Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
1NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Bank financing for SMEs from 2014 to 2019 : effect of changes in the law on lending
Ch. PietteM.-D. Zachary
Introduction
TheโLawโofโ21 December 2013,โamendedโbyโtheโLawโofโ21 December 2017,โaimsโtoโfacilitateโaccessโtoโbankโfinanceโforโsmallโandโmedium-sizedโenterprisesโ(SMEs).โToโthatโend,โitโcomprisesโaโrangeโofโprovisionsโintendedโtoโ correctโ anyโ informationโ asymmetriesโwhichโ couldโ affectโ theโ commercialโ relationshipโbetweenโ lendersโ andโborrowersโbyโspecifyingโtheโtransparencyโobligationsโincumbentโonโbothโparties.โThus,โfirmsโapplyingโforโcreditโhaveโ toโ supplyโ sufficientlyโ exhaustiveโ dataโ onโ theirโ financialโ statements,โ amongโ otherโ things.โ Forโ theirโ part,โlendersโmustโmakeโ availableโ toโ applicantsโ standardโ documentsโ correspondingโ toโ theโ variousโ formsโ ofโ creditโwhichโ theyโ offer 1.โ Inโ addition,โ aโ proposedโ creditโ agreementโ hasโ toโ beโ accompaniedโ byโ aโ briefโ informationโdocumentโ settingโoutโallโ theโcharacteristicsโofโ theโcontract.โTheโ lenderโmustโalsoโascertainโ theโmostโ suitableโtypeโofโcredit,โtakingโaccountโofโtheโfirmโsโfinancialโpositionโatโtheโtimeโofโconclusionโofโtheโcontractโandโtheโpurposeโofโtheโcredit.โTheseโprovisionsโaimโinโparticularโtoโstimulateโcompetitionโbetweenโcreditโinstitutionsโbyโmakingโitโeasierโforโfirmsโtoโcompareโtheโvariousโoptionsโavailableโtoโthem.โInโtheโeventโofโaโrefusalโtoโgrantโcredit,โtheโlenderโmustโinformโtheโfirmโofโtheโessentialโreasonsโforโthatโrefusalโorโtheโfactorsโinfluencingโtheโriskโassessment,โandโthatโinformationโmustโbeโtransparentโandโcomprehensibleโforโtheโfirm.โTheโlawโalsoโincludesโanโ articleโ onโ theโ rulesโ forโ fixingโ theโ prepaymentโ penaltyโ whichโ theโ lenderโ isโ entitledโ toโ demandโ fromโ anyโborrowersโwhoโdecideโtoโrepayโtheirโloanโearly.โTheโloanโagreementโmustโexplicitlyโmentionโtheโamountโofโtheโpenalty.โTheโlatterโisโdeterminedโbyโaโstandardisedโprocedureโaccordingโtoโaโmethodโofโcalculationโdefinedโinโaโcodeโofโconduct,โrevisedโin 2018โfollowingโchangesโtoโtheโlawโandโagreedโbetweenโemployersโโorganisationsโrepresentingโtheโinterestsโofโself-employedโworkersโ(theโSNI)โandโSMEsโ(UnizoโandโtheโUCM),โonโtheโoneโhand,โandโFebelfinโonโtheโother.โForโloansโforโanโinitialโsumโofโnoโmoreโthanโโฌโ2 million,โtheโpenaltyโmustโnotโexceedโsixโmonthsโโ interest.โAtโtheโtime,โtheโfinancialโsectorโhadโexpressedโcertainโreservationsโonโthisโ lastโprovision,โraisingโconcernsโofโaโrelativeโincreaseโinโtheโinterestโrateโonโloansโtoโSMEs.
Theโ lawโ alsoโ statesโ thatโ itsโ ownโ effectsโmustโ beโ reviewedโ everyโ twoโ years.โ Inโ thatโ connection,โ theโNationalโBankโisโaskedโtoโgiveโaโโpreliminaryโopinionโ 2.โTheโRoyalโDecreeโofโ10 April 2016,โwhichโlaysโdownโtheโreviewโprocedures,โalsoโspecifiesโthatโtheโBankโshallโprovideโstatisticalโdataโonโtheโloansโconcernedโbyโtheโlaw.
1โ Since 2018,โthisโrequirementโnoโlongerโappliesโtoโloansโofโlessโthanโโฌโ25,000,โprovidedโtheyโdoโnotโcompriseโaโpenaltyโclauseโandโareโnotโcoveredโbyโcollateralโorโguarantees.
2 AnโopinionโisโalsoโrequestedโfromโtheโFSMA,โtheโfinancialโdisputeโmediatorโOMBUDSFIN,โandโtheโorganisationsโrepresentingโsmallโbusinesses.
2NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Theโaimโofโ thisโarticleโ isโ toโ summariseโandโcommentโonโ theโvariousโdataโ consideredโ relevantโ forโmonitoringโlendingโ toโSMEsโbyโ residentโbanks,โnamelyโdataโonโ theโvolumeโofโ suchโ loans,โ theโ interestโ ratesโappliedโandโtheโconditionsโforโgrantingโloans.โHowever,โtheโexaminationโofโtheseโdataโisโnotโsufficientโinโitselfโtoโdetermineโwhetherโ structuralโ effectsโ emergedโ followingโ theโ lawโsโ entryโ intoโ forceโ;โ itโ isโ thereforeโ supplementedโ byโ anโeconometricโanalysis,โwhichโisโsummedโupโinโtheโsecondโpartโofโtheโarticle.โFinally,โtheโconclusionโdiscussesโanyโeffectsโofโtheโlawโonโtheโdynamicsโofโlendingโtoโSMEs.
Whenโthisโarticleโwasโwritten,โtheโdataโtakenโintoโaccountโforโtheโanalysisโwereโavailableโupโtoโtheโendโof 2019โforโsomeโvariablesโandโupโtoโFebruary 2020โforโothers,โi.e.โbeforeโtheโeruptionโofโtheโCovid-19 crisis.โThatโcrisisโisโliableโtoโresultโinโfundamentalโchangesโinโtheโsituationโfromโtheโpointโofโviewโofโbothโcreditโinstitutionsโandโfirms.โTheโconclusionsโdrawnโatโ theโendโofโ thisโ studyโwillโ thereforeโneedโ toโbeโ reassessedโ inโaโ fewโmonthsโโtime.โMeanwhile,โweโreferโinterestedโreadersโtoโtheโspecificโmonitoringโbyโtheโBankโconductsโpublishedโonโitsโwebsiteโasโpartโofโtheโObservatoryโforโCreditโtoโNon-financialโCorporations 1.
1. Recent developments in business credit
1.1 Volume of bank loans
Lendingโ byโ residentโ creditโ institutionsโ toโ residentโ non-financialโ corporationsโ isโ usuallyโ monitoredโ viaโ twoโstatisticalโ sources,โnamelyโ theโdataโobtainedโ fromโbankโbalanceโ sheetsโ andโ thoseโobtainedโ fromโ theโCentralโCorporateโCreditโRegister.โInโprinciple,โtheโfirstโofโtheseโtwoโsourcesโisโexhaustive,โbutโitโhasโtheโdisadvantageโofโonlyโprovidingโinformationโatโanโaggregateโlevel.โThisโdefectโisโovercomeโbyโtheโsecondโsource,โwhichโpermitsโaโbreakdownโofโtheโstockโofโbusinessโloansโaccordingโtoโtheโfirmsโโsizeโandโtheirโbranchโofโactivity.โTheseโstatisticsโareโcompiledโusingโseparateโcollectionโmethodsโandโareโbasedโonโdifferentโdefinitions 2โ:โnaturally,โthatโgivesโriseโtoโsomeโstatisticalโvariations.โItโshouldโalsoโbeโpointedโoutโthatโtheโanalysisโconcernsโcreditโusedโbyโfirmsโtakenโintoโ accountโ inโ theโCentralโ Creditโ Register 3โ andโ focusesโ onโ non-financialโ corporations,โ thusโ excludingโ firmsโclassifiedโinโtheโfinancialโservicesโbranch 4.
