韓國(guó)央行-考慮異質(zhì)性家庭利率敞口的加息對(duì)消費(fèi)的影響（英）

38

Ⅱ.KeyIssues

ImpactofInterestRateIncreasesonConsumption

ConsideringtheHeterogeneousHouseholds’InterestRateExposure

KEYTAKEAWAYS

I.Privateconsumptionremainssluggishduetohighpricesandinterestrates.Sincetheendof2022,themomentumofconsumptionrecoveryhasweakened,andrealinterestrateshaveclimbedabovethelevelsseenpriortotheBaseRatehikes.Inaddition,fromearlylastyear,theinterestratesonmortgageloansandunsecuredloansheldbyhouseholdsrosetoalevel2%pto3%phigherthanthelevelbeforetheratehikes,implyingthattheinterestrateincreasesareconstrainingconsumption.

II.[Averagetreatmenteffect:intertemporalsubstitution]Thedirectimpactofhigherinterest

ratesonconsumptioncanusuallybeexplainedbyhouseholds’intertemporalsubstitution(increasedsavings,decreasedcurrentconsumption),andconsideringthatinterest-bearingassetsofhouseholdssuchasdepositshaveincreasedsignificantlysinceearlylastyear,theintertemporalsubstitutioneffectappearstobeinplace.

III.[Individualeffect:interestrateexposure]Amidthismacroeconomicimpact,theimpactof

interestratechangeonhouseholdconsumptionvariesnoticeably,dependingontheextenttowhichhouseholdsareexposedtointerestraterisks.Toidentifythiseffect,anexaminationoftheinterestrateexposureandconsumptionofindividualhouseholdswasconducted,whichfoundthat,infact,theconsumptionofhouseholdswhoseshort-termfinancialliabilitiesexceedtheirshort-termfinancialassets(“l(fā)osersofhigherinterestrates”)contractedconsiderably.Amongthesehouseholds,thelargestshareswereoccupiedbyfamilieswithhouseholdheadsintheir30sto40s,thoseinthemid-orhigh-incomebrackets,andthosewithhighconsumption.Ontheotherhand,the“winnersofhigherinterestrates,”whichincludedlargersharesofpeopleintheir60sandpeoplewithhighincomeandlargeassets,increasedtheirconsumption.

IV.Thisdistributioncharacteristicofinterestrateexposureamonghouseholdsappearstobe

anotherconstraintontotalconsumptioninadditiontotheintertemporalsubstitutioneffect.Thisisattributedtothefactthat,whereasthe“l(fā)osersofhigherinterestrates”includearelativelylargershareofhouseholdswithhighpropensitytoconsume,the“winnersofhigherinterestrates”containmoreofhouseholdswithlowpropensitytoconsume.Theeffectofinterestrateincreasesonslowingconsumptionestimatedbythismodel(intertemporalsubstitution+interestrateexposure)isfoundtobeover20%largerthanwhenonlytheintertemporalsubstitutioneffectisconsidered.

V.Ifinterestratesdeclineaspricesstabilize,householdconsumptionisexpectedtorecover

gradually.Still,higherpricesdrivenbyaccumulatedinflationarypressurearelikelytounderminethepaceofconsumptionrecoverygoingforward.Ifinterestratesfallamidahighdebtratioofhouseholdsaged30-40,adequatepolicyeffortswillbeneededtokeephouseholddebtfromresurging.

39

Background

1.Recoveryinconsumptionremainedslowerthanexpectedthroughoutlastyear,duetotheimpactofhigherpricesandinterestrates.Themomentumofrecoveryinprivateconsumptionweakenedfromtheendof2022,andinthe4thquarteroflastyear,privateconsumptionRemainedwellbelowthelevelanticipatedbeforeglobalinflationaswellasthepandemic.Consumptionisinfluencedbymultiplefactors:economicfactorssuchashouseholdincomeandinflationexpectationsandstructuralfactorssuchasdemographicstructure,potentialgrowthrate,andhouseholdconsumptionpatternsafterthepandemic.However,consideringthatinterestratespaidbyhouseholdsremainedabout2%pto3%phigherthanbeforetheinterestratehikesandthatrealinterestratesalsoclimbedbyaround1.5%p,itislikelythattheeffectofhigherinterestsonconsumptionhasemerged.Thus,thissectionexaminestheimpactofinterestratehikesonconsumption.

