismobilityonthehousingmarket.Wecomparetwocases.Inonecase,mobilityonthehousingmarketishigh,butlowintheothercase.
Figure6showshowinflationandthepolicyrateareaffectedifhouseholddebtdeclineswhenmobilityishigh.Reducingtheloan-to-valuecaportaxreliefonmortgageinteresthassimilareffectsonthepolicyrateandinflation.Inflationrisesatmostbyjustunder0.2percentandthepolicyrateby0.2percent.Tighteningtheamortisationrequirementhastheoppositeeffectonthepolicyrateandinflation,however,asbothdecreaseswiththismeasure.
Inthecasewhenmobilityinsteadislow,theeffectsonthepolicyrateandinflationaredifferent,asshowninFigure7.Inflationdeclinesirrespectiveofthemeasureandbyuptoalmosthalfapercentagepointwhenamortisationrequirementsaretightened.Thecentralbankthereforecutsthepolicyratetodampenthedropininflation.
Theimpactonmonetarypolicythereforedoesnotonlydependonwhichmeasureisimplemented,butalsotheextentofmobilityonthehousingmarket.Ifmobilityislow,householdsareforcedtocutdownontheirconsumptionofothergoods.Inthatcase,inflationwillbelowerandmonetarypolicymoreexpansionary.However,thisalsoleadstoadropinmortgagerates,whichpartiallycounteractstheeffectsofthemeasuresonindebtedness.Thefindingsthusillustratethatthedifferentmeasureshaveimplicationsformonetarypolicythatmaydiverge,andthatitisimportanttoalsoconsidermobilityonthehousingmarket.
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 81
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
-0.2
-0.2
-0.1
-0.1
0.0
0.1
0.1
0.2
1 4 7 10 13 16 19 22 25 28 31 34 37 40 1 4 7 10 13 16 19 22 25 28 31 34 37 40
Amortisation requirement Mortgage capTax relief on mortgage interest
Figure 6. A tightening of the loan-to-value cap, tax relief on mortgage interest and amortisation requirement with high mobility on the housing market Deviations from long-term levels, in percentage points
Source: Own calculations
Repo rate Inflation
1 4 7 10 13 16 19 22 25 28 31 34 37 40 1 4 7 10 13 16 19 22 25 28 31 34 37 40
Figure 7. A tightening of the loan-to-value cap, tax relief on mortgage interest and amortisation requirement with low mobility on the housing market Deviations from long-term levels, in percentage points
Repo rate Inflation
-0.5
-0.4
-0.3
-0.2
-0.1
0.0
0.1
-0.5
-0.4
-0.3
-0.2
-0.1
0.0
0.1
Amortisation requirement Mortgage capTax relief on mortgage interest
Source: Own calculations
82 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
7Howdoestheleveloflong-termdebtaffectmonetarypolicy?TherapidlyincreasingindebtednessofSwedishhouseholdssincethemid-1990shasprobablymademanyhouseholdsmoresensitivetochangesininterestrates.Ifinterestratesrise,thishasadirecteffectonthehouseholds’disposableincome,andbringsaboutaredistributionofresourcesfromborrowerstolenders.Thehighertheindebtedness,thegreatertheeffectofaratehikeonborrowers’interestexpenseanddisposableincome,whichinturnreduceshouseholds’scopeforotherconsumption.However,theextentoftheimpactdependsonseveralfactors,suchashowlongaratehikelasts,andhowthedebtsaredistributedamongthehouseholds.
Accordingtothisargument,theimpactofapolicyratehikeondemand–andhenceoninflation–wouldbegreaternowthanbefore.Inordertoillustratethepotentialextentoftheeffects,wecomparehowtheeffectsdifferdependingonthreedifferentloan-to-incomeratiolevels,seeFigure8.Inallthreecases,thereporateisincreasedby1percentagepointtobethenadjustedbacktotheoriginallevelaccordingtothemonetarypolicyrule(seeequation(7)).Inthefirstcaseweassumethattheloan-to-incomeratioisatthepresentlevel,i.e.closeto180percent,whentheratehikeoccurs.Inthatcase,theaggregateconsumptionofhouseholds(bothborrowersandlenders)declinesbyaround0.7-0.8percentinthefirstyear.Buttheborrowersaretheoneshardesthitbytheratehike.Theirconsumptiondropsbyalmost2percent,whilethelenders’consumptionislargelyunaffected.Wecanalsoseethatinflationdropsbyjustover0.5percentagepoints.