Nonetheless,โtheseโtwoโsources 5โprovideโaโpictureโwhichโtalliesโoverallโwithโdevelopmentsโinโoutstandingโbankโloansโtoโfirmsโlocatedโinโBelgium.โAsโillustratedโinโchart 1,โthoseโdevelopmentsโclearlyโbearโtheโmarksโofโcrisisโepisodesโ:โfirst,โ theโgreatโ recessionโof 2008 and 2009,โandโ thenโ theโ sovereignโdebtโ crisisโwhichโaffectedโ theโeuroโ areaโ in 2011 and 2012.โ Theseโ twoโ eventsโ hadโ aโ negativeโ impactโ onโ bothโ demandโ forโ loansโ andโ theโsupplyโofโcredit.โLaggingโsomewhatโbehindโtheโtrendโinโeconomicโactivity,โtheโvolumeโofโcreditโusedโbyโfirmsโhasโthusโcontractedโonโtwoโoccasionsโinโtheโpastโtenโyearsโ:โfirst,โbetweenโmid-2009 andโmid-2010,โandโthenโin 2013 or 2014,โdependingโonโtheโdataโsourceโconsidered.โAfterโthat,โ itโbeganโrisingโagainโtowardsโtheโendโof 2014,โandโthatโexpansionโacceleratedโfrom 2015โtoโmid-2018,โbeforeโslowingโdownโupโtoโFebruary 2020.
However,โtheโmovementsโinโtheโtotalโvolumeโofโloansโmaskโhighlyโheterogeneousโvariationsโaccordingโtoโfirmโsize.โ Inโ reality,โ theโ procyclicalโ characterโ ofโ theโ trendโ inโ creditโ primarilyโ concernsโ lendingโ toโ largeโ firms.โ Theโreductionโinโtheโtotalโvolumeโofโloansโduringโtheโtwoโcrisisโperiodsโisโalmostโentirelyโattributableโtoโthem.โTheโ
1โ https://www.nbb.be/doc/dq/kredobs/fr/ko_home.htm.2 First,โtheโCentralโCorporateโCreditโRegisterโisโconstantlyโupdatedโbyโcreditโinstitutions,โbutโtheโdataโmayโbeโrevisedโretrospectivelyโduringโtheโtwelveโmonthsโfollowingโtheirโsubmission.โThatโimpliesโthatโtheโexistingโstatisticsโforโtheโperiodโfromโMarch 2019โtoโFebruary 2020โmayโyetโbeโamended.โNext,โsomeโtransactionsโregardedโasโloansโunderโSchemeโAโareโexcludedโfromโtheโCentralโCorporateโCreditโRegisterโsโstatisticsโ:โthatโapplies,โforโinstance,โtoโclaimsโresultingโfromโrepoโtransactionsโinโsecurities.โFinally,โtheโcriteriaโforโgrantingโloansโviaโanโassociationโorโsyndicateโtoโaโcounterpartyโwithinโtheโassociationโareโnotโtheโsame,โasโtheโbanksโhaveโdirectโinformationโwhichโisโnotโcurrentlyโpassedโonโtoโtheโCentralโCorporateโCreditโRegisterโ;โthatโsometimesโleadsโtoโdifferencesโinโgeographicalโandโ/โorโsectoralโclassification.
3โ Creditโusedโisโdifferentโfromโcreditโauthorised,โwhichโconcernsโcreditโlinesโmadeโavailableโtoโfirmsโbutโnotโnecessarilyโusedโinโfull.4โ Moreโspecifically,โthisโconcernsโtheโbranchesโunderโNACEโcodeโโKโ.5โ TheโperiodโcoveredโendsโinโFebruary 2020โowingโtoโtheโavailabilityโofโdataโatโtheโtimeโofโconcludingโthisโanalysis.
3NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
volumeโofโbankโloansโcontractedโbyโsmallโfirms 1โshowsโmuchโmoreโstableโgrowth.โThereโhaveโbeenโtwoโdistinctโphasesโinโrecentโyearsโ:โ(1)โanโaccelerationโinโtheโcreditโexpansionโrateโfrom 2015โtoโtheโendโof 2017,โ(2)โfollowedโbyโaโdecelerationโwhichโpersistedโupโtoโFebruary 2020โ(latestโavailableโdata).โDuringโthatโsameโperiod,โtheโvolumeโofโloansโtoโmedium-sizedโfirmsโdeclinedโforโ12 consecutiveโmonths,โbetweenโDecember 2016โandโNovember 2017,โwhileโlendingโtoโlargeโfirmsโremainedโpositiveโalmostโthroughoutโtheโperiodโfrom 2015โtoโFebruary 2020.
Overall,โsinceโtheโbeginningโof 2014โโโi.e.โsinceโtheโentryโintoโforceโofโtheโLawโofโ21 December 2013 onโSMEโfinancingโโโ theโvolumeโofโ creditโusedโbyโmicroโcompaniesโandโ smallโfirmsโ takenโasโaโwholeโhasโ risenโbyโanโannualโ averageโofโ4.6โ%โ (similarโ toโ theโ riseโduring 2008-2013 whenโannualโgrowthโaveragedโ4.8โ%โ forโ thisโcategoryโ ofโ firms),โwhileโ lendingโ toโmedium-sizedโ firmsโ expandedโ byโ 0.6โ%โ (seeโ chart 2).โ Althoughโ lendingโtoโmedium-sizedโfirmsโdidโ increase,โ theโgrowthโ rateโwasโmuchโ lowerโ thanโ inโ theโ sixโprecedingโ yearsโ (2008-2013),โwhenโ itโaveragedโ7.6โ%.โLendingโtoโ largeโfirmsโexpandedโbyโanโaverageโofโ4.2โ%โbetween 2014โandโFebruary 2020 2,โfollowingโaโ1.3โ%โriseโbetween 2008 and 2013.
1โ Inโthisโstudy,โtheโtermโโsmallโfirmsโโincludesโbothโmicroโcompaniesโ(usingโtheโmicroโmodelโforโannualโaccounts)โandโsmallโfirmsโ(filingโaccountsโinโtheโabbreviatedโformat).
2 Inโaddition,โcreditโdevelopmentsโmayโvaryโgreatlyโfromโoneโbranchโofโactivityโtoโanotherโ(seeโannexโ2).โForโexample,โinโtheโcaseโofโsmallโfirms,โbankโlendingโexpandedโfasterโbetween 2014โand 2020โinโtheโenergyโsector,โrealโestateโactivitiesโandโtheโconstructionโindustry.โConversely,โgrowthโwasโslowerโinโtheโwholesaleโandโretailโtradeโsector.โAnnexesโ1 andโ2 provideโdetailedโstatisticsโonโcreditโdevelopmentsโinโtheโvariousโbranchesโofโactivityโandโforโtheโdifferentโfirmโsizeโcategories.
Chart 1
Growth rate of bank lending to firms
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ15
โ10
โ5
0
5
10
15
20
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ25
โ20
โ15
โ10
โ5
0
5
10
15
20
Bank loans according to credit institution balance sheets 1 and the Central Corporate Credit Register ยฒ(in %)
Growth rate of used credit: breakdown by firm size 3
(in %)
Credit institution balance sheets 1
Central Corporate Credit Register 2
Micro companies and small firms
Medium-sized firms
Large firms
Sourceโ:โNBBโ(CentralโCorporateโCreditโRegisterโandโcreditโinstitutionโbalanceโsheets).1โ Sumโofโflowsโoverโ12 monthsโinโrelationโtoโtheโoutstandingโtotalโrecordedโaโyearโearlier.2 Movingโaverageโoverโ4 quartersโofโtheโgrowthโrateโoverโthreeโmonths,โwhichโisโthenโannualised.โNoโaccountโwasโtakenโofโcreditโusedโbyโ
firmsโclassifiedโinโtheโfinancialโservicesโsector.3โ Firmsโusingโmicroโmodelsโforโtheirโannualโaccountsโareโconsideredโtoโbeโmicroโcompaniesโandโfirmsโsubmittingโanโabbreviatedโmodelโ
areโconsideredโtoโbeโsmallโfirms.โFirmsโsubmittingโfull-formatโaccountsโareโregardedโasโlargeโorโmedium-sizedโaccordingโtoโwhetherโtheyโexceedโoneโorโmoreโofโtheโthresholdsโdefinedโinโtermsโofโtheโnumberโofโworkersโ(50 FTE),โturnoverโ(โฌโ9โ000 000)โandโbalanceโsheetโtotalโ(โฌโ4โ500 000). Forโmoreโdetails,โseeโhttps://www.nbb.be/doc/ba/infomail/mail_f_50.pdf.
4NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
However,โ itโ shouldโ beโ notedโ that,โ inโ viewโ ofโ theโ procyclicalityโ ofโ theโ statisticalโ seriesโ describedโ here,โ andโmoreโspecificallyโthoseโconcerningโlargeโfirms,โdescriptiveโdataโcannotโbeโusedโonโtheirโownโtoโinterpretโtheseโdevelopmentsโ asโ beingโ genuinelyโ specificโ toโ theโ periodโ inโwhichโ theโ lawโ hasโ applied.โ Forโ thatโ purpose,โweโconductedโaโmoreโdetailedโanalysisโusingโanโeconometricโmethodโwhichโenablesโusโtoโdistinguishโbetweenโtheโeffectsโ specificโ toโ thisโperiodโandโ thoseโ relatingโ toโ theโchangesโ inโ theโeconomicโenvironmentโandโborrowingโrates.โTheโdetailsโareโsetโoutโinโtheโsecondโpartโofโthisโarticle.
1.2 Interest rates charged by credit institutions
Theโinterestโratesโchargedโbyโbanksโgrantingโcreditโtoโnon-financialโcorporationsโcanโbeโtrackedโviaโtheโresultsโofโtheโMIRโ(MonetaryโfinancialโinstitutionโInterestโRate)โsurvey.โChart 3 presentsโtheโfourโmainโseriesโestablishedโonโtheโbasisโofโthatโsurvey.โTheyโareโcalculatedโasโweightedโaveragesโofโvariousโtypesโofโinterestโrates 1,โnamelyโvariableโorโshort-termโratesโโโforโloansโofโlessโthanโorโmoreโthanโโฌโ1 millionโ(i.e.โloansโwithโanโinitialโfixed-rateโperiodโofโ lessโ thanโoneโyear),โmedium-termโ interestโ ratesโ initiallyโfixedโ forโaโperiodโofโbetweenโoneโandโfiveโyears,โandโfinally,โlong-termโratesโinitiallyโfixedโforโmoreโthanโfiveโyears.โInโBelgium,โtheโmedium-โandโlong-termโweightedโaverageโ interestโ ratesโareโonlyโestablishedโforโsumsโofโ lessโ thanโโฌโ1 millionโ (becauseโtheโvolumeโofโtheseโmaturitiesโisโtooโsmallโforโamountsโinโexcessโofโโฌโ1 million).
Chart 2
Annualised growth rate of used credit : breakdown by firm size 1, 2, 3
(inโ%)
2002-2007
0
1
2
3
4
5
6
7
8
2008-2013 2014-2020 2002-2020
Micro companies and small firms Medium-sized firms Large firms Total
Sourceโ:โNBBโ(CentralโCorporateโCreditโRegister).1โ Noโaccountโwasโtakenโofโcreditโusedโbyโfirmsโclassifiedโunderโfinancialโservices.2 Theโannualisedโgrowthโrateโisโdeterminedโasโfollowsโ:โanโaverageโisโcalculatedโonโtheโbasisโofโtheโquarterlyโgrowthโratesโ/โoverโaโthree-monthโperiodโ(fromโJanuary 2013,โtheโdateโfromโwhichโmonthlyโdataโareโavailable).โ(EncoursโTโ/โEncoursโT-1
-1),โdisregardingโtheโquartersโaffectedโbyโaโbreakโinโtheโseriesโ(namelyโtheโsecondโquarterโof 2012,โandโtheโfourthโquarterโof 2014).โThisโaverageโisโthenโannualised.
3โ Firmsโusingโmicroโmodelsโforโtheirโannualโaccountsโareโconsideredโtoโbeโmicroโcompaniesโandโfirmsโsubmittingโanโabbreviatedโmodelโareโconsideredโtoโbeโsmallโfirms.โFirmsโsubmittingโfull-formatโaccountsโareโregardedโasโlargeโorโmedium-sizedโaccordingโtoโwhetherโtheyโexceedโoneโorโmoreโofโtheโthresholdsโdefinedโinโtermsโofโtheโnumberโofโworkersโ(50 FTE),โturnoverโ(โฌโ9โ000 000)โandโbalanceโsheetโtotalโ(โฌโ4โ500 000). Forโmoreโdetails,โseeโhttps://www.nbb.be/doc/ba/infomail/mail_f_50.pdf.
1โ Forโmoreโdetailsโonโtheโmethodology,โgoโtoโhttps://www.nbb.be/doc/dq/mir/pdf/facteurs_en.pdf.
5NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Chart 3
MIR rates and OIS rates 1 : breakdown according to the initial fixation period(inโ%)
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ1
0
1
2
3
4
5
6
7
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ1
0
1
2
3
4
5
6
7
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ1
0
1
2
3
4
5
6
7
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ1
0
1
2
3
4
5
6
7
GapBorrowing rate 1-year OIS rate GapBorrowing rate 1-year OIS rate
GapBorrowing rate 3-year OIS rate GapBorrowing rate 5-year OIS rate
Loans of less than โฌ 1 million, at fixed interest rates, with an initial fixation period of between 1 and 5 years
Loans of less than โฌ 1 million, at fixed interest rates, with an initial fixation period of more than 5 years
Loans of less than โฌ 1million, at variable interest rates, with an initial fixation period of less than 1 year
Loans of more than โฌ 1 million, at variable interest rates, with an initial fixation period of less than 1 year
Sourcesโ:โNBBโ(MIRโsurvey),โEikon.1โ OvernightโindexโswapโโโOISโ:โtheโinterestโratesโonโloansโtoโprimeโbanks,โandโconsequentlyโanโapproximationโofโtheโratesโatโwhichโbanksโcanโraiseโfinance.โInterestโratesโonโshort-termโloansโwereโcomparedโwithโtheโ1-yearโOIS,โratesโonโmedium-termโloansโwereโcomparedโwithโtheโ3-yearโOISโandโratesโonโlong-termโloansโwereโcomparedโwithโtheโ5-yearโOIS.
6NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Theโ movementโ inโ theseโ interestโ ratesโ largelyโ reflectsโ theโ variousโ monetaryโ policyโ measuresโ adoptedโ inโ theโEurosystem.โInโaccordanceโwithโtheโtasksโentrustedโtoโitโbyโtheโTreatyโonโtheโFunctioningโofโtheโEuropeanโUnion,โtheโEurosystemโsโprimaryโaimโ isโ toโmaintainโpriceโstability.โTheโpolicyโ interestโ ratesโsetโbyโ theโECBโGoverningโCouncilโthereforeโgenerallyโmirrorโinflationโexpectations.โTheseโwereโrevisedโdownwardsโfollowingโtheโeruptionโofโtheโeconomicโandโfinancialโcrisisโin 2008,โasโwasโtheโoutlookโforโdemand.โTheโGoverningโCouncilโthereforeโmadeโsubstantialโcutsโ inโtheโpolicyโinterestโratesโfromโthenโon.โMoreover,โ inโsettingโitsโdepositโfacilityโ interestโratesโatโaโnegativeโ levelโ fromโJune 2014,โandโestablishingโ itsโextendedโsecuritiesโpurchaseโprogrammeโatโ theโbeginningโof 2015,โtheโEurosystemโtriedโtoโencourageโportfolioโrebalancingโinโfavourโofโlendingโand,โultimately,โtheโfinancingโofโtheโrealโeconomy.โAlthoughโtheโassetโpurchaseโprogrammeโwasโsuspendedโinโDecember 2018,โitโwasโresumedโinโNovember 2019โinโorderโtoโreinforceโtheโaccommodativeโeffectsโofโtheโinterestโrates.โInโaddition,โtheโtargetedโlonger-termโrefinancingโoperationsโ(TLTRO)โalsoโhelpedโtoโboostโbankโlendingโtoโbusinessesโ(andโconsumers)โinโtheโeuroโarea,โandโsimilarlyโinโBelgium.โTheseโmeasuresโmaintainedโfavourableโcreditโconditionsโinโtheโeuroโarea.โTheโfirstโseriesโofโtargetedโlonger-termโrefinancingโoperations,โintendedโtoโstimulateโlendingโtoโeuroโareaโcreditโinstitutions,โwasโannouncedโinโJune 2014โandโextendedโoverโaโ2-yearโperiod.โTheโsecondโseriesโofโoperationsโ(TLTROโII)โbeganโinโMarch 2016,โandโtheโthirdโseriesโ(TLTROโIII)โinโMarch 2019.
