1.
Short term career goals: I am very interested in English learning and dedicated to English writing and translation. However, as a graduate, I lack some social experience and professional expertise. Therefore, within a short term, it is necessary to adapt myself to the working environment from campus life. Furthermore, I should cultivate myself to be competent for my position and get a general knowledge about the development trend of modern education. And at least I should be familiar with my job.
Long term career goals: I want to become an expert in my field, or I should be proficient in English writing and translation as well as in the business about study abroad. To become proficient, it will take a long time and need hard working. Meanwhile, a good relationship with the colleagues does a favor for work efficiency, so I will pay attention to communication with others in order to improve my comprehensive abilities.
2. Essay 1
I set my career goals on the basis of personal interests and professional knowledge so as to pursue a better future. The aims are lying ahead, and how to achieve them counts a lot. In order to realize my short term and long term goals, I made a detailed plan as follows. Firstly, I should adjust my mind to adapt to the working environment. The campus atmosphere is different from working context. The former is free and leisure and the latter is of high pressure and competitive, so I should have a positive and right attitude towards my job. Secondly, I should learn the knowledge about American universities and its education system. Although, I have learnt some knowledge about American education system, it is not complete and professional, and my understanding of American colleges is not profound. In order to write a suitable and excellent essay, it is a must for me to master adequate background knowledge. Thirdly, learning English is a life-long process and I should study every day in my spare time to improve my English. Meanwhile, I will also focus on my working process so as to enhance my abilities.
There may be some barriers on the road to my goals. One
challenge is how to communicate with the students who would like to apply for American top universities. Communication is of importance in the working process, so I will practice my communication skills to facilitate my work. Another challenge is the writing task. According to different purposes the essays serve, there are different writing styles. As a green hand, I am not very familiar with the correct writing style which is a big challenge to my academic knowledge. Fortunately, I learnt writing skills during my college study and hold the belief that practice makes perfect. Through the systematic training and practice, I believe I will be able to make it. Moreover, the third challenge is how to control and manage my time. The schedule management does matter a lot and it affects working quality and efficiency. As a result, I should have a clear timetable in my mind and separate priority from ordinary things. It is lucky enough that I was aware of cultivating my time sense when I was in college. Meanwhile, I have the professional spirit and good sense of responsibility, which encourages me to fulfill my duty in time.
Working is another way of learning and it is common to
meet lots of problems and challenges, but what matters is to find the right methods with a positive attitude. Hardships make people go ahead not fall down, and I think I can go through the difficulties with determination and perseverance.
第二篇:Long-run equilibrium, short-term adjustment, and spillover effects across Chinese segmented stock ma
Available online at
Int.Fin.Markets,Inst.andMoney18(2008)425–437
Long-runequilibrium,short-termadjustment,andspillovereffectsacrossChinesesegmentedstock
marketsandtheHongKongstockmarket
ZhuoQiaoa,d,ThomasC.Chiangb,?,Wing-KeungWongc,d
aResearchInstituteofEconomicsandManagement,SouthwesternUniversityofFinanceandEconomics,PRChina
bDepartmentofFinance,LeBowCollegeofBusiness,DrexelUniversity,Philadelphia,PA,USA
cDepartmentofEconomics,HongKongBaptistUniversity,HongKong
dDepartmentofEconomics,NationalUniversityofSingapore,Singapore
Received7June2006;accepted24May2007
Availableonline6June2007
Abstract
ThispaperadoptsanovelFIVECM-BEKKGARCHapproachtoexaminethebilateralrelationshipsamongtheA-shareandB-sharestockmarketsinChinaandtheHongKongstockmarket.Theevidenceshowsthatthesestockmarketsarefractionallycointegrated.AnalysesofthespillovereffectsacrossthesemarketsindicatethattheA-sharemarketsaremostin?uential.TherelaxationofgovernmentrestrictionsonthepurchaseofBsharesbydomesticresidentsacceleratesthemarketintegrationprocessofA-sharemarketswiththeB-shareandHongKongmarkets.TheeffectsoftheAsiancrisisonthestock-returndynamiccorrelationsvaryacrossthesemarkets.
