WorkingPaperNo.1019

AnInquiryConcerningJapaneseYenInterestRateSwapYields

TanweerAkram

SeniorVicePresident/SeniorEconomistatCitibank

and

KhawajaMamun

AssociateProfessoratSacredHeartUniversity(SHU)

May2023

TanweerAkramAddress:Citibank,6400LasColinasBlvd.,Irving,TX75039,USA.Phone:+1(703)4058656.Email:tanweer.akram@.ORCID:/0000-0002-6646-0902;KhawajaMamunAddress:SacredHeartUniversity,JackWelchCollegeofBusiness&Technology,Fairfield,CT06825,USA.Phone:+1(203)676-5147.Email:mamunk@.ORCID:/0000-0001-9669-2897

Theauthorsthankparticipantsatvariousworkshopsfortheirvaluablecommentsandsuggestions.Theauthors’institutionalaffiliationsareprovidedsolelyforidentificationpurposes.Viewsexpressedaremerelythoseoftheauthors.Thestandarddisclaimerholds.Thisresearchdidnotreceiveanyspecificgrantfromfundingagenciesinthepublic,commercialornot-for-profitsectors.Thedatasetusedintheempiricalpartofthispaperisavailableuponrequesttobonafideresearchersforthereplicationandverificationoftheresults

TheLevyEconomicsInstituteWorkingPaperCollectionpresentsresearchinprogressbyLevyInstitutescholarsandconferenceparticipants.Thepurposeoftheseriesistodisseminateideastoandelicitcommentsfromacademicsandprofessionals.

LevyEconomicsInstituteofBardCollege,foundedin1986,isanonprofit,nonpartisan,independentlyfundedresearchorganizationdevotedtopublicservice.Throughscholarshipandeconomicresearch,itgeneratesviable,effectivepublicpolicyresponsestoimportanteconomicproblemsthatprofoundlyaffectthequalityoflifeintheUnitedStatesandabroad.

LevyEconomicsInstitute

P.O.Box5000

Annandale-on-Hudson,NY12504-5000

Copyright?LevyEconomicsInstitute2023Allrightsreserved

ISSN1547-366X

1

ABSTRACT

ThispapereconometricallymodelsJapaneseyen(JPY)–denominatedinterestrateswapyields.Itexamineswhethertheshort-terminterestrateexertsaninfluenceonthelong-termJPYswapyieldaftercontrollingforseveralkeymacroeconomicvariables,suchascoreinflation,thegrowthofindustrialproduction,thepercentagechangeintheequitypriceindex,andthepercentagechangeintheexchangerate.ItalsotestswhethertherearestructuralbreaksinthedynamicsofJapaneseswapyieldsandrelatedvariables.Theestimatedeconometricmodelsshowthattheshort-terminterestrateexertsanimportantinfluenceonthelong-termswapyieldinsomeperiodsbutnotinotherperiodsinwhichcoreinflationexertsamarkedinfluenceontheswapyield.ThefindingsfromtheeconometricmodelsrevealadiscernablerelationshipbetweenthecallrateandtheswapyieldofdifferentmaturitytenorsclearlyheldpriortoApril2014butdidnotinthesubsequentperiod.ThesefindingshighlightthelimitsandscopeofJohnMaynardKeynes’scontentionthatthecentralbank’spolicyratecommandsadecisiveinfluenceoverthelong-termmarketratethroughtheshort-terminterestrate.Thepolicyimplicationsoftheestimatedmodels’resultsarediscussed.

KEYWORDS:InterestRateSwaps;SwapYields;CallRate;Inflation;BankofJapan(BOJ);Japan

JELCLASSIFICATIONS:E43;E50;E58;E60;G10;G12

2

SECTIONI:INTRODUCTION

Japaneseyen(JPY)–denominatedinterestrateswapsareanimportantcomponentoftheglobalswapsmarket.JPYswapsplayavitalroleintheglobalfinancialsystemandJapanesefinancialmarkets.Asof2021,thenotionalvalueofJPYinterestrateswapsamountedtonearly$35trillion,whiletheirgrossmarketvalueamountedto$265billion,accordingtotheBankforInternationalSettlements(BIS)(2022).