TheseโvariousโmeasuresโconsiderablyโreducedโtheโfundingโcostsโofโcreditโinstitutionsโinโBelgiumโandโinโtheโeuroโareaโasโaโwhole.โChart 3 showsโtheseโcostsโaccordingโtoโ theโOISโ rates 1โ forโmaturitiesโsimilarโ toโ thoseโofโ theโweightedโ averageโ interestโ ratesโ obtainedโ fromโ theโMIRโ survey.โ Theโdeclineโ inโ fundingโ costs,โwhichโoccurredโmainlyโfromโtheโsecondโhalfโof 2008,โwasโnotโentirelyโreflectedโinโtheโborrowingโratesโthatโBelgianโbanksโofferedโtoโresidentโenterprises.โTheโgapโbetweenโtheseโtwoโtypesโofโinterestโratesโwidenedโaโlittleโfurtherโfollowingโtheโnewโcutsโinโkeyโinterestโratesโintroducedโfrom 2011.โHowever,โinโtheโcaseโofโloansโatโvariableโinterestโratesโ(withโanโinitialโfixationโperiodโofโlessโthanโoneโyear),โthatโgapโstabilisedโfrom 2012 atโaโlevelโcloseโtoโ2โ%.โFurthermore,โtheโgapโisโhardlyโanyโgreaterโforโloansโofโlessโthanโโฌโ1 millionโcomparedโtoโloansโofโmoreโthanโโฌโ1 million,โwhichโsuggestsโthatโloansโforโsmallerโamountsโโโtypicallyโsoughtโbyโSMEsโโโwereโnotโaffectedโbyโaโhigherโinterestโrateโowingโtoโtheโlimitโonโprepaymentโpenalty.
Medium-โandโlong-termโfixedโinterestโratesโdeclinedโbyโmoreโthanโtheโinterbankโratesโin 2012 and 2013,โcausingโaโsimilarโreductionโinโtheโgapโinโrelationโtoโtheโcorrespondingโOISโrate.โSpreadโwithโrespectโtoโtheโOISโcontinuedโtoโ fall,โ reachingโaโ lowโpointโ in 2017โbeforeโedgingโbackโupโ in 2018โand 2019.โDuring 2019,โmedium-โandโlong-termโinterestโratesโfellโ toโhistoricallyโ lowโlevels,โ respectivelyโdecliningโtoโ1.20โ%โ(inโApril)โandโ1.46โ%โ(inโOctober),โindicatingโparticularlyโfavourableโcreditโconditionsโforโbusinesses.
1.3 Credit conditions
Apartโ fromโ theโ accommodativeโmonetaryโ policy,โ otherโ factorsโ alsoโ contributedโ toโ theโ cutsโ inโ interestโ ratesโchargedโbyโbanks,โandโmoreโgenerally,โtheโeasingโofโtheโconditionsโtoโbeโmetโbyโfirmsโseekingโtoโborrow.โAsโwellโasโinterestโratesโandโcommercialโmargins,โtheseโconditionsโincludeโmaximumโloanโamountsโ/โloanโperiods,โasโwellโasโtheโcollateralโrequiredโandโnon-interestโrateโcharges.
Theseโconditionsโwereโ tightenedโ initiallyโ in 2008,โwhenโ theโeconomicโandโfinancialโ crisisโ erupted,โandโagainโin 2011-2012 atโtheโtimeโofโtheโsovereignโdebtโcrisisโ;โthisโappliedโtoโbothโSMEsโandโlargeโfirms.โAccordingโtoโtheโresultsโofโtheโBankโLendingโSurveyโcoveringโtheโfourโlargestโcreditโinstitutionsโoperatingโinโBelgium,โtheirโsupplyโofโ loansโbecameโmoreโ limitedโ in 2008 owingโ toโ theโ risksโ (whichโ theyโ sawโasโhavingโ increasedโconsiderably),โhigherโfundingโcostsโandโtighterโbalanceโsheetโconstraints.โThisโlastโaspectโwasโconnectedโwithโtheโprocessโofโbalanceโsheetโconsolidationโandโsizeโreductionโonโwhichโtheโbanksโhadโembarkedโafterโtheโstartโofโtheโcrisis.
1โ Overnightโindexโswapโ(OIS).โThisโisโtheโinterestโrateโonโloansโtoโprimeโbanks.
7NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Betweenโ theโ endโ of 2013 andโ mid-2018,โ theโ successiveโ keyโ interestโ rateโ cutsโ andโ otherโ monetaryโ policyโmeasuresโdidโmuchโtoโencourageโbanksโtoโeaseโtheirโloanโcriteria.โFurthermore,โtheโupturnโinโeconomicโactivityโin 2014โand 2015โledโtoโaโdownwardโrevisionโofโtheโassessmentโofโcreditโrisk.โFinally,โtheโresultsโofโtheโbankโsurveysโprimarilyโrevealโanโincreaseโinโtheโpressureโfromโcompetitionโbetween 2015โand 2018,โpromptingโtheโbanksโ toโ offerโmoreโ favourableโ creditโ conditionsโ forโ borrowers.โ Theseโ developmentsโ areโ notโ specificโ toโ theโBelgianโmarketโ;โaโsimilarโpictureโalsoโemergedโinโotherโeuroโareaโcountries.โIn 2019,โtheโbanksโbeganโtoโtightenโtheirโcreditโconditionsโforโbothโlargeโfirmsโandโSMEs,โmainlyโowingโtoโtheโdeterioratingโriskโperceptionโinโtheโwakeโofโtheโslowdownโinโeconomicโactivity.
TheโgenerallyโfavourableโcharacterโofโcreditโconditionsโforโSMEsโinโrecentโyearsโisโconfirmedโbyโsurveysโamongโentrepreneurs,โsuchโasโtheโSurveyโonโtheโAccessโtoโFinanceโofโEnterprisesโ(SAFE)โconductedโeveryโsixโmonthsโinโtheโeuroโareaโcountries.โTheโresultsโofโthatโsurveyโ(presentedโinโchart 5)โinโfactโsuggestโaโmarkedโreductionโinโtheโobstaclesโtoโaccessโtoโbankโloansโbetweenโtheโfirstโhalfโof 2017โandโinโlimitedโapprovalsโtheโfirstโhalfโof 2018,โinitiallyโreflectedโinโaโdeclineโinโrejectionโratesโaccompanied,โfrom 2017,โbyโaโreductionโinโlimitedโapprovalsโofโloanโapplications.โAccordingโtoโtheโlatestโdata,โtheโobstaclesโareโstillโrelativelyโlowโinโhistoricalโterms,โalthoughโtheyโhaveโincreasedโsinceโmid-2018.
Chart 4
Credit standards applied to non-financial corporations : breakdown by firm size(weightedโnetโpercentages 1)
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ100
โ80
โ60
โ40
โ20
0
20
40
60
80
100
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ100
โ80
โ60
โ40
โ20
0
20
40
60
80
100
Cost of funds and balance sheet constraints 1
Trend in credit standardsPressure from competition 1
Risk perception 1
Risk tolerance ยฒ
Expected trend in credit standards
Large firms SMEs
Sourceโ:โNBBโ(Bankโlendingโsurvey).1โ Aโpositiveโ(negative)โpercentageโcorrespondsโtoโtighteningโ(easing)โofโcreditโconditionsโorโaโfactorโcontributingโtoโtheโtighteningโ(easing)โofโthoseโconditions.
2 Factorโincludedโforโtheโfirstโtimeโinโtheโsurveyโinโtheโfirstโquarterโof 2015.
8NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Chart 5
Obstacles impeding access to bank financing for SMEs ยน, ยฒ(percentagesโofโrespondents)
2010โS1
2010โS2
2011โS1
2011โS2
2012โS1
2012โS2
2013โS1
2013โS2
2014โS1
2014โS2
2015โS1
2015โS2
2016โS1
2016โS2
2017โS1
2017โS2
2018โS1
2018โS2
2019โS1
2019โS2
0
5
10
15
20
25
Did not apply because of possible rejection
Total
Received only a limited part of the amount requested
Refused because the cost was too high
Application rejected
Sourceโ:โECBโ(SAFE).1โ Fewerโthanโ250 workers.2 Proportionโofโfirmsโnotโapplyingโforโbankโcreditโbecauseโofโpossibleโrejection,โorโapplyingโforโaโloanโbutโonlyโreceivingโaโlimitedโpartโofโtheโamountโrequested,โrefusingโcreditโbecauseโtheโcostโwasโtooโhigh,โorโhavingโtheirโapplicationโrejected.