78

Consumptionrecoverymomentumhasweakenedsincetheendof2022

Figure1.Trend1)2)andperformanceofprivate

consumption

Notes:1)Dottedlineindicatesthetrendof2015-2019.

2)Atrendlineestimatedusingthemid-term

projection(2023-2026)oftheIMF(April2021).Source:BankofKorea.

Realinterestratesandeffectivehouseholdinterestratesrisewithtimelag

Figure2.BaseRate,realinterestrates,1)and

effectivehouseholdloaninterestrates2)

Notes:1)Inflationdeductedfromhouseholdloaninterestrates(basedonbalance).

2)Basedonloanbalanceofdepositorybanks.Source:BankofKorea.

2.Thetrajectoryofhouseholdsincreasingtheirsavingsanddecreasingtheirconsumptionduringtheinterestrateincreases(intertemporalsubstitution)isappliedtoalleconomicagentsinthesamedirection,butactualchangesinconsumptionmayvaryduetothedifferentinterestrateriskscausedbydifferentfinancialconditionsofindividualhouseholds.Recently,ithasoftenbeensaidthattheeffectofinterestratesonconsumptionmayhaveexpandedduetothechangeintheenvironment,suchasaccumulatedhouseholddebt.Thus,thissection①identifieshouseholdsthatexperiencedasignificantconstraintonconsumptionbyanalyzingthemicrodataonhouseholds’assetsandliabilitiesanddetermineswhetherthesehouseholds’consumptionchangeshaveactuallycausedaggregateprivateconsumptiontoslow.Basedonthis,②thissectionassessesthelikelihoodthattheimpactoftherecentinterestratehikeshasbecomestrongerthantheaverageresponsesofthepast.Lastly,③theimpactofanticipatedinterestratedeclineonconsumptionisestimatedandimplicationsarederived.

78Thissectionanalyzestheimpactofinterestrateincreasesonconsumptionbylimitingittotheimpactofhouseholds’decision-makingbetweensavingsandconsumptionandchangesininterestearnings(interestburden).Thus,theimpactonconsumptionofchangeinrealhouseholdincomeduetointerestratehikesisexcludedfromthisanalysis.

40

RecentTrendsofConsumptionandSavings

3.Domesticprivateconsumptionremainedinaslumpsincelastyear,whichisprevalentinvarioussectorsregardlessofitemandhouseholdcharacteristics.Whiletheconsumptionofallgoods,suchasdurableandsemi-durablegoods,wassluggishfromlastyear,

79

servicesconsumption,whichhadbeenrelativelysolid,hasseenitsrecoverymomentumweakensincelastyearduetotheexhaustionofpent-updemand.Byhouseholdcharacteristics,aswell,despiteanincreaseinincomeacrossallincomequintilescomparedtopre-pandemic,consumptiongrowthfellshortofincomegrowth,leadingtoanoveralldeclineinthepropensityofhouseholdstoconsume.Thissuggeststhattheslowdownofconsumptioncannotbeexplainedbyonlythedeteriorationofincome.

Goodsconsumptionissluggishacrossallitems

Figure3.Goodsconsumptionby

typeofgoods1)

Notes:1)Dottedlineindicatestrendof2015-2019.

2)Figuresin()indicateweights.Source:BankofKoreaNationalAccounts.

Servicesconsumptionrecoverymoderates

Figure4.Goodsandservices

consumption

Note:1)Figuresin()indicateweights.Source:BankofKoreaNationalAccounts.

Households’propensitytoconsumeshowsanoverallweakening

Figure5.Change1)inincome2)and

consumptionexpenditurebyincomequintile

Notes:1)FromQ1toQ3,adjustedbyconsumerpriceindex(2020).

2)Sumofcurrentincome

(earned,business,propertyincome)andnon-currentincome.