Wecancomparethatcasewithwhattheeffectsonconsumptionandinflationwouldbehadtheloan-to-incomeratioinsteadbeenatthesamelevelaswhentheinflationtargetwasintroducedinthemid-1990s,i.e.around90percent.Inthiscase,theeffectonconsumptionwill,asexpected,besmaller,andinparticulartheeffectontheborrowers’consumptionwillbemuchsmaller.Theirconsumptiononlydropsbyacoupleoftenthsofapercentagepoint.Theeffectoninflationisalsosmaller,althoughthedifferenceisrelativelysmall.
Inthethirdcaseweassumealoan-to-incomeratioof210percent.Thatlevelcouldbereachedwithin10yearsiftheloan-to-incomeratiocontinuestoincreaseatthecurrentratefromtoday’slevel.Withthatdebtlevel,theeffectsofaratehikeonconsumptionwouldbemuchgreater.Forexample,aninterestratethatisonepercentagepointhighercouldleadtoadropinborrowers’consumptionofalmost6percent.Theimpactoninflationwouldalsobegreater.
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 83
Thesecalculationsprovideanindicationoftheimportanceofthehouseholdloan-to-incomeratioonthetransmissionmechanismofmonetarypolicy.Thehighertheloan-to-incomeratio,thegreatertheshareofborrowers’incomethatwillbespentoninterestpaymentsandconsumptionofhousingservices,andthelowertheamountthatcanbespentonconsuminggoods.
1 4 7 10 13 16 19 22 25 28 31 34 37 40 1 4 7 10 13 16 19 22 25 28 31 34 37 40
Figure 8. Effects of hiking the repo rate at different loan-to-income ratios Deviations from long-term levels in per cent, in percentage points (repo rate, inflation) and per cent (consumption)
Source: Own calculations
Repo rate Consumption
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
Loan-to-income ratio in the mid-1990sLoan-to-income ratio 2024 if the rate of increase continues at the current rateCurrent loan-to-income ratio
1 4 7 10 13 16 19 22 25 28 31 34 37 40 1 4 7 10 13 16 19 22 25 28 31 34 37 40
Borrowers’ consumption Inflation
-6
-5
-4
-3
-2
-1
0
1
2
-0.25
-0.20
-0.15
-0.10
-0.05
0.00
0.05
84 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
8ConcludingremarksWehaveshownthat,overtime,householddebtcouldbereducedbytighteningdifferentmacroprudentialtoolsandreducingthetaxreliefonmortgageinterest.Thiswouldalsoleadtoaredistributionofresourcesfromlenderstoborrowers.Thisredistributioneffectiscontrarytotheconventionalwisdomaccordingtowhichmacroprudentialmeasuresarenegativeforborrowers.ForlatiandLambertini(2014)describesimilarfindings.TheystudytheeffectsofthenewloancontractswithlowerdownpaymentsandloweramortisationrequirementsthatcameintouseintheUSinthelatterhalfofthe1990s.Intheirmodel,easieraccesstocreditpushesborrowerstoborrowmoreanddemandmorehousingservices,withoutconsideringtheimpactoftheirbehaviouronhousingprices.Intheaggregatetheresultinghigherhousingdemandwillpush-uphouseprices.Asaresult,borrowerswillhavetocutdowntheirconsumptionandworkmore,therebyreducingtheirwelfareinthelongterm.
Wehavealsoshownthatareductioninthetaxreliefonmortgageinterestaffectshouseholdsindifferentways,dependingonhowtheGovernmentchoosestousethemoneyinstead.Ifitcompensatestheborrowers,suchameasurecanbepositivefortheirconsumptionotherthanhousing.Hence,thosethatmaybehurtfinanciallyofareductioninthetaxreliefcouldbecompensatedwithothertransfers.
Acaveatofouranalysisisthatwedisregardtheimpactofhouseholddebtontheriskofenteringafinancialcrisis.Themainpurposeofmacroprudentialpolicyisindeedtopreventfinancialcrises,andthatisalsoimportantfromamacroeconomicperspective.Thisishoweversomethingwecannotquantifyinthemodelwehaveused.