Chart 6
Firmsโ assessment of conditions governing access to credit : breakdown by firm size ยน(balanceโofโtheโpercentagesโofโfavourableโ(+)โandโunfavourableโ(โ)โresponses,โ4-quarterโmovingโaverage)
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ30
โ20
โ10
0
10
20
30
40
50
Small firms
Large firms
Medium-sized firms
Very large firms Sourceโ:โNBBโ(Quarterlyโsurveyโofโcorporateโcreditโconditions).1โ Smallโ=โ1-49 workersโ;โmediumโ=โ50-249 workersโ;โlargeโ=โ250-499 workersโ;โveryโlargeโ=โ500 workersโorโmore.
9NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Theโ improvementโ inโ creditโ conditionsโ isโ likewiseโ borneโ outโ byโ theโ quarterlyโ surveyโwhichโ theโNationalโ BankโconductsโonโaโsampleโofโBelgianโentrepreneursโ(seeโchart 6).โTheโresultsโofโthatโsurveyโshowโthatโtheโimprovementโwhichโbeganโin 2013 andโpersistedโupโto 2016โ(sinceโwhen,โentrepreneursโโsatisfactionโhasโgenerallyโstabilisedโatโhistoricallyโhighโ levels)โ concernedโbothโSMEsโandโ largeโfirms.โThatโ improvementโwasโalsoโapparentโ inโ theโvariousโaspectsโofโloanโagreements,โbeโitโinโtermsโofโinterestโrates,โnon-interestโrateโcharges,โloanโvolumesโandโcollateralโrequiredโfromโborrowersโ(seeโchart 7).โSMEsโthereforeโdoโnotโseemโtoโhaveโbeenโpenalisedโinโanyโwayโasโregardsโlendingโconditionsโcomparedโtoโtheโtermsโappliedโtoโlargeโfirmsโbetween 2016โand 2019.
Chart 7
Firmsโ assessment of credit criteria : breakdown by size 1
(balanceโofโpercentagesโofโresponsesโreportingโimprovementโ(+)โorโdeteriorationโ(โ),4-quarterโmovingโaverage)
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ30
โ20
โ10
0
10
20
30
40
50
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ35
โ30
โ25
โ20
โ15
โ10
โ5
0
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ35
โ30
โ25
โ20
โ15
โ10
โ5
0
5
10
15
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
โ45
โ40
โ35
โ30
โ25
โ20
โ15
โ10
โ5
0
5
Small firms
Interest rates
Medium-sized firms Large firms Very large firms
Non-interest rate charges
Collateral requirementsSize of the loan
Sourceโ:โNBBโ(Quarterlyโsurveyโofโcorporateโcreditโconditions).1โ Smallโ=โ1-49 workersโ;โmediumโ=โ50-249 workersโ;โlargeโ=โ250-499 workersโ;โveryโlargeโ=โ500 workersโorโmore.
10NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
2. Econometric analysis
Theโdevelopmentsโdescribedโinโtheโpreviousโsectionโareโattributableโtoโaโcombinationโofโvariousโfactors.โ Inโparticular,โtheโexpansionโofโcorporateโcreditโisโnaturallyโinfluencedโpartlyโbyโtheโmovementโinโinterestโratesโchargedโbyโbanks.โAsโ theโdescriptiveโ statisticsโ suggest,โ itโmayโalsoโbeโgreatlyโ influencedโbyโ theโeconomicโclimateโviaโbothโdemandโeffectsโandโsupplyโeffects.โDemandโeffectsโariseโinโparticularโbecauseโfirmsโinvestโmoreโ(less)โ inโperiodsโofโstrongโ(weak)โactivity,โandโbecauseโofโtheโprocyclicalityโofโtheirโ liquidityโneeds,โasโpurchasesโofโintermediateโgoodsโandโwageโpaymentsโmoveโinโlineโwithโeconomicโactivityโandโemployment.โInโ addition,โ asโ aโ deterioratingโ economicโ situationโ isโ accompaniedโ byโ anโ increasedโ creditโ riskโ concerningโsomeโ firms,โ banksโ tendโ toโ tightenโ restrictionsโ onโ thoseโ whichโ areโ struggling,โ e.g.โ byโ imposingโ higherโriskโ premiums.โ Regardingโ interestโ rateโ levelsโ andโ trends,โweโ haveโ seenโ earlierโ thatโ theyโ closelyโ trackโ theโmovementโinโbanksโโfundingโcosts,โwhichโareโthemselvesโdeterminedโmainlyโbyโtheโinterestโratesโprevailingโonโtheโmoneyโmarket.
Asโ alreadyโ pointedโout,โ identifyingโhowโ theโ Lawโofโ 21 December 2013 hasโ affectedโ lendingโ volumesโ andโinterestโrateโlevelsโentailsโisolatingโitsโimpact,โifโany,โfromโthatโofโtheโotherโfactorsโmentionedโabove.โHowever,โsuchโ anโ exerciseโ isโ notโ easyโ sinceโ thatโ lawโmayโ influenceโ businessโ creditโ inโ variousโwaysโ:โ bothโ positively,โbyโ strengtheningโ competitionโ betweenโ creditโ institutions,โ promptingโ themโ toโ adaptโ theirโ tariffโ strategiesโtoโmakeโ themโmoreโ favourableโ toโ borrowers,โ andโ negatively,โ byโ regulatingโ theโ fixingโ ofโ theโ prepaymentโpenalty,โ perhapsโ causingโ banksโ toโ includeโ anโ excessโ riskโ premiumโ inโ borrowingโ rates.โ Butโ theseโ opposingโeffectsโ areโ notโ directlyโ quantifiableโ onโ theโ basisโ ofโ theโ dataโ availableโ toโ theโ Nationalโ Bank.โ Theโ chosenโapproachโthereforeโconsistedโinโexaminingโdevelopmentsโinโbusinessโcreditโandโinterestโratesโandโidentifyingโdivergencesโcomparedโtoโwhatโmightโnormallyโbeโexpectedโinโviewโofโtheโeconomicโenvironmentโandโbanksโโfundingโcosts.
Thisโ meantโ conductingโ anโ econometricโ analysisโ enablingโ usโ toโ separateโ anyโ structuralโ changesโ afterโ theโbeginningโof 2014โโโwhenโtheโlawโcameโintoโforceโโโfromโcyclicalโeffectsโmeasuredโonโtheโbasisโofโtheโBankโsโoverallโ syntheticโ indicatorโ andโ changesโ connectedโ withโ theโ movementโ inโ fundingโ costsโ onโ theโ interbankโmarkets,โexaminedโbyโmeansโofโtheโswapโrates.โDivergencesโspecificโtoโtheโperiodโbeginningโ in 2014โwereโassessedโwithโtheโaidโofโaโbinaryโvariableโintroducedโintoโeachโmodel.โTheโvariousโequationsโtakeโtheโformโofโautoregressiveโdistributedโlagโmodels.โHowever,โtheโonesโrelatingโtoโinterestโratesโwereโconvertedโintoโerrorโcorrectionโmodelsโinโorderโtoโtakeโaccountโofโanyโlong-termโlinkโbetweenโtheโinterestโratesโchargedโbyโbanksโonโloansโtoโbusinessesโandโtheโcostโofโfundsโassessedโaccordingโtoโtheโswapโrates.
Theโ specificationsโ ofโ theโ variousโmodelsโ areโ shownโ inโ anโ annexโ (Annexโ 3),โ asโ areโ theโ detailedโ resultsโ ofโtheโestimationsโ(Annexesโ4 andโ5).โTheโmodelโconcerningโ(year-on-year)โgrowthโofโlendingโtoโnon-financialโcorporationsโisโdeclinedโinโfourโversionsโ:โoneโforโallโlendingโandโthreeโothersโforโtheโvariousโsizeโcategories,โnamelyโmicroโandโsmallโfirms,โgroupedโ inโaโsingleโcategory,โandโmedium-sizedโandโ largeโfirms.โTheโmodelโconcerningโinterestโratesโwasโestimatedโforโtheโfourโweightedโaverageโratesโrepresentedโinโchart 3.