?2007ElsevierB.V.Allrightsreserved.
JELclassi?cation:G10;C32;F36
Keywords:Stockmarketsegmentation;Cointegration;FIVECM;MultivariateGARCH
1.Introduction
AsamechanismfordevelopingtheChinesestockmarket,theChinesegovernmenthasadoptedamarketsegmentationpolicythatdividesitsstockmarketintoadomesticboardandaforeignboardtocatertotheneedsofdifferentinvestors.CompaniescanissueAshares,whichonly?Correspondingauthor.Tel.:+12158951745;fax:+16092650141.
E-mailaddress:Chiangtc@drexel.edu(T.C.Chiang).
1042-4431/$–seefrontmatter?2007ElsevierB.V.Allrightsreserved.
doi:10.1016/j.int?n.2007.05.004
426Z.Qiaoetal./Int.Fin.Markets,Inst.andMoney18(2008)425–437
ChinesecitizenslivinginmainlandChinacanbuy;theyarealsoallowedtoissueBshares,whichcanbeboughtbyforeigninvestors,includingChineseinvestorsresidinginHongKong(HK),Macau,orTaiwan.1AandBsharesarelistedontheShanghai(SH)StockExchange(SHSE)andtheShenzhen(SZ)StockExchange(SZSE),namely,SHA,SHB,SZA,andSZBinmainlandChina.Asharesaredenominatedinthelocalcurrency(RMB),whileBsharesaredenominatedinU.S.dollarsontheSHSEandinHKdollarsontheSZSE.BecauseoftheisolationofChinesecurrencyfromforeigncurrencies,differentinformationenvironments,diverseregulatorypolicies,andheterogeneousinvestors,thesegmentedmarketshaveshownvariouspatternsofevolution.
HongKongisanimportantpartnerofmainlandChinaforherlocation,economicdevelopment,andpoliticalrelationship.Frombothgeographicalandstrategicpointsofview,HongKongactsasanintermediaryforChina’sinternationaltradethroughre-exportsandoffshoretransactions.Inaddition,asubstantialamountofcapitalto?nanceChina’seconomicexpansionhasbeenraisedthroughHongKong’schannels.However,thisintermediaryroleformanagingandengaginginter-nationaltradeandcapitalalsoshapesHongKong’seconomicstructure,leadingtohereconomicprosperity.Thesemultilateraleconomicactivitieshelpinbridgingknow-howgap,transferringtechnology,disseminatingandprocessinginformation,creatinginvestmentopportunities,andgeneratinghigherreturn,buttheyalsoassumehigherriskforstockmarketsinbothHongKongandmainlandChina.Itisthisuniquesettingaswellastheincreasinglyimportantroleplayedinworld?nancialmarketsthatpromptsustoexploretheexplicitlong-runequilibrium,short-runadjustment,andspillovereffectsacrossthemainlandChinastockmarketsontwostockexchanges(SHSEandSZSE)andtheHKstockmarket.
Inthisstudy,weincorporateafractionallyintegratedvectorerrorcorrectionmodel(FIVECM)intotheBEKKGARCHframework(EngleandKroner,1995)toexaminethebilateralrelation-shipsbetweeneachofthefollowingsixpairsofstockmarkets:HK–SHA,SHB–SHA,SHB–HK,HK–SZA,SZB–SZA,andSZB–HK.This?ndingwillbeveryusefultoinvestors,sincethepresenceofthefractionalcointegrationimpliestheexistenceofbothlong-runequilibriumandlong-periodiccomovementsbetweenthetwomarkets.Asaresult,itwouldaffectinvestors’assetallocationstrategiesinthelongandmediumterms(CheungandLai,1995).Atthesametime,thepresenceofafractionalcointegratingrelationshipbetweentwostockmarketshasanimpor-tantimplicationfortheirshort-runlinkages.AsageneralizationofthestandardlinearVECM,whichallowsonlythe?rst-orderlagofthecointegrationresidualtoaffecttheequilibriumrela-tionship,theFIVECMspeci?cationisappealing,sinceitnotonlyhelpsinvestorstoobserveshort-runadjustmentsandlong-termequilibriumrelationshipsamongco-integratedvariables,butalsoaccountsforthepossiblelongmemoryinthecointegrationresidualseriesthatother-wisemightdistorttheestimation(Dingetal.,1993).Finally,incorporatingtheFIVECMintoabivariateBEKKformulationallowsustocapturethesecondmomentautocorrelationsofthereturnseriesandanalyzethe?rstandsecondmomentspillovereffectsacrossthesestockmarketssimultaneously.