Source:HouseholdIncomeandExpenditure

SurveyofStatisticsKorea.

4.Atthesametime,thesavingsrateofhouseholdsremainedhigherthantheaverageofthepast,withinterest-bearingassetssuchasdepositsraisinghouseholds’propensitytosave.Koreanhouseholds’netsavingsraterosesharplyduringtheearlystageofthepandemicandthendeclinedfortwoyears.After2022,whenthemajorityoftheinterestratehikesoccurred,therateofdeclineinthesavingsrateslowed,remainingabovethelevelobservedbeforethepandemic.Inaddition,becausehouseholdsincreasedtheirinterest-earningassets,suchasdepositsandbonds,andreducedtheirinterest-bearingliabilities,suchasloans,theratioofhouseholds’interest-earningassetstointerest-bearingliabilitiesclimbedsharplylastyear,suggestingthatintertemporalsubstitutionistakingplace(Figure7).

79Fordetaileddiscussiononresponsestointerestrateincreasesbyconsumptionpattern,referto“ImpactonDomesticDemandbySectorfromRateIncreases”(BOKIssueNoteNo.2022-27).

41

Households’netsavingsrateremainshigherthanthatpre-pandemiclevel

Figure6.Households’netsavingsrate

Source:BankofKorea.

Ratioofhouseholds’interest-bearingassetstoliabilitiesrosesharplylastyear

Figure7.Ratioofhouseholds’interest-bearing

assetstoliabilities1)andbaserate

Note:1)Assets:transferabledepositsandnon-transferabledeposits,bonds,derivatives,andotherfinancialassets.

Liabilities:loans,governmentloans,derivatives,

andotherfinancialliabilities.Source:BankofKorea.

5.Giventhatconsumptionisbroadlysluggishregardlessofitemandhouseholdcharacteristics,theintertemporalsubstitutionchannelisjudgedtobeontracktoacertainextent.Thatis,theintertemporalsubstitutionduetohigherinterestrates,aswellasotherfactorssuchasslowingwagesandhigherprices,appearstohavedriventherecentslumpinprivateconsumption.Furthermore,consideringthefactthattheshareofvariable-rateloansindomestichouseholddebtisrelativelylarge,

80

householdsinKoreawouldbemoreaffectedbyhighinterestratesthaninothermajorcountries.

80Portionofvariable-ratemortgageloansbycountry(Q42022,basedonloanbalance,%):Korea65.1,Sweden51.0,Denmark35.8,UK13.1,UnitedStates6.1

42

ImpactofUnhedgedInterestRateExposureHeterogeneity

6.Whiletheintertemporalsubstitutionchannelaffectsallhouseholdsinthesamedirection,thereisachannelthroughwhichinterestratehikesexertdifferentimpactsondifferenthouseholds.

7.Toexaminethis,theunhedgedinterestrateexposure(URE)ofhouseholdsismeasured.

81

Itisfoundthattherearehouseholdsthatearnfinancialgainuponaninterestrateincrease(positiveURE)andhouseholdsthatsufferloss(negativeURE)(Figure8).Classifyingthehouseholdsinto10decilesbasedontheirUREs,thecloseronesaretothefirstdecile,i.e.theyhavealargernegativeURE,themorelikelytheyaretobeadverselyaffectedbyhigherinterestrates.Conversely,thecloseronesaretothe10thdecile,themorelikelytheyaretobenefitfromit.Lookingatassetsizeandcomposition,whilehouseholdsinthe9th-10thdecilehavemoreliquidandilliquidassets,thoseclosertothe1stdeciletendtoholdmoreilliquidassets

82

andlessliquidassets,showinganunbalancedassetstructure.Themiddledecilesareoverallneutraltointerestratechangebecausetheirassetsanddebtsaresmallduetotheirlimitedaccesstothefinancialmarket.Low-incomevulnerablehouseholdsusuallybelongtothesemiddledeciles.

TheimpactofinterestratehikesonhouseholdsaredifferentiatedbytheirUREs

Figure8.DistributionofUREsamonghouseholds

(bydecile)1)

Note:1)SurveyofHouseholdFinancesandLiving

Conditions(2023),calculatedbasedonhouseholdswhosehouseholderisaged30to50.