Theโ resultsโ obtainedโ (chart 8)โ doโnotโ showโanyโ significantโ structuralโ changeโ inโ theโ creditโ situationโ thatโcouldโpotentiallyโbeโlinkedโtoโtheโentryโintoโforceโofโtheโlaw.โInโgeneral,โtheโoutstandingโamountโofโloansโtoโbusinessesโfollowsโaโrelativelyโstableโtrendโalthoughโitโisโactuallyโinfluencedโbyโcyclicalโdevelopmentsโtoโsomeโextent,โandโafterโaโ timeโ lag.โAsโalreadyโ shownโbyโ theโdescriptiveโ statisticsโ inโ theโfirstโ section,โ theโcreditโ usedโ byโmicro-โ andโ smallโ firmsโ isโ lessโ sensitiveโ toโ theโ economicโ environmentโ thanโ loansโ grantedโtoโ largerโfirms.โMoreover,โbankโ loansโtoโ largeโfirmsโseemโtoโbeโtheโonlyโonesโ influencedโbyโ interestโrateโchanges,โbutโtheโ impactโ isโnotโveryโsignificant.โPartโofโtheโreasonโmayโbeโthat,โunlikeโmostโSMEs,โsomeโlargeโfirmsโareโableโtoโaccessโmarketโfinancingโwhenโtheโ interestโ ratesโchargedโbyโcreditโ institutionsโareโdeemedโtooโhigh.
11NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Chart 8
Average gaps between loans and interest rates compared to their โexpectedโ trend at the beginning of 2014(effectsโspecificโtoโtheโperiod 2014-2020,โestimatedโbyโmeansโofโeconometricโmodelsโandโ95โ%โconfidenceโintervals)
โ6
โ4
โ2
0
2
4
6
8
10
12
โ20
โ15
โ10
โ5
0
5
10
Totalcredit
Smallfirms
Medium-sizedfirms
Largefirms
Varia
ble
rate
s <1
yea
r,am
ount
s <โฌ
1 m
illio
n
Varia
ble
rate
s <1
yea
r,am
ount
s >โฌ
1 m
illio
n
Fixe
d ra
tes
>1 y
ear
and
<5 y
ears
, am
ount
s <โฌ
1 m
illio
n
Fixe
d ra
tes
>5 y
ears
, a
mou
nts
<โฌ 1
mill
ion
Estimated gaps for annual growth of credit used(percentage points)
Estimated gaps for interest rates(basis points)
Sourceโ:โNBB.Noteโ:โโTheโdashesโinโtheseโchartsโrepresentโtheโeffectsโspecificโtoโtheโperiodโfromโtheโbeginningโof 2014โtoโtheโfinalโquarterโof 2019โ(forโ
loans)โorโtoโFebruary 2020โ(forโinterestโrates).โTheseโeffectsโareโinterpretedโasโdeviationsโfromโwhatโtheโmodelโpredictsโsolelyโonโtheโbasisโofโtheโfundamentalโfactorsโtakenโintoโaccountโinโtheโmodels.โTheโverticalโlinesโrepresentโtheโconfidenceโintervals.โIfโtheโvalueโโ0โโonโtheโy-axisโisโincludedโinโtheโinterval,โtheโassociatedโeffectโisโconsideredโinsignificantโatโaโ5โ%โlevel.
Accordingโ toโ theโestimatesโobtainedโ forโmodelsโ relatingโ toโ theโ interestโ ratesโappliedโbyโbanks,โ thoseโ ratesโareโdeterminedโmainlyโbyโtheโcostโofโfunds.โTheโresultsโinโfactโconfirmโthatโinโmostโcasesโtheseโinterestโratesโmirrorโ theโ trendโ inโ theโ keyโ interestโ ratesโ andโmoneyโmarketโ rates,โ exceptโ forโmedium-termโ interestโ rates.โMoreover,โtheโeconomicโclimateโdoesโnotโappearโtoโhaveโaโsignificantโeffectโonโmostโinterestโrates,โatโ leastโnotโduringโtheโperiodโcoveredโbyโtheโdata.โHowever,โtheyโdidโdeviateโfromโtheโtrendโinโmoneyโmarketโratesโafterโeruptionโofโtheโcrisisโinโtheโsecondโhalfโof 2008.
Whileโborrowingโratesโcontinuedโtoโfallโbetweenโtheโendโof 2014โandโtheโendโofโFebruary 2020,โthereโ isโnoโsignificantlyโnegativeโeffectโexceptโ inโ theโcaseโofโ interestโ ratesโatโmoreโ thanโoneโyear.โThatโchangeโprobablyโreflectsโaโreductionโinโintermediationโmarginsโdueโtoโincreasedโpressureโfromโcompetition,โasโwasโalsoโrevealedโbyโtheโBankโLendingโSurveyโ(seeโabove),โratherโthanโbeingโdueโtoโtheโLawโofโ21 December 2013.
Finally,โanโanalysisโofโtheโgapsโbetweenโtheโvaluesโpredictedโbyโtheโvariousโmodelsโandโtheโobservedโvaluesโ(notโreportedโhere)โalsoโfailedโtoโshowโanyโtemporaryโeffects,โasโtheโerrorโtermsโofโtheโmodelsโrelatingโtoโcreditโandโinterestโratesโforโtheโperiodโcommencingโin 2014โremainedโwithinโtheโusualโmargins.
12NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Conclusions
ThisโarticleโreviewedโtheโdynamicsโofโtheโvolumeโofโloansโgrantedโbyโbankingโinstitutionsโoperatingโinโBelgiumโtoโresidentโenterprises,โandโtheโinterestโratesโandโotherโcreditโconditions,โsinceโtheโentryโintoโforceโofโtheโLawโofโ21 December 2013 onโSMEโfinancing.
LoansโtoโSMEsโโโgenerallyโlessโprocyclicalโthanโtheโvolumeโofโcreditโusedโbyโlargeโfirmsโโโincreasedโbetween 2014โandโ theโ beginningโ of 2020.โ Duringโ thatโ period,โ theโ growthโ ofโ lendingโ toโmicro-companiesโ andโ smallโ firmsโaveragedโ4.6โ%โyear-on-year,โcomparedโtoโ0.6โ%โforโmedium-sizedโfirms.โThisโsituationโisโdueโpartlyโtoโtheโcreditโsupply,โwhichโexpandedโthroughoutโtheโperiod,โsupportedโbyโtheโvariousโmeasuresโtakenโbyโtheโEurosystemโtoโstimulateโeconomicโactivity.โUpโtoโtheโendโof 2018,โtheโexpansionโofโcreditโwasโalsoโpropelledโbyโrisingโdemandโfromโbothโSMEsโandโlargeโfirms.
Interestโratesโonโbusinessโloansโareโdeterminedโmainlyโbyโinterbankโmarketโrates,โwhichโareโthemselvesโstronglyโinfluencedโbyโmonetaryโpolicy.โForโsomeโyearsโnow,โtheโEurosystemโsโmonetaryโpolicyโhasโindirectlyโdoneโmuchโtoโ facilitateโ accessโ toโ bankโ financeโ forโ businessesโ (andโ households).โ Theโmeasuresโ implementedโ since 2014โhaveโcutโtheโcostโofโfundsโforโcreditโinstitutions.โTheโlatterโalsoโhaveโaccessโtoโloansโatโadvantageousโratesโviaโtheโtargetedโlonger-termโrefinancingโoperations.โFurthermore,โtheโassetโpurchaseโprogrammeโmadeโitโeasierโforโthemโtoโfreeโupโliquidityโwhichโcouldโbeโallocatedโtoโnewโloans.โFinally,โasโaโresultโofโcompetition,โtheโreductionโinโ banksโโ fundingโ costsโ ledโ toโ firmsโ beingโ offeredโ lowerโmedium-termโ andโ long-termโ borrowingโ rates,โ andโcausedโshort-termโratesโtoโstabiliseโatโaโlowโlevel.
Accordingโtoโvariousโsurveys,โfirmsโhaveโinโfactโreportedโthatโtheirโcreditโconditionsโhaveโimprovedโsince 2014,โandโ wereโ particularlyโ favourableโ between 2016โ and 2019.โ However,โ owingโ toโ theโ increasedโ risks,โ theseโconditionsโwereโtightenedโslightlyโatโtheโendโofโtheโperiod,โandโthatโaffectedโallโcategoriesโofโfirmsโregardlessโofโsize.
13NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Annex 1
Outstanding total of credit used by size and branch of activity(in โฌ billion, endโofโyear data)
2002 2008 2013 2016 2019
All firms 1, 2 68.4 98.4 109.7 117.0 138.9
Small firms (including micro companies) 1 29.3 45.8 54.1 59.2 70.7
of which :
Manufacturing industry 2.9 3.5 3.4 3.5 4.4
Electricity, gas, steam and air conditioning supply 0.0 0.1 0.2 0.4 0.6
Water supply ; sewerage, waste management and remediation activities 0.1 0.2 0.2 0.2 0.3
Construction 3.1 5.4 6.5 7.3 9.0
Wholesale and retail trade ; repair of motor vehicles and motor cycles 7.7 10.3 10.8 11.1 12.1
Transportation and storage 1.3 1.6 1.5 1.5 2.0
Accommodation and food service activities 1.4 1.8 2.2 2.3 2.7
Information and communication 0.5 0.8 1.1 1.2 1.4
Real estate activities 5.5 9.3 11.5 13.8 16.6
Mediumโsized firms 1 13.1 18.4 24.4 22.2 24.9
of which :
Manufacturing industry 3.5 3.1 2.8 2.8 3.1
Electricity, gas, steam and air conditioning supply 0.1 0.3 0.7 0.5 0.8
Water supply ; sewerage, waste management and remediation activities 0.5 0.5 0.4 0.3 0.3
Construction 1.2 1.9 3.1 2.5 2.9
Wholesale and retail trade ; repair of motor vehicles and motor cycles 3.4 3.6 4.5 4.4 4.3
Transportation and storage 1.0 1.3 1.4 1.3 1.5
Accommodation and food service activities 0.2 0.2 0.3 0.4 0.3
Information and communication 0.3 0.2 0.3 0.3 0.4
Real estate activities 1.3 2.7 4.1 3.9 4.6
Large firms 1 19.8 28.1 24.2 27.3 31.5
of which :
Manufacturing industry 7.4 5.1 4.0 4.3 6.0
Electricity, gas, steam and air conditioning supply 2.1 3.9 5.2 4.0 4.6
Water supply ; sewerage, waste management and remediation activities 1.0 1.9 2.4 2.2 2.2
Construction 0.8 1.1 1.3 1.3 1.5
Wholesale and retail trade ; repair of motor vehicles and motor cycles 4.8 6.3 5.7 5.9 5.1
Transportation and storage 1.0 1.3 1.6 2.3 2.0
Accommodation and food service activities 0.1 0.1 0.1 0.1 0.1
Information and communication 0.8 0.4 0.3 0.4 0.7
Real estate activities 0.3 0.2 0.2 0.5 0.5
Source : NBB (Central Corporate Credit Register).1 No account was taken of credit used by firms classified under financial services.2 The sum of the loans to small, medium and large firms is less than the total credit recorded because some of the loans are made to firms
for which no size information is available (because they have not yet filed a balance sheet or are not required to do so).
14NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Annex 2
Annualised growth rate 1 of used credit, by size and branch of activity(in %)
2002โ2007 2008โ2013 2014โ2020 3 2002โ2020 3
All firms 2 4.0 3.1 4.4 4.1
Small firms (including micro companies) 2 6.4 4.8 4.6 5.0
of which :
Manufacturing industry 1.9 0.6 3.8 2.8
Electricity, gas, steam and air conditioning supply 18.3 37.5 16.1 20.7
Water supply ; sewerage, waste management and remediation activities 2.7 6.9 5.1 5.0
Construction 7.6 6.1 5.6 6.1
Wholesale and retail trade ; repair of motor vehicles and motor cycles 3.9 2.3 1.7 2.3
Transportation and storage 1.8 1.3 5.1 3.7
Accommodation and food service activities 5.0 3.9 3.7 4.0
Information and communication 7.9 7.6 4.9 6.0
Real estate activities 8.4 6.0 6.3 6.7
Mediumโsized firms 2 1.1 7.6 0.6 2.1
of which :
Manufacturing industry โ4.9 1.1 0.8 โ0.2
Electricity, gas, steam and air conditioning supply 21.7 57.0 1.9 16.7
Water supply ; sewerage, waste management and remediation activities 4.4 0.7 โ5.5 โ2.3
Construction 1.4 8.9 0.7 2.4
Wholesale and retail trade ; repair of motor vehicles and motor cycles โ1.1 5.7 โ0.1 0.8
Transportation and storage 4.6 10.3 โ0.3 2.8
Accommodation and food service activities 1.3 9.5 2.9 3.9
Information and communication 3.8 6.0 3.6 4.1
Real estate activities 8.5 9.4 2.7 5.2
Large firms 2 0.1 1.3 4.2 2.8
of which :
Manufacturing industry โ6.5 โ4.4 8.3 2.9
Electricity, gas, steam and air conditioning supply 41.6 14.1 โ0.8 10.5
Water supply ; sewerage, waste management and remediation activities 8.7 9.1 0.0 3.5
Construction 4.7 8.9 5.7 6.1
Wholesale and retail trade ; repair of motor vehicles and motor cycles 0.2 2.6 โ1.7 โ0.5
Transportation and storage 3.1 11.2 5.5 6.2
Accommodation and food service activities 22.9 6.1 17.6 16.4
Information and communication 2.2 โ2.7 15.1 9.1
Real estate activities โ7.8 79.5 31.9 33.5
Source : NBB (Central Corporate Credit Register).1 The annualised growth rate is determined as follows : a 12โmonth moving average is calculated on the basis of the quarterly growth rates
(Encoursโtโ/โEncoursโtโโโ1โโโ1), disregarding the quarters affected by a break in the series (namely the second quarter of 2012, and the fourth quarter of 2014). This average is then annualised.
2 No account was taken of credit used by firms classified under financial services.3 Data taken into account up to February 2020.
15NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Annex 3
Description of the econometric models and detailed results
Theโautoregressiveโdistributedโlagโmodelsโdescribingโtheโrelationshipโbetweenโcreditโgrowthโonโtheโoneโhand,โandโbankโinterestโratesโandโtheโbusinessโcycleโonโtheโother,โareโdefinedโbyโtheโfollowingโequationโ:
whereโ
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
representsโtheโgrowthโofโusedโcreditโcomparedโtoโtheโcorrespondingโquarterโofโtheโpreviousโyearโforโaโcategoryโofโfirmsโ(microโ/โsmall,โmediumโorโlarge),โ
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
โisโtheโlong-termโinterestโrateโ(correspondingโtoโloansโofโlessโthanโโฌโ1 millionโwithโtheโrateโinitiallyโfixedโforโmoreโthanโfiveโyears)โandโ
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
โisโtheโstateโofโeconomicโactivityโmeasuredโ accordingโ toโ theโquarterlyโ averageโofโ theโBankโsโ overallโ syntheticโ indicator.โ
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
isโ aโ binaryโ variableโ intendedโ toโ captureโ structuralโ changesโ inโ thatโ relationshipโ which,โ ifโ negative,โ couldโ beโattributedโtoโtheโcrisisโwhichโbeganโinโtheโthirdโquarterโof 2008.โTheโsecondโbinaryโvariable,โ
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
โ,โisโusedโtoโassessโanyโstructuralโchangesโspecificโtoโtheโperiodโfollowingโtheโentryโintoโforceโofโtheโlawโonโSMEโfinancing.โTheโlastโtwoโbinaryโvariables,โ
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
โandโ
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
โhaveโnoโeconomicโinterpretation.โTheyโ serveโonlyโ toโabsorbโ statisticalโbreaksโdueโ toโmethodologicalโ changesโ inโ theโcompilationโofโ theโCentralโCreditโRegisterโsโstatistics.
Inโessence,โtheโmodelsโusedโtoโidentifyโtheโfactorsโdeterminingโchangesโinโinterestโratesโareโalsoโautoregressiveโdistributedโlagโmodelsโ:
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
Whereโ
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
nowโsymbolisesโoneโofโtheโfourโinterestโratesโillustratedโinโchart 3,โandโ
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
โstandsโforโtheโswapโinterestโratesโ usedโ asโ theโ referenceโ forโ eachโ ofโ themโ inโ thatโ sameโ chart.โ Theโ otherโ variablesโ areโ definedโ inโ theโsameโwayโasโinโtheโmodelsโrelatingโtoโusedโcredit.โUnlikeโthoseโmodels,โwhichโareโbasedโonโquarterlyโdata,โtheโmodelโrelatingโtoโinterestโratesโisโestimatedโonโtheโbasisโofโmonthlyโdata.โHowever,โitโisโnotโestimatedโasโsuchโ;โitโwasโreformulatedโasโaโgeneralisedโerrorโcorrectionโmodelโ inโorderโtoโtakeโaccountโofโtheโpossibleโexistenceโofโaโtrendโlinkโbetweenโbankโinterestโratesโonโloansโtoโbusinessesโandโswapโinterestโratesโ:
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
Whereโ
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
โandโ
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
.