ThispapereconometricallymodelsJPYinterestrateswapyieldsusingmacroeconomicandfinancialvariables.Itexamineswhetherthecurrentshort-terminterestrateexertsadecisiveinfluenceonthelong-termswapyieldaftercontrollingforseveralkeymacroeconomicvariables,suchascoreinflation,thegrowthofindustrialproduction,thepercentagechangeintheequitypriceindex,andthepercentagechangeintheexchangerate.ItalsotestswhethertherearestructuralbreaksinthedynamicsofJPYswapyieldsandrelatedvariablesthatinfluencethebehaviorofJPYswapyields.

Figure1laysouttheevolutionofoutstandingJPYinterestrateswapsbothintermsoftheirnotionalamountandgrossmarketvalue.JPY-denominatedinterestrateswapsconstitutemorethan95percentofallJPYinterestratederivativesasofthefirsthalfof2021,accordingtotheBIS(2022).(ThesourceforthedatainFigure1istheBIS,whilethesourcesofthedatainremainingfiguresinthispaperarelistedinTable1.)

3

Figure1:TheEvolutionofJPYInterestRateSwaps,2000–21

WhileJPYinterestrateswapsarecrucialfinancialinstrumentsintheuniverseofJapanesefinancialassets(aswellasinthefunctionandoperationoftheJPY-denominatedfinancialandbankingsystem),thereisapaucityofempiricalanalysesofJPYswapyields.ThispaperfillsalacunaintherelatedliteratureoninterestrateswapsbyexploringthemacroeconomicdeterminantsofJPYswapyields.JohnMaynardKeynes(1930)positedthatthecentralbank’smonetarypolicydecisions—particularlyitssettingofthepolicyrate—influencetheyieldsoflong-termgovernmentbondsthroughtheireffectsontheshort-terminterestrate.Recentresearchhasvindicatedthehypothesisthatthecurrentshort-terminterestrateisakeydriverofJapanesegovernmentbond(JGB)yields(AkramandLi2020a,b).ThisandrelatedfindingsholdnotjustforJapan,butalsoforotheradvancedcountries,suchastheUnitedStates(Akram2021a,AkramandLi2020c).Takingacuefromthisliterature,thispaperexamineswhetherthecurrentshort-terminterestratehasadecisiveeffectonswapyieldsofdifferentmaturitytenorsinJapan,aftercontrollingforkeymacroeconomicandfinancialvariables.Thisisanimportanttheoreticalandpolicyquestionbecauseithasconsequentialimplicationsfortheefficacyofmonetarypolicyandthemonetarytransmissionmechanism,financialmarkets,financialintermediation,thefinancialservicesindustry,corporatefinance,fiscalpolicy,andfiscal-monetarypolicycoordination.

Thispaperisarrangedasfollows.SectionIIundertakesabriefoverviewoftheliteratureoninterestrateswapsandtheliteratureontheempiricsofinterestratedynamicsfromtheKeynesianperspective.SectionIIIprovidesasynopsisoftheevolutionofJPYswapyieldswith

4

referencetoJapan’smacroeconomicandfinancialdevelopmentsduringthepasttwodecades.SectionIVdescribesthedatausedintheeconometricmodeling;italsoundertakesunitrootandstationaritytests.SectionVeconometricallymodelsthedynamicsoftheJPYswapyieldbasedonkeymacroeconomicandfinancialvariables.SectionVIdeliberatesthepolicyimplicationsoftheempiricalfindings.SectionVIIsummarizesandconcludes.

SECTIONII:ABRIEFOVERVIEWOFTHELITERATURE

Theliteratureoninterestrateswapsisvast.Corb(2012)providesadetailedprimeroninterestrateswaps,coveringtheirfunctions,pricing,applications,andrecentinnovations.BickslerandChen(1986),KimandKoppenhaver(1993),SmithJr.,Smithson,andWakeman(1988),andVisvanathan(1998)renderadditionalanalysisoftheapplicationsofswapsinbusinessandfinance,thefunctionoftheswapmarket,andthevarioustypesofswapusers.BIS(2022)givesdetailedstatisticsaboutswapsdenominatedinmajorcurrencies,includinginterestrateswaps,providingtime-seriesinformationonthenotionalandgrossmarketvaluesofoutstandingswaps.

Theempiricalliteraturemodelingswapyieldsinquantitativefinancehasfallenshortincrucialaspects.DuffieandHunag(1996),DuffieandSingleton(1997),andLekkosandMilas(2001)areamongthemostnotableempiricalstudiesofswapsandswapspreads.However,theseandmostotherempiricalstudiesofswapyieldsinquantitativefinancehavebeendeficientintwocrucialaspects.First,themodelersoftendonotrelateswapyieldstofundamentalmacroeconomicvariables.Second,themodelersdonotassesswhetherKeynes’sconjecture,whichconnectsthelong-terminterestratetothecurrentshort-terminterestrate,isapplicableforswapyields.