Sources:BankofKorea,StatisticsKorea,

BankofKoreaResearchDepartment.

SizeandcompositionofhouseholdassetsareclearlydifferentiatedbyURE

Figure9.LiquidandIlliquidassetsbyUREdecile1)2)

Notes:1)SurveyofHouseholdFinancesandLivingConditions(2023).

2)Forclassificationofliquidandilliquidassets,refertoFootnote82.

Sources:BankofKorea,StatisticsKorea,

BankofKoreaResearchDepartment.

8.Marginalpropensitytoconsume

83

isalsodifferentiatedbythisdistinction.Generally,consumptionofhouseholdswhohaverelativelylessliquidassetismoresensitivetoincomeshocks(highmarginalpropensitytoconsume).Infact,marginalpropensitytoconsume,estimatedusinghouseholdpaneldata,wasinlinewiththisexpectation.

84

Thatis,whilehouseholdswithnegativeURE(losersofhigherinterestrates)

81Interestrateexposure(URE)wasmeasuredfornetfinancialassets,basedonassetsandliabilitiesthatmaturewithinoneyear,accordingtoAuclert(2019)(Fordetails,refertoBox1).

82Cashandsavingsdepositsareclassifiedasliquidassets,whileotherfinancialassetsandrealassetsareclassifiedasilliquidassets.

83Thisisdefinedastheproportionofanincreaseinincomethataconsumerspendsonconsumptionandiscalculatedasthechangeinconsumptiondividedbythechangeinincome.

84Theestimationofmarginalpropensitytoconsumeduetotransitoryincomeshocks,usinglong-termhouseholdpaneldata(KoreanLaborandIncomePanelStudy,13thto23rdyear),foundthatthemarginalpropensitytoconsumeincreasesasliquidassetholdingsdecrease,showingamonotonicfunction(Fordetails,refertoBox2).

43

orneutralURE(vulnerablegroup)haveahighermarginalpropensitytoconsume,householdsintheURE9th-10thdeciles(winnersofhigherinterestrates)showedlowermarginalpropensitytoconsume.Reflectingthesecharacteristicsofassetsandmarginalpropensitytoconsumeaswellasacademicdiscussions,

85

asinTable1,householdsaredividedintolosersofhigherinterestrates(URE1st-3rddeciles:wealthyhand-to-mouth,W-HtM),vulnerablegroup(5thdecile:poorhand-to-mouth,P-HtM),andwinnersofhigherinterestrates(URE9th-10thdeciles:wealthy).

86

Table1.ClassificationofHouseholdsbytheirUREs

Classification

Unhedgedinterestrateexposure

Illiquidassets

Liquidassets

Marginalpropensitytoconsume

Losersofhigherinterestrates(W-HtM)

Negative(-)

More

Less

High

Vulnerablegroup(P-HtM)

-

Less

Less

High

Winnersofhigherinterestrates(wealthy)

Positive(+)

More

More

Low

9.Losersofhigherinterestrates(W-HtM)areconcentratedamonghouseholdswithhouseholdheadsintheir30sto40s,thoseintheuppermiddle-incomebracket(income4th-7thquintiles),andthoseinthehigh-consumptionbracket(consumption6th-10thdeciles).Inaddition,theyshowedahigherproportionofhomeowners,higherproportionofresidencyinmetropolitancities,higherdebt,andlargershareofrealestate-securedloans.

87

Comparedwiththewealthygroup,theW-HtMgroupisyoungonaverage,withmoderatelylowerincomethanthewealthygroup,butthereisnotmuchdifferenceintermsoftheproportionofhomeownersandconsumptionlevel.

W-HtMishighinboth

consumptionandmarginalpropensitytoconsume

Figure10.Consumption1)andmarginal

propensitytoconsume2)byURE

Notes:1)Excludingexpenditureondurablegoodssuchasautomobiles.

2)Forestimationmethod,refertoBox2.Sources:BankofKorea,StatisticsKorea,

BankofKoreaResearchDepartment.