Theโestimatesโobtainedโforโtheseโtwoโtypesโofโmodelsโonโtheโbasisโofโtheโordinaryโleastโsquaresโmethodโareโsetโoutโinโtheโfollowingโtables.
17
Annex 3 - Description of the econometric models and detailed results
The autoregressive distributed lag models describing the relationship between credit growth on the one hand, and bank interest rates and the business cycle on the other, are defined by the following equation:
ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ๐๐๐๐๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
3
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4 + ๐ฟ๐ฟ๐ฟ๐ฟ3๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1
+ ๐ฟ๐ฟ๐ฟ๐ฟ4๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐3
where ฮ4๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก represents the growth of used credit compared to the corresponding quarter of the previous year for a category of firms (micro/small, medium or large), ๐๐๐๐๐ก๐ก๐ก๐ก โ1 is the long-term interest rate (corresponding to loans of less than โฌ 1 million with the rate initially fixed for more than five years) and ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐ก is the state of economic activity measured according to the quarterly average of the Bankโs overall synthetic indicator. ๐ท๐ท๐ท๐ท2008๐๐๐๐3โ2019๐๐๐๐4, is a binary variable intended to capture structural changes in that relationship which, if negative, could be attributed to the crisis which began in the third quarter of 2008. The second binary variable, ๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2019๐๐๐๐4, is used to assess any structural changes specific to the period following the entry into force of the law on SME financing. The last two binary variables, ๐ท๐ท๐ท๐ท2012๐๐๐๐2โ2013๐๐๐๐1 and ๐ท๐ท๐ท๐ท2014๐๐๐๐4โ2015๐๐๐๐4 have no economic interpretation. They serve only to absorb statistical breaks due to methodological changes in the compilation of the Central Credit Registerโs statistics.
In essence, the models used to identify the factors determining changes in interest rates are also autoregressive distributed lag models:
๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐i๐ก๐ก๐ก๐กโ1 + ๐ฝ๐ฝ๐ฝ๐ฝ0๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐ฝ๐ฝ๐ฝ๐ฝ1๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 + ๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2+ ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐๐ก๐ก๐ก๐ก now symbolises one of the four interest rates illustrated in chart 3, and ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก stands for the swap interest rates used as the reference for each of them in that same chart. The other variables are defined in the same way as in the models relating to used credit. Unlike those models, which are based on quarterly data, the model relating to interest rates is estimated on the basis of monthly data. However, it is not estimated as such; it was reformulated as a generalised error correction model in order to take account of the possible existence of a trend link between bank interest rates on loans to businesses and swap interest rates:
ฮ๐๐๐๐๐ก๐ก๐ก๐ก = ๐๐๐๐ + ๐๐๐๐โ(i๐ก๐ก๐ก๐กโ1 โ ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1) + ๐ฝ๐ฝ๐ฝ๐ฝ0ฮ๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐ก + ๐๐๐๐๐ ๐ ๐ ๐ ๐ก๐ก๐ก๐กโ1 +๏ฟฝ๐พ๐พ๐พ๐พ๐๐๐๐๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ถ๐ก๐ก๐ก๐กโ๐๐๐๐
9
๐๐๐๐=0
+ ๐ฟ๐ฟ๐ฟ๐ฟ1๐ท๐ท๐ท๐ท2008๐๐๐๐6โ2020๐๐๐๐2 + ๐ฟ๐ฟ๐ฟ๐ฟ2๐ท๐ท๐ท๐ท2014๐๐๐๐1โ2020๐๐๐๐2
Where ๐๐๐๐โ = (๐๐๐๐ โ 1) and ๐๐๐๐ = (๐๐๐๐+ ๐ฝ๐ฝ๐ฝ๐ฝ0 + ๐ฝ๐ฝ๐ฝ๐ฝ1 โ 1).
The estimates obtained for these two types of models on the basis of the ordinary least squares method are set out in the following tables.
16NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Annex 4
Estimated parameters for equations relating to the annual growth of credit used by firms(ordinary least squares estimates over a period from the second quarter of 2003 to the final quarter of 2019)
Dependent variable : ฮ4Credt All firms Microโ and small firms
Mediumโsized firms
Large firms
ฮผ โ0.713 3.228 โ7.243 โ7.837
ฮ4Credtโโโ1 0.619*** 0.761*** 0.560*** 0.422***
itโโโ1 0.772 โ0.240 2.112 2.445*
Conjt 0.047 0.029 0.270 โ0.018
Conjtโโโ1 0.065 โ0.031 โ0.080 0.263
Conjtโโโ2 โ0.016 0.045 0.005 โ0.133
Conjtโโโ3 0.161* 0.038 0.215 0.475**
D2008T3โโโ2019T4 0.696 โ0.540 4.017* 2.119
D2014T1โโโ2019T4 0.886 โ0.605 1.037 4.645
D2012T2โโโ2013T1 0.413 โ1.033 3.603 1.254
D2014T4โโโ2015T3 โ1.081 โ0.818 โ4.788** 4.217*
Sum of longโterm elasticities of cyclical effects
1
!โ ๏ฟฝ! (1 โ ๏ฟฝ)โ"!#$ *
0.673*** 0.341* 0.931*** 1.015***
Rยฒ 0.810 0.825 0.714 0.753
Standard error of the regression 1.915 1.427 3.967 3.934
Observations 67 67 67 67
Sourceโ:โโNBBโ(CentralโCorporateโCreditโRegister).*โ โ โโCoefficientโsignificantโatโaโlevelโofโ10โ%.**โ โโCoefficientโsignificantโatโaโlevelโofโ5โ%.***โโCoefficientโsignificantโatโaโlevelโofโ1โ%.
17NBB Economic Review ยก June 2020 ยกโ BankโfinancingโforโSMEsโfromโ2014โtoโ2019:โeffectโofโchangesโinโtheโlawโonโlending
Annex 5
Estimated parameters for equations relating to interest rates on loans to nonโfinancial corporations(ordinary least squares estimates over a period from February 2003 to February 2020)
Dependent variable : ฮit Variable interest rates initially fixed
for less than 1 year
(amount โค โฌ 1 million)
Variable interest rates initially fixed
for less than 1 year
(amount > โฌ 1 million)
Rate initially fixed for more than 1 year but less than 5 years
(amount โค โฌ 1 million)
Rate initially fixed for more than 5 years
(amount โค โฌ 1 million)
ฮผ 0.446*** 0.158*** 0.109 0.214***
itโโโ1 โ stโโโ1 โ0.258*** โ0.151*** โ0.027 โ0.104***
ฮst 0.184*** 0.245*** 0.237*** 0.137***
stโโโ1 โ0.019** โ0.015 โ0.017 โ0.023**
Conjt โ0.003 โ0.001 0.000 โ0.001
Conjtโโโ1 0.007 0.009 0.011* 0.004
Conjtโโโ2 0.000 0.001 0.003 0.000
Conjtโโโ3 โ0.001 0.002 โ0.003 0.002
Conjtโโโ4 โ0.007 โ0.010* โ0.005 โ0.006*
Conjtโโโ5 0.002 0.004 โ0.007 0.001
Conjtโโโ6 0.002 0.000 0.003 โ0.003
Conjtโโโ7 0.000 โ0.002 0.007 0.002
Conjtโโโ8 0.004 0.006 โ0.003 0.002
Conjtโโโ9 โ0.002 โ0.002 0.001 0.000
D2008M7โโโ2020M2 0.063*** 0.086*** 0.022 0.085***
D2014M1โโโ2020M2 0.005 0.031 โ0.069** โ0.109***
Sum of longโterm elasticities of cyclical effects
!โ ๏ฟฝ! (1 โ ๏ฟฝ)โ%!#$ * 0.008 0.041*** 0.304 0.009
Rยฒ 0.608 0.516 0.382 0.433
Standard error of the regression 0.079 0.097 0.110 0.063
Observations 205 205 182 174
Source : NBB (Central Corporate Credit Register).* Coefficient significant at a level of 10 %.** Coefficient significant at a level of 5 %.*** Coefficient significant at a level of 1 %.