Keynes’sconjectureoninterestratedynamicshasaplausibletheoreticalandbehavioralbasis(Keynes1930,[1936]2007)andhasfoundsupportinthedata.TherecentempiricalliteraturehasgivencredencetoKeynes’sconjectureregardingtheconnectionbetweentheshort-terminterestrateandthelong-terminterestratebasedonKeynes’sowntheoreticalperspectiveandRiefler’s(1930)trailblazingstatisticalanalysis.Agoodnumberofempiricalstudies,suchas

5

AkramandLi(2020a,b,c),Atesogulu(2003–4,2005),Chakraborty(2016),DeleidiandLevrero(2020),Gabrisch(2021),Kim(2021),LiandSu(2021),Payne(2006–7),Simoski(2019),andVinod,Chakraborty,andKarun(2014),havecoveredbothadvancedeconomiesandemergingmarkets.Thesestudiesreportthatthereisstrongevidenceofstatisticallysignificantandeconomicallymeaningfulpassthroughfromtheshort-terminterestratetotherelevantlong-termmarketinterestrateaftercontrollingfortheappropriatemacroeconomicandfinancialfactors.Akram(2021b)hasformalizedKeynes’sconjecturelinkingthelong-terminterestratetotheshort-terminterestratesinaquantitativeframework.Mostresearchontheempiricalmodelingofinterestratedynamicshasbeenlimitedtotheexaminationoftherelationshipbetweentheshort-terminterestrateandthelong-termbenchmarkgovernmentbondyield.However,AkramandMamun(2022a,b)haverecentlyevincedthatKeynes’sconjectureisapplicableforunderstandingthedynamicsoftheyieldsoflong-termswapsdenominatedinUSdollars(USD)andBritishpounds(GBP).Inasimilarvein,thispaperexamineswhetherKeynes’sconjectureappliesforJapaneseyen(JPY)swaps.

SECTIONIII:THEEVOLUTIONOFJPYSWAPYIELDSINJAPAN’SMACROECONOMICCONTEXT

Figure2displaystheevolutionofJPYswapyieldsandthecallrateinJapan.Itshowsthattheswapyieldsandcallrateusuallymovetogetherovertime.Forinstance,asthecallraterosebetweenearly2006tolate2008,swapyieldsbegantorise,andasthecallratedeclinedbetweenearly2009andtolate2010,swapyieldsbegantodecline.Asthecallratedeclinedtonear-zerolevels,swapyieldsfellsharplyandstayedlowbetweenearly2017tolate2021.However,swapyields,particularlyinthefrontendoftheswapyieldcurve,roseinearly2022andcontinuedthroughouttheremainderoftheyear.

6

Figure2:TheEvolutionofInterestRateSwapYieldsandtheCallRateinJapan,2003–22

Source:SeeTable1

Figure2indicatesthatswapyieldsandthecallratehadstructuralbreaksduringthestudyperiod.Aroundmid-2000,swapyieldsroseandstayedhighuntilthelate2000s.However,theswapyieldsfelltonearzerointhe2010s.Inalatersection,econometrictestsareconductedtoidentifythebreakpointsduringthestudyperiod.

Figure3providesthecoevolutionofthe10-yearswapyieldandthecoreconsumerpriceindex(CPI)inflationbetween2003and2022.Itsuggeststhattheconnection,ifany,betweentheswapyieldandcoreinflationisweak.

7

Figure3:TheCoevolutionofthe10-yearSwapYieldandCoreCPIInflation,2003–22

Source:SeeTable1

Figure4depictsthegrowthofindustrialproductioninJapan.Industrialproductionwasgrowingbetween2003andmid-2008.Industrialproductionfellsharplyduringtheglobalfinancialcrisis.ItgrewastheJapaneseeconomyrecoveredfromtheglobalfinancialcrisis.ItdeclinedagainintheaftermathofJapan’searthquakeandtsunamiin2011.Followingabriefrecovery,industrialproductionfellinlate2012andearly2013.Between2014andmid-2019,industrialproductionmoderatedwithsomemonthsofgrowthandsomemonthsofcontraction.Bylate2019,theJapaneseeconomywasslowingdownandindustrialproductionwascontracting.Withtheonsetoftheglobalpandemicandlockdown,industrialproductionfellsharply.However,itstartedgrowingagaininearly2021astheglobalpandemicsubsidedandrestrictionswereeased.Inlate2021,industrialproductionfellagainwiththeonsetoftheomicronvariantofCOVID-19andtheslowdownineconomicactivity;itremainedweakuntiltheendofthestudyperiodin2022.