Ourempiricalresultsshowthatallsixpairsofstockmarketsarefractionallycointegrated.Ineachofthesixpairs,onlyonemarketadjuststoreturntoequilibrium.Wealso?ndbi-directionalvolatilityspillovereffectsbetweentheA-sharemarketsandtheB-sharemarketsandbetweentheB-sharemarketsandtheHongKongmarket.However,we?ndonlyunidirectionalvolatilityThisrestrictionwasrelaxedon22February2001,whenitbecamepermissiblefordomesticcitizenstobuyandsellBshares.1
Z.Qiaoetal./Int.Fin.Markets,Inst.andMoney18(2008)425–437427
spillovereffectsfromtheA-sharestockmarketstotheHongKongstockmarket.TheevidenceconcludesthattheAsharesarethemostin?uentialmarketsinbothmeanandvolatilityspillovereffects.Investigationofthedynamicpathofcorrelationcoef?cientssuggeststhatrelaxationofgovernmentrestrictionsonthepurchaseofBsharesbydomesticresidentsincreasedthecorrelationbetweentheA-andB-sharemarketsandacceleratedthemarketintegrationprocessoftheA-sharemarketswiththeHongKongstockmarket.OurresultsalsosuggestthattheAsiancrisishadadifferentspillovereffectonstock-returndynamiccorrelationsacrossChinesesegmentedmarketsandtheHongKongstockmarket.
Theremainderofthispaperisorganizedasfollows:Section2offersareviewoftherelevantliterature,Section3discussesthedataandmethodology,Section4providesempiricalresults,andSection5summarizesourconclusionsandcomments.
2.Literaturereview
MuchempiricalworkhasbeendoneonanalyzingtheChinesesegmentedstockmarketsandthelinkagesbetweenChinesesegmentedstockmarketsandinternationalstockmarkets.Forinstance,Lietal.(2006)?ndthattheriskpremiumsassociatedwiththeHongKongandmainlandChinesemarketsinatwo-factormodelsuccessfullyexplainthecross-sectionofreturnsontheAandHshares.TheyconcludethattheriskpremiumsassociatedwiththesegmentedA-shareandH-sharemarketsexertcrucialimpactsonthepricedifferentialsbetweenthetwoclassesofshares.Chakravartyetal.(1998)reportthebivariatereturncorrelationsamongtheA-andB-shareindices,aswellasHongKong,Japanese,andU.S.marketindicesandsuggestthattheChinesemarketisstillisolated,evenaftertheintroductionofBshares.WangandFirth(2004)?ndaunidirectional-returnsspillovereffectfromdevelopedstockmarketstostockindicesintheGreaterChinaeconomiczone.
SeveralgroupshaveappliedGrangercausalityteststodeterminethelead–lagrelation-shipsbetweentheA-shareandB-sharemarkets.Forexample,KimandShin(2000)?ndthattheA-sharemarketsleadtheB-sharemarketsbefore1996,buttherelationshipeitherdis-appearsorreversesafter1996.Ontheotherhand,Laurenceetal.(1997)observeacausalrelationshipfromtheSHBtoallotherChinesemarketsandfeedbackfromSHAandSZBtoSHB.Inaddition,adoptingVARandbivariateGARCH-Mmodels,Yehetal.(2002)?ndthattheunexpectedchangesinthepremiumratioofA-sharepriceoverB-sharepricecon-tributetothereturnvolatilityofbothAandBshares.Chiangetal.(inpress)presentevidencethatthecorrelationcoef?cientsbetweenA-shareandB-sharestockreturnsaretimevary-ing.Theirresultssuggestthatthetime-varyingcorrelationsaresigni?cantlyassociatedwithexcessivetradingactivitymeasuredbyexcessivetradingvolumesandhigh-lowpricedifferen-tials.