ProportionofhomeownershiprisesatbothendsoftheUREdistribution

Figure11.Shareofhomeownership

andageofhouseholderbyURE

Sources:BankofKorea,StatisticsKorea,BankofKoreaResearchDepartment.

W-HtMincludesahigherproportionofhouseholderswhoareintheir30sto40s

Figure12.AgedistributionofallhouseholdsandW-HtMhouseholds

Sources:BankofKorea,StatisticsKorea,BankofKoreaResearchDepartment.

85Householdmarginalpropensitytoconsumeisknowntodeclineasincomeincreases,and,inacademia,householdswithhighmarginalpropensitytoconsumearecalled“hand-to-mouth”consumers.However,Kaplanetal.(2014)foundthat,besideshouseholdswithsmallassetsizes,therearemanyhouseholdswithahighmarginalpropensitytoconsumeastheyholdlittleornoliquidassetsbutsizableamountsofilliquidassets,anddefinedthesehouseholdsas“wealthyhand-to-mouth.”

86BasedonPark(2017)andSong(2020),whichestimatedKorea’sproportionofW-HtMat25%to30%(majorcountriessuchastheUnitedStatesputitataround20%),theURE1st-3rddecilesareclassifiedasW-HtM,andURE9th-10thdecilesaswinnersofhigherinterestrates.NoqualitativedifferencewasfoundevenifthegroupisadjustedtotheURE8th-10thdeciles.Householdsthatarelikelytomovetoothergroupsarenotclassified.

87Shareofhomeownership:Total69%,W-HtM79%;Shareofmetropolitancities:Total51%,W-HtM57%;Shareofrealestate-secureddebtamongtotaldebt:Total50.4%,W-HtM58.8%

44

10.Thevulnerablegroup(P-HtM)andwinnersofhigherinterestrates(wealthy)aremadeupoflargerproportionsoftheelderly.Notably,intermsofassets,theelderlyoccupiesadisproportionatelyhighshareofthelowest(1stdecile)andtop(10thdecile)assetpositions,showingpolarization.Ontheotherhand,intermsofincomeandconsumption,theelderlyisconcentratedinthelowerdeciles,provingthattheelderly,eventhosewithsizableassets,donotshowahighlevelofconsumption.

Theelderlyoccupieslarger

proportionsoftheP-HtMandwealthygroups

Figure13.Classificationofhouseholdsbyagegroup

Thedistributionofassetsamongthoseintheir60sispolarized

Figure14.Distributionofassetsofthoseintheir60sbyassetdecile1)

Householdsexhibitdistinctivecharacteristicsbyage

Table2.Averagestatisticsbyage

group1)2)

Sources:BankofKorea,StatisticsKorea,Note:1)Distributionofthoseintheir60s

BankofKoreaResearchDepartment.relativetodistributionofallage

groupsbyassets,i.e.,ifthe

distributionisthesameasthatofallagegroups,10%isallocatedtoeachdecile.

Sources:BankofKorea,StatisticsKorea,BankofKoreaResearchDepartment.

(KRWmillion)

Age

Income

Consumption

Assets

30s

65

28

363

(49)

(14)

(407)

40s

76

36

513

(69)

(18)

(623)

50s

76

33

542

(62)

(19)

(730)

60s

61

25

541

(54)

(15)

(785)

Notes:1)SurveyofHouseholdFinancesandLivingConditions,averageof2020-2023.

2)Figuresin()indicatestandarddeviations.

Sources:BankofKorea,StatisticsKorea,BankofKoreaResearchDepartment.