8

Figure4:TheGrowthofIndustrialProductioninJapan,2003–22

Source:SeeTable1

Figure5presentstheevolutionoftheJPY(￥)againsttheUSD($).Theexchangeratestoodataround￥120/$inearly2003.Itappreciatedto￥103/$inearly2005butitrevertedbackto￥120/$laterthesameyearandremainedstableuntilmid-2007.However,itgraduallyappreciatedfromlate2007tolate2012toaround￥75/$.WiththeadventofAbenomicsandquantitativeandqualitativemonetaryeasing(QQME),theJPYdepreciatedfromearly2013tolate2015.Bylate2015,theJPYhadappreciatedtoaround￥120/$.Inthefollowingmonths,theJPYappreciatedsteadilyuntilitreachedaround￥100/$.Thiswasfollowedbyadepreciationtoaround￥115/$inearly2017.Between2017andlate2021theJPYhoveredbetween￥100/$and￥115/$.However,in2022theJPYbegantodepreciatenoticeablyand,bythethirdquarterof2022,ithaddepreciatedtonearly￥150/$.Bytheendof2022,itappreciatedslightly,toabout￥135/$.

9

Figure5:TheExchangeRate(USDJPY),JapaneseYenperUSDollar,2003–22

Source:SeeTable1

Figure6displaystheevolutionoftheNikkei225stockpriceindexduringtheperiodcoveredinthispaper.TheNikkeiindexwasaround8,500atthestartoftheperiod.Itcontinuedtoriseuntilmid-2007toaround18,000.TheNikkeiindexdeclinedduringtheglobalfinancialcrisis,bottomingoutataround7,800inearly2009.Itremainedsteadyfromearly2009tolate2012,graduallyrisingfrom10,000inJanuary2014tonearly30,000inDecember2021.Butin2022,theNikkeiindexexperiencedsomecorrectionanddeclinedabit.

Figure6:TheStockPriceIndex,Nikkei225,2003–22

Source:SeeTable1

10

SECTIONIV:DATADESCRIPTIONANDUNITROOTANDSTATIONARITYTESTS

Table1,below,providesasummaryofthedatausedinthispaper.Thefirstcolumnliststhevariables.Thesecondcolumnprovidesadescriptionofthedataandthedaterangeofeachvariable.Thethirdcolumnshowsthefrequencyofthedataandwhetherhigh-frequencydatahavebeenconvertedtolow-frequencyequivalents.Thefinalcolumngivesthesourcesofthedata.

Thevariableusedfortheshort-terminterestrateisthecallmoneyrate.Forlong-termswapyields,swapsofthreedifferentmaturitytenorsarechosen:2-year,5-year,and10-yearinterestrateswaps.Coreinflationismeasuredastheyear-over-yearpercentagechangeintheCPI,excludingfood,non-alcoholicbeverages,andenergy.Economicactivityisgaugedbytheyear-over-yearpercentagechangeintheindexofindustrialproduction.ThevalueofthecurrencyisgivenbythespotexchangerateoftheJPYperUSDandthenominaleffectiveexchangerateoftheJPY.ThestockmarketdataarecoveredbytheNikkeiindexandtheTopixindex.Thenaturallogarithmoftheexchangerateandthenaturallogarithmofthestockpriceindexareused.Themonthlydatausedintheempiricalportionsofthepaper,whichcoversJanuary2003toDecember2022,havemorethan240observations.