Ourworkextendstheexistingliteratureinthefollowingtwoways.First,ourpaperisthe?rsttoexaminetheequilibriummechanismamongsegmentedChinesestockmarketsandtheHongKongstockmarket.TheFIVECMapproachisamoregeneralspeci?cationbecauseitcontainsboththetraditionalVECMandtheeffectsofthelongmemoryofthecointegratingrelationship,whichisimportantforrevealingthetruerelationshipsamongmarkets.Second,bycombiningtheFIVECMwithabivariateBEKKGARCHformulation,thismodelallowsustoinvestigateChinesestock-returnlinkagesinamultivariateframework.Differentfrompreviousstudiesthatmainlyuseunivariatemodels,thebivariateBEKKGARCHmethodenablesustodetecttheconditionalcorrelationsbetweenthesemarkets.Withinthisnovelframework,empiricallead-lagrelationshipsinthemeanaswellasvolatilityinacross-marketsettingcanbesimultaneouslyestimated.
428Z.Qiaoetal./Int.Fin.Markets,Inst.andMoney18(2008)425–437
Theempiricalresultsderivedfromthisresearchrevealthenatureofthecomplicatedstructurebetweentwodifferentmarkets,which,inturn,providesadditionalinformationtoinvestorsandfundmanagersfortheirinvestmentdecisionsandstrategiesinthesemarkets.Our?ndingsarealsousefulforpolicymakersinsettingregulationsforthesemarkets.
3.Dataandmethodology
3.1.Data
ThedatainthisstudyincludepriceindicesofShanghaiA-share(SHA),ShenzhenA-share(SZA),ShanghaiB-share(SHB),ShenzhenB-share(SZB),andHongKongHangSeng(HK).AlldataaretakenfromDataStreamInternational,coveringJanuary1995throughDecember2005.Inlightoftheevidenceoftheunusualpatternoftheday-of-the-weekeffectobservedbyCaietal.(2006),inthispaper,weemployweeklyWednesdayindicestoalleviatetheimpactofnoisecreatedbyusingdailydataandtoavoidday-of-the-weekeffects.2
3.2.Methodology
Oneoftheprincipaltasksinthispaperistoexaminestock-returnbehaviorbyexploringtheshort-rundynamicsinrelationtothelong-runequilibriuminacross-marketsetting.Toachievethisend,weemploythecointegrationtesttoexaminewhethertwoseriesthatdriftedapartinlong-runequilibriumhaveatendencytobebroughtbacktogetheragain.Usually,thedisequilibriumerrorusedintheVECMframeworkisneitherI(1)norI(0)butfollowsafractionallyintegratedprocess,I(d),where?0.5<d<0.5(EngleandGranger,1987).Withoutaccountingforthelongmemory(whend<0.5)featureofthedisequilibriumerror,thetruerelationshipsamongcointe-gratedvariablesdisclosedbytraditionalVECMmaybemisspeci?ed.Tocircumventthisproblem,weemployafractionallyintegratedVECM(Baillie,1996)tostudythenatureofcomovementsforeachpairofstock-returnseries.
First,weemploytheEngleandGranger(1987)two-stepapproachby?ttingthefollowingdynamicordinaryleastsquaredmodel(DOLS)(Saikkonen,1991)tothepairsofstockindicesandthereafterobtaintheestimatedcointegratingresidualz?t:
y1t=α+βy2t+p??ωj??y2t?j+vt,(1)
j=?p
wherey1tandy2tareapairofstockindicesinnaturallogarithms;eachcouldrepresentSHA,SHB,SZA,SZB,andHK.Byusingthisprocedure,wecanremovethedeleteriouseffectofshort-run?,whichissuper-consistentaswelldynamicsintheequilibriumerrorνtandobtaintheestimateβasef?cient(StockandWatson,1993).