Anotherconclusionfromtheanalysisis,notsurprisingly,thatthedifferentmeasureshaveimplicationsformonetarypolicythatcangoindifferentdirectionsandareaffectedbytheextentofmobilityonthehousingmarket.Usingthemodel,wehavealsoillustratedhowthelevelofhouseholddebtaffectstheimpactofmonetarypolicyontherestoftheeconomy.
Alimitationinthisstudyisthatwehavedisregardedforeigntradeandpossibilitiesofcross-bordertransactions.TheresultsinChenandColumba(2016),however,showthatthisdoesnotaffecttheresultstoanyconsiderableextentinamodelsimilartoours.Theyalsoanalysetheeffectsofvariousmacroprudentialmeasuresonhouseholddebt,andtheirfindingsareinlinewithours.
TheNationalInstituteofEconomicResearch(2016)hasrecentlyconductedastudytoassessthemacroeconomicimpactofintroducingaloan-to-incomecap.Assumingthatallborrowersborrowuptothelimit,theyconcludethatthe
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 85
introductionofthenewpolicywillleadtoadecreaseinthehouseholdloan-to-incomeratioby11-14percentagepointsafter10yearswhilenominalGDPwillbe0.1-1.7percentlower,andthepricelevelmeasuredbytheCPIF0.1-0.7percentlower.Accordingtoourmodel,realGDPdeclinesby0.4percentinthelongterm.
Anotherstudythatlooksathowhouseholddebtisaffectedbymacroprudentialandfiscalpolicies,i.e.apropertytaxandtaxreliefonmortgageinterest–isAlpandaandZubairy(2016).TheirmodeliscalibratedonUSdataandhasarelativelydetaileddescriptionofthemortgagemarket,butlacksabankingsectorwhichintermediatesfundsbetweenlendersandborrowers.Accordingtotheiranalysis,themostefficientpolicyisreducingthemortgageinterestduduction,followedbytighteningtheloan-to-valuecapandintroducingapropertytax.
Inourmacromodel,wemakethesimplifiedassumptionthatborrowersarealwayslimitedbyaloanrestriction.BothHull(2015)andSvensson(2016)divergefromthatassumptionandstudyeffectsofamortisationrequirementsusing–insomeregards–moredetailedmodelsoftheloancontracts.Hull(2015)studiestheamortisationrequirementintroducedbyFinansinspektionenon1June2016.Inhismodel,besidesamortisationrequirements,householdsmustalsomeetadebtserviceratiocap.Theanalysisisconductedinan“overlappinggenerationmodel”with60generations,inwhichtheloancontractsofthehouseholdsectorarelargelyconsistentwiththeactualdesignofloancontractsinSwedenwithSwedish-stylemortgagecontracts.AccordingtoHull’sanalysis,amortisationrequirementsleadtosomewhatsmallereffectsthaninourmodel(thehouseholdloan-to-incomeratiodecreasesbyjustover2percentagepointsatmost),becausethehouseholdscanoptimallychoosetoregularlyrenegotiatetheirloans.
AccordingtoSvensson(2016)householddebtcould,onthecontrary,increaseratherthandecreaseifanamortisationrequirementisintroduced.Thisisbecausehouseholdscanchoosethesamesavingsaswithoutamortisationrequirementsbyborrowingmoreinitially,investingthesurplusinasavingsaccountandthenmakingwithdrawalsattheraterequiredtomeettheamortisationrequirement.Theaveragedebtwillthereforebehigherthanwithoutamortisationrequirements.AcrucialassumptioninSvensson’sanalysisisthatmortgagesarenotlimitedatthesametimebyloan-to-valuecaps,loan-to-incomecapsorotherfactorsthatpreventborrowersfromfinancingtheiramortisationswithloans,suchasthecostofrenegotiatingloans.
Justlikeinthetheoreticalliterature,theempiricalliteraturecontainsahighdegreeofuncertaintyaboutwhicheffectsvariousmacroprudentialtoolshave.Guibourgetal.(2015)summarisethefindingsofsomeimportantempiricalstudies.Severalofthesestudiesshowthatbothloan-to-valuecapsandloan-
86 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
to-incomecapscanbeeffectiveinpreventinghouseholdmortgagesfromrisingtooquickly.Loan-to-valuecapsandloan-to-incomecapsalsoappeartobethemacroprudentialtoolsmostusedtoreducedemandforcredit,andoftenincombinationwitheachother.Housing-relatedtaxessuchaspropertytaxortaxreliefonmortgageinterestalsohaveeffectsonmortgages.Housing-relatedtaxesandloan-to-valuecapsalsoappeartohavesignificanteffectsonhouseprices.