11

Table1:SummaryoftheData

Variables

Datadescription,

daterange

Frequency

Sources

Short-terminterestrates

CALLRATE

Callmoneyrate,%,

September2002–December2022

Daily;

convertedto

monthly

AssociationofCall&

Discount

Companies/Nikkei

TIBOR

Tokyointerbankofferrate(TIBOR),threemonths,%,

September2002–December2022

Daily;

convertedto

monthly

Refinitiv

Long-termswapyields

JPSWAP2Y

Interestrateswap,2-year,%,September2002–December2022

Daily;

convertedto

monthly

Refinitiv

JPSWAP5Y

Interestrateswap,5-year,%,September2002–December2022

Daily;

convertedto

monthly

Refinitiv

JPSWAP10Y

Interestrateswaprate,10-year,%,September2002–December2022

Daily;

convertedto

monthly

Refinitiv

Coreinflation

CORECPI

Consumerpriceindex,allitems

excludingfood,non-alcoholic

beverages,andenergy,

%change,y/y,

September2002–December2022

Monthly

MinistryofInternal

Affairsand

Communications

Economicactivity

IPYOY

Industrialproduction,seasonally

adjusted,

%change,y/y,

September2002–December2022

Monthly

OrganizationforEconomicCooperationandDevelopment

Currency

USDJPY

Theexchangerate,spot,JapaneseyenperUSdollar,USDJPY,

September2002–December2022

Daily;convertedtomonthly

BankofJapan

NEER

Thenominaleffectiveexchangerate,September2022–December2022

Monthly

BankofJapan

Financialmarkets

NIKKEI

Stockpriceindex,Nikkei225,index,September2002–December2022

Daily;convertedtomonthly

TheFinancialTimes

TOPIX

StockpriceIndex,Topix,cashpriceindex,

September2002–December2022

Daily;

convertedto

monthly

TheFinancialTimes

12

ThesummarystatisticsofallvariablesintheirlevelsandfirstdifferencesareprovidedinTables2Aand2B,respectively.Themeanoftheswapyieldsincreaseswiththematuritytenors,ashighermaturityindicateshigherrisk.Theskewnessmeasuresindicatethatthe2-yearswaprate,callrate,andcoreCPIareright-skewed.Othervariablesarenotskewedineitherside.Allthetimeseriesareleptokurtic,indicatinganarrower,bell-shapeddistribution.Lastly,theJarque-BeratestsindicatenoneofthevariablesarenormallydistributedinTable2A.

Table2A:SummaryStatisticsoftheVariables

Vars

Obs.

Mean

Std.Dev.

Max.

Min.

Skewness

Kurtosis

J-B

Probability

JPSWAP2Y

244

0.27

0.33

1.28

-0.17

1.17

3.49

57.68

0.00

JPSWAP5Y

244

0.47

0.46

1.65

-0.19

0.80

2.74

26.73

0.00

JPSWAP10Y

244

0.86

0.62

2.14

-0.08

0.20

1.80

16.17

0.00

CALLRATE

244

0.20

0.24

1.08

-0.05

1.83

5.49

199.18

0.00

CORECPI

244

-0.14

0.80

2.30

-1.90

1.01

4.83

75.28

0.00

IPYOY

244

0.33

8.27

27.32

-33.33

-0.88

7.17

208.41

0.00

LNNIKKEI

244

9.61

0.38

10.31

8.95

0.05

1.81

14.39

0.00

LNTOPIX

244

7.15

0.30

7.63

6.59

-0.34

1.83

18.70

0.00

LNUSDJPY

244

4.66

0.13

4.99

4.34

-0.73

3.28

22.49

0.00

LNNEER

244

4.46

0.10

4.71

4.26

0.46

2.82

9.08

0.01

Table2Bshowsthesummarystatisticsofallthevariablesattheirfirstdifference.Allthevariablesaremorevolatileattheirfirstdifferences.Noneofthevariableshaveanormaldistribution,accordingtotheJarque-Beratests.Thehigher-maturityswapyieldsareright-skewedandthedistributionsforthefirstdifferenceofthestockindicesareleft-skewed.Allthevariablesareleptokurtic,showinganarrow,bell-shapeddistribution.ThechangeinthegrowthofindustrialproductionshowsalargedeclineinMarch2011,followingtheTohokuearthquakeandtsunami.

Table2B:SummaryStatisticsfortheFirstDifferencesoftheVariables

Vars

Obs.

Mean

Std.Dev.

Max.

Min.