11.Giventhesecharacteristics,changeinconsumptionduetotheUREchannelisexpectedtobemostprominentamongthelosersofhigherinterestrates(W-HtM)andrelativelylimitedamongthevulnerablegroup(P-HtM)andwinnersofhigherinterestrates(wealthy).TheW-HtMgroupfacesgreaterfinancialstrainduetotheirliquidliabilitiesaftertheinterestratehikesandtheirhighmarginalpropensitytoconsumefurtheramplifiesthenegativeimpactonconsumption.Inaddition,astheW-HtMgroupalsohasahighabsolutelevelofconsumption,theircontributiontotheslowdownofaggregateconsumptionislikelytobesignificant.Ontheotherhand,whiletheP-HtMgrouphashighmarginalpropensitytoconsume,thefinancialeffectofhigherinterestratesisneutral,andtheirshareofaggregateconsumptionisnotlarge.Asforthewealthygroup,althoughinterestratehikeshavealargepositiveeffectonthem,itisdifficultfortheeffecttocarryovertoanincreaseinaggregateconsumptionasthegroup’smarginalpropensitytoconsumeislow.

45

12.Asforchangesinhouseholdconsumptionafterthepandemic,

88

therecoveryofconsumptionisfoundtobethemostsluggishamongthelosersofhigherinterestrates(W-HtM).Ifconsumptiongrowthin2022isexaminedexcludingincome,thedeclineinconsumptioninthevulnerablegroup(P-HtM)wasrelativelymoderatecomparedtotheW-HtMgroup,whilethewinnersofhigherinterestrates(wealthy)increasedtheirconsumptionslightly(Figure15).Thisconsumptionvariationisinlinewiththeexpectedpatternoftheimpactofinterestratehikes,includingtheUREchannel.Thatis,withtheintertemporalsubstitutionchannelconstraininghouseholdconsumptionoverall,theconsumptionofW-HtM,whoaresignificantlyaffectedbytheURE,remainsremarkablysluggish.ThissuggeststhatinterestrateincreasesarealsoinfluencinghouseholdconsumptionthroughtheUREredistributionchannel.

Consumptionafterthepandemicisthemostsluggishamonglosersofhigherinterestrates,inlinewithexpectationsassociatedwithinterestratehikes

Figure15.ChangesinconsumptionbyURE

(Actualdata,2022vs.2019)1)2)

Notes:1)CommonpanelhouseholdsintheSurveyofHouseholdFinancesandLivingConditions,2020-2023.Consumptiongrowthrateiscalculatedaftercontrollingandremovingincomefromhouseholdexpenditure.

2)EachdecileisdeterminedbasedontheUREasofMarch2020.Hence,variationinconsumptionforeachdecilereflectschangesintheUREitselfassociatedwithchangesintheconfigurationofassetsandliabilitiesduringtherelevantperiod.

Sources:BankofKorea,StatisticsKorea,

BankofKoreaResearchDepartment.

Figure16.ChangesinconsumptionbyURE

(Predictionsofthemodelfora1%riseininterestrates1))

Note:1)CommonpanelhouseholdsintheSurveyofHouseholdFinancesandLivingsConditions,2020-2023.Predictedchangeinconsumptionduetoa1%interestratehikeshock.Theinterestrateshockofthemodelassumesanexperimentalshockwhereinterestratesriseby1%inthecurrentyearandtheincreasedisappearsinthefollowingyear(assumingtheintertemporalelasticityofsubstitutionparameterσ=0.3).Itneedstobenotedthatthisisdifferentfroman

actualinterestrateshock.

Source:BankofKoreaResearchDepartment.

88IntheconsumptionexpenditureoftheSurveyofHouseholdFinancesandLivingConditions,the2022dataarethelatest,andthusthechangeinconsumptionfrom2023isdifficulttoidentify.However,thedataonconsumercreditcardspendingshowthattheconsumptionexpenditureofhouseholdsholdinghighdebtcontinuedtobeparticularlysluggishin2023aswell,implyingthattheconsumptionslumpamongthelosersofhighinterestrateslikelypersistedinthatyear.

46

13.IftheimpactoftheUREredistributionthroughthemodelissummedupforallhouseholds,theneteffectoftheUREchannelmovesinthesamedirectionastheintertemporalsubstitutionchannel,thatis,insuchawaythatinterestratehikesreduceaggregatehouseholdconsumption.ThisisduemainlytothefactthatthedecreaseinconsumptionintheW-HtMgroupwithhighmarginalpropensitytoconsumeoutpacedtheincreaseinconsumptionofthewealthygroup.Thenegativeeffectofhigherinterestratesonconsumption,aspredictedbythismodel(intertemporalsubstitution+URE),isfoundtobe20%largerthanwhenonlytheintertemporalsubstitutionisconsidered,implyingthatthemagnitudeofUREisnotsmall

89

(Figure16).