Inthisstudy,wehavequantifiedbyhowmuchdifferentmacroprudentialpoliciesandmortgageinterestdeductionshavetobetightenedinordertoattainacertaineffectontheloan-to-incomeratio.Themacroeconomiceffectsofthesedifferentmeasureswillbefairlyequalforthepopulationonaverage,butdifferbetweenborrowersandlenders.
S v e r i g e S r i k S b a n k e c o n o m i c r e v i e w 2016:2 87
ReferencesAlpanda,SamiandSarahZubairy(2016),“Addressinghouseholdindebtedness:monetary,fiscalormacroprudentialpolicy?”,revisedBankofCanadaWorkingPaperno.2014-58.
Chen,JiaqianandFrancescoColumba(2016),“MacroprudentialandmonetarypolicyinteractionsinaDSGEmodelforSweden”,IMFWorkingPaper16/74.
Emanuelsson,Robert(2015),“SupplyofhousinginSweden”,Economic Review,2015:2,SverigesRiksbank,pp.47-73.
Emanuelsson,Robert,OlaMelanderandJohanMolin(2015),“Financialrisksinthehouseholdsector”,Economic Commentaries,no.6.
Englund,Peter(2016),“Moreneutralcapitaltaxation:Distributioneffectsoflimitedtaxreliefonmortgageinterest”[En mer neutral kapitalbeskattning: Fördelningseffekter av begränsade ränteavdrag],ReportfortheSwedishFiscalPolicyCouncil2016/3.
Finansinspektionen(2015),“TheSwedishmortgagemarket2015”,reportfrom14April2015.
Finansinspektionen(2016),”TheSwedishmortgagemarket”,reportfrom14April2016.
Finocchiaro,Daria,MagnusJonsson,ChristianNilssonandIngvarStrid(2016),“Macroeconomiccostsofdeleveraging”,mimeo,SverigesRiksbank.
Forlati,ChiaraandLuisaLambertini(2014),“Mortgageamortizationandwelfare”,ÉcolePolytechniqueFédéraledeLausanneWorkingPaper.
Gerali,Andrea,StefanoNeri,LucaSessaandFedericoM.Signoretti(2010),“CreditandbankinginaDSGEmodeloftheeuroarea”,Journal of Money, Credit and Banking,vol.42,pp.107-141.
Guibourg,Gabriela,MagnusJonsson,BjörnLagerwallandChristianNilsson(2015),“Macroprudentialpolicy–effectsontheeconomyandtheinteractionwithmonetarypolicy”,Economic Review,2015:2,SverigesRiksbank,pp.29-46.
Hull,Isaiah(2015),“Amortizationrequirementsandhouseholdindebtedness:anapplicationtoSwedish-stylemortgages”,WorkingPaper298,SverigesRiksbank.
Iacoviello,Matteo(2005),“Houseprices,borrowingconstraints,andmonetarypolicyinthebusinesscycle”,American Economic Review,vol.95(3),pp.739-764.
NationalInstituteofEconomicResearch(2016),“Short-termmacroeconomiceffectsofcredit-limitingmacroprudentialmeasures”[Kortsiktiga makroekonomiska effekter av kreditbegränsande makrotillsynsåtgärder],Occasionalstudyno.50,May.
Kydland,Finn,PeterRupertandRomanSustek(2016),“Housingdynamicsoverthebusinesscycle”,International Economic Review, tobepublished.
88 M a c r o e c o n o M i c e f f e c t s o f r e d u c i n g h o u s e h o l d d e b t
Svensson,LarsE.O.(2016),”Amortizationrequirementsmayincreasehouseholddebt: asimpleexample”,IMFWorkingPaper16/83.
SverigesRiksbank(2014),MonetaryPolicyReport,July2014.
SverigesRiksbank(2015),MonetaryPolicyReport,September2015.
SverigesRiksbank(2016),FinancialStabilityReport2016:1.