Skewness

Kurtosis

J-B

Probability

?JPSWAP2Y

243

0.0004

0.04

0.19

-0.22

0.13

8.78

339.08

0.00

?JPSWAP5Y

243

0.0003

0.07

0.32

-0.23

0.91

6.96

192.06

0.00

?JPSWAP10Y

243

-0.0015

0.09

0.43

-0.23

1.14

6.61

184.06

0.00

?CALLRATE

243

0.0001

0.10

0.59

-0.67

-0.91

16.63

1913.93

0.00

?CORECPI

243

0.01

0.27

1.70

-1.70

0.29

21.19

3351.77

0.00

?IPYOY

243

-0.03

3.48

18.27

-16.66

0.39

9.36

415.96

0.00

13

?LNNIKKEI

243

0.004

0.05

0.12

-0.29

-1.19

7.92

302.45

0.00

?LNTOPIX

243

0.003

0.05

0.10

-0.24

-1.04

6.41

161.34

0.00

?LNUSDJPY

243

0.0005

0.02

0.07

-0.06

0.15

3.84

7.98

0.02

?LNNEER

243

-0.0006

0.02

0.11

-0.06

0.44

5.75

84.42

0.00

TheunitrootandstationaritytestresultsarepresentedinTables3Aand3B.Table3Aexhibitstheunitrootandstationaritytestsofthevariablesatthelevel.ItpresentsboththeaugmentedDickey-Fuller(ADF)(DickeyandFuller1979,1981)unitroottestsandtheKwiatkowski-Phillips-Schmidt-Shin(KPSS)(Kwiatkowskietal.1992)stationaritytests.ThenullhypothesesfortheADFandKPSStestsaredifferent.Theunitroottestsindicatethatmostofthevariablesarenonstationaryorhaveaunitroot.Theonestrongexceptionisthegrowthinindustrialproduction,whichshowsthepresenceofnounitrootbybothtypesoftests.

Table3A:UnitRootandStationarityTestsoftheVariables

Variablesat

level

ADFunitroottests(H0:unitroot)

KPSStests(H0:stationarity)tests

None

Intercept

Trend

Intercept

Trend

JPSWAP2Y

–1.06

–1.50

–2.11

0.95***

0.22***

JPSWAP5Y

–1.00

–1.54

–2.52

1.32***

0.20**

JPSWAP10Y

–0.91

–1.35

–2.45

1.61***

0.21**

CALLRATE

–1.76*

–2.31

–2.66

0.54**

0.16**

CORECPI

–2.35**

–2.28

–2.52

0.40*

0.12

IPYOY

–3.43***

–3.42**

–3.45**

0.09

0.04

LNNIKKEI

1.15

–1.19

–1.97

1.34***

0.27***

LNTOPIX

0.82

–1.51

–1.83

0.82***

0.25***

LNUSDJPY

0.08

–1.54

–1.72

0.35*

0.31***

LNNEER

-0.26

–1.95

–1.99

0.18

0.17**

Note:Significancelevels:***for1percent,**for5percent,and*for10percent

Table3Bshowstheunitrootandthestationaritytestsofthevariablesintheirfirstdifference.AllthevariablesbecomestationaryattheirfirstdifferenceperbothADFandKPSStests.TheKPSStestsrejectedthenullhypothesisofstationaritywiththetrendassumptionfortheswapyield.However,theoverallpictureprovidesprettystrongsupportforstationarityatthefirstdifference.

14

Table3B:UnitRootandStationarityTestsoftheFirstDifferencesoftheVariables

ADFunitroottests(H0:unitroot)

KPSStests(H0:stationarity)tests

None

Intercept

Trend

Intercept

Trend

?JPSWAP2Y

–11.02***

–11.00***

–10.98***

0.17

0.15**

?JPSWAP5Y

–12.40***

–12.37***

–12.35***

0.15

0.15*

?JPSWAP10Y

–12.50***

–12.47***

–12.45***

0.13

0.13*

?CALLRATE

–23.26***

–23.21***

–23.17***

0.08

0.06

?CORECPI

–15.82***

–15.81***

–15.79***

0.06

0.05

?IPYOY

–8.73***

–8.72***

–8.70***

0.02

0.02

?LNNIKKEI

–12.65***

–12.72***

–12.69***

0.07

0.06

?LNTOPIX

–12.41***

–12.43***

–12.41***

0.08

0.07

?LNUSDJPY

–11.90***

–11.88***

–12.01***

0.27

0.06

?LNNEER

–11.75***

–11.73***

–11.75***

0.11

0.06

SECTIONV:ECONOMETRICMODELSANDFINDINGS

Threedifferentmodelsoftheswapyieldwithstructuralbreaksareestimated.Inthefirstmodel,theswapyieldisjustafunctionoftheshort-terminterestrate,coreinflation,andthegrowthofindustrialproduction.Inthesecondmodel,theswapyieldisafunctionoftheshort-terminterestrate,coreinflation,thegrowthofindustrialproduction,thepercentagechangeintheNikkeistockindex,andthepercentagechangeintheexchangerate.Inthethirdmodel,theswapyieldismodeledasafunctionoftheshort-terminterestrate,coreinflation,thegrowthofindustrialproduction,thepercentchangeintheTopixstockindex,andthepercentagechangeinthenominaleffectiveexchangerate.Foreachmodel,swapyieldsofthreedifferentmaturitytenors—namely2-year,5-year,and10-year—areusedasthedependentvariablesintheregressionequations.