14.TheeffectofinterestratehikesincorporatingtheUREchannelislikelytobesomewhatlargerthantheestimatesbasedonmacroeconomicdata.Althoughthereishighuncertainty,theresponseofconsumptiontointerestrateshocksunderAuclert(2019)largelyexceedstheresponsesestimatedbyamacroeconomicmodelbasedon20yearsofdata.TheseresultssuggestthattheeffectofinterestratesonconsumptionisreinforcedthroughtheredistributionchannelandcouldfurtherexpandiftheUREsaredisproportionatelynegativeduetohighhouseholddebtwithagreatershareofvariable-rateloans.

89Themagnitudeoftheelasticityofconsumptiontointerestrates,basedontheredistributionchannelofURE,isabout-0.06,meaningthata1%pincreaseininterestratesreducesthehouseholdconsumptiongrowthratebyabout0.06%p,andtherelativesizeinthebasicmodelisestimatedatabout23%oftheintertemporalsubstitutionchannel.Thisfigureisthemedianoftheparametersoftheintertemporalelasticityofsubstitutionbyapplyingtherangewidelyusedintheliterature(0.1to0.5).Theestimatesofinterestrateelasticitybychannelareasfollows.

?UREchannel:?Intertemporalsubstitutionchannel:

ci:consumptionexpenditureofhouseholdi,MPC:marginalpropensitytoconsume,σ:parameterofelasticityofintertemporalsubstitution

Iftherange(0.1to0.5)isappliedtoσ,therelativesizeis15%to70%,whichisslightlylowerthantheU.S.estimate(30%to130%)(Auclert2019).Itisnoted,however,thattheestimatesarehighlyuncertainduetouncertaintysurroundingthemarginalpropensitytoconsume,differencesbetweennationaldata,andpossibilityoferrorinmodeling.

47

ConclusionandImplications

15.Thehigherinterestratesarefoundtohaveslowedprivateconsumption.Whilehouseholdsdecreasetheirexpenditureonaverage,householdswithahighershareofshort-termfinancialdebtareexpectedtoexperienceagreaterconsumptionslowdown.Householdsthatreducedtheirconsumptionduetothedirectimpactofhigherinterestratesarefoundtobethelosersofhigherinterestrates(W-HtM),mostofwhichhavehouseholdheadsintheir30sto40sandareintheuppermiddle-incomebrackets.Ontheotherhand,thewealthyhouseholds,whichcanexpecttoearnprofitsfromassetsowingtointerestratehikes,havelowermarginalpropensitytoconsumeanddidnotincreasetheirconsumptionsignificantly.Thelow-incomevulnerablehouseholds(P-HtM)werenotsignificantlyaffectedbyinterestratehikes,becausetheyholdsmallamountsoffinancialassetsandliabilities.

90

Hence,ifalltheimpactsofhigherinterestratesonhouseholds’financialconditionsaresummedup,theUREchannelisalsofoundtohavecontractedconsumption,inadditiontotheintertemporalsubstitutionchannel.

16.Goingforward,declininginterestratesandstabilizingpricesareexpectedtohaveapositiveeffectonhouseholdconsumption.However,theelevatedlevelofpricesduetoaccumulatedpriceincreaseswilllikelyweighontherecoveryofconsumption.Ifinterestratesfallasthedebtratioofhouseholdswithhouseholdheadsintheir30sto40sremainshigh,policyeffortswillbeneededtopreventhouseholddebtfromrisingagain,

91

becausethesurgeindebtsofeconomicagentsactivelyengaginginconsumptioncouldraisehouseholds’exposuretointerestrates,limitingconsumptioncapacityinthelongtermandunderminingthevitalityofdomesticdemand.

90Theconsumptionofthelow-incomevulnerablegroupislikelytohavebeenaffectedbyhigherpricesandslowingincomeratherthanhi