EconometricResults

Tables4A,4B,and4CreporttheBai-Perronbreaktests(BaiandPerron2003)ofthethreemodelsforeachmaturitytermoftheswaps.The2-yearswapyieldshowedonebreakinApril2014.However,thehighermaturity-termswapyields,namelythe5-yearand10-yearswap,havetwobreakpoints.Forthe5-yearswap,breaksoccurinAugust2007andApril2014.The10-yearswapyieldshavebreakpointsinJuly2009andApril2014.Thus,allswapmaturitiesshoweda

15

breakpointinApril2014.However,highermaturityswaprateshaveoneadditionalbreakpointthatcomesearlier,duringthe2007–9globalfinancialcrisis.

Table4A:Bai-PerronBreakTestsforthe2-yearSwapYield

SW2Y

SW2Y

SW2Y

F(1|0)

79.11*

85.01*

85.27*

F(2|1)

4.32

4.49

4.42

F(3|2)

Numberofbreaks

1

1

1

Nonbreakingcontrols

C,IPYOY

C,IPYOY,

?LNNIKKEI,

?LNUSJPY

C,IPYOY,

?LNTOPIX,

?LNNEER

Breakdates

2014M04

2014M04

2014M04

Note:Bai-Perron(2003)criticalvaluesfor0to1,1to2,and2to3breakpointsare11.47,12.95,and14.03,respectively.

Table4B:Bai-PerronBreakTestsforthe5-yearSwapYield

SW5Y

SW5Y

SW5Y

BreakTests

F(1|0)

14.30*

15.45*

15.49*

F(2|1)

43.13*

36.83*

35.80*

F(3|2)

1.35

1.83

1.78

Numberofbreaks

2

2

2

Nonbreakingcontrols

C,IPYOY

C,IPYOY,

?LNNIKKEI,

?LNUSJPY

C,IPYOY,

?LNTOPIX,

?LNNEER

Breakdates

2007M08

2014M04

2007M08

2014M04

2007M08

2014M04

Note:Bai-Perron(2003)criticalvaluesfor0to1,1to2,and2to3breakpointsare11.47,12.95,and14.03,respectively.

Table4C:Bai-PerronBreakTestsforthe10-yearSwapYield

SW10Y

SW10Y

SW10Y

F(1|0)

53.34*

17.76*

17.87*

F(2|1)

14.58*

54.29*

52.39*

F(3|2)

1.86

4.48

4.44

Numberofbreaks

2

2

2

Nonbreakingcontrols

C,IPYOY

C,IPYOY,

?LNNIKKEI,

?LNUSJPY

C,IPYOY,

?LNTOPIX,

?LNNEER

Breakdates

2009M07

2014M04

2009M07

2014M04

2009M07

2014M04

Note:Bai-Perron(2003)criticalvaluesfor0to1,1to2,and2to3breakpointsare11.47,12.95,and14.03,respectively.

The2-yearswapyieldmodelsarepresentedinTable5A.Inthepre-breakpointperiod,a100-basispointincreaseinthecallratewillincreasethe2-yearswapyieldby121–122basispoints.Theeffectisremarkablystablewithdifferentnonbreakingcontrolvariables.ThecoreinflationratehasanegativeimpactontheswapratesbeforeApril2014.However,inthepost-break

16

period,fromApril2014toDecember2022,thecallratedoesnotstatisticallyaffectthe2-yearswapyield.Thecoreinflationratereversedthesign,andhasapositiveandstatisticallysignificantrelationshiptotheswapyield,albeitthesizeoftheestimateissmallerthanbeforethebreakpoint.Noneofthecontrolvariableshaveanyimpactontheswapyield.VariousmodeldataarealsodisplayedinTable5A.TheadjustedR2impliesthatmuchofthevarianceintheswapyieldisexplainedbythecallrate,core