Next,totestfortheexistenceofanylongmemoryinthez?tseries,weuseLo’smodi?edR/Stest(Lo,1991).Ifwecon?rmthatz?tfollowsanI(d)(?0.5<d<0.5)process,theny1tandy2tareCaietal.(2006)?ndthataverageMondayreturnsfromA-shareindicesaresigni?cantlynegativeduringthethirdandfourthweeksbutaverageTuesdayreturnsonmostoftheA-shareandB-shareindicesarenegativeduringthesecondweekofthemonth.2
Z.Qiaoetal./Int.Fin.Markets,Inst.andMoney18(2008)425–437429
saidtobefractionallycointegrated.Inthissituation,weproceedto?tthefollowingautoregressivefractionallyintegratedmovingaverage(ARFIMA)model3:
?t=atΨ(L)?1Φ(L)(1?L)dz(2)
whereΨ(L)andΦ(L)areMAandARpolynomials,Lisabackwardshiftoperator,andatisani.i.d.noise.Ifthereisanycointegrationrelationshipamongthevariables,aVECMrepresentationcanbeestablishedtoadequatelycapturetherelevantlong-runandshort-termrelationships(EngleandGranger,1987).WeincorporatetheVECMintothefollowingbivariateFIVECMtoaccountforthefractionalintegrationpropertyinz?tseriesbyemployingtheARFIMAmodel(2):
??y1t=c1+α1at?1+
??y2t=c2+α2at?1+m??i=1m??
i=1iφ21??y1t?i+iφ11??y1t?i+m??i=1iφ12??y2t?i+ε1t,m??i=1iφ22??y2t?i+ε2t.(3)
Both??y1tand??y2tinEq.(3)representthereturnseriesforeachpairofstockindices,namely,HK–SHA,SHB–SHA,SHB–HK,HK–SZA,SZB–SZA,andSZB–HKinthisstudy;εt=(ε1t,ε2t)??isthevectoroferrorterms;thecoef?cientsα1andα2indicatetheshort-rundynamicadjustmentswiththeirmagnitudesrepresentingthespeedsoftheadjustment.AVAR(m)structureintheVECMmodel,inparticularm=1,isemployedinthispaper.
Tocapturetheheteroskedasticityinthereturnseriesandtoensurethatthevariancematrixoferrortermsispositivede?nite,weapplythefollowingbivariateBEKK(1,1)model(EngleandKroner,1995)4:
??)fort=1,2,...T;εt|Ωt?1~N(0,t????????σ11,tσ12,t????????=A0A0+A1(εt?1εt?1)A1+B1=B1,(4)tt?1σ21,tσ22,t
whereεtisassumedtofollowabivariatenormaldistributionconditionalonthepastinformationsetΩt?1;??tdenotesthevariance–covariancematrixofεt,whichissymmetricandpositivesemi-de?nite;A0isalowertriangularmatrix;A1andB1areunrestrictedsquarematrices.Onthebasisofthisframework,thevolatilityspillovereffectsacrossreturnseriesindicatedbytheoff-diagonalentriesofcoef?cientmatricesA1andB1areestimated.TheBEKKspeci?cationisamoregeneraland?exiblemultivariateGARCHmodelastherearenorestrictionsimposedonthecoef?cients.Inthispaper,weestimatetheFIVECM-BEKKmodel(i.e.,systems(3)and(4)jointly)inwhichthecoef?cientestimateswouldbemoreef?cient,andtherelationshipsamongtheserieswouldbedelineatedmoreaccurately.
For0<d<0.5,theARFIMAprocessissaidtopossessalongmemoryorlong-rangedependence.Whend=0,anARFIMAprocesscanbereducedtotheconventionalARMAprocess.For?0.5<d<0,ithasashortmemory.For0.5<d<1,theprocessismean-revertingbecausethereexistsanon-long-runeffectofaninnovationonthefuturevaluesoftheprocess.Ford>1,theprocessisnotmean-reverting,sinceanyshocktotheprocesscouldmakeitdriftawayfromitsequilibriumpermanently.SeeBaillie(1996)formoreinformation.
4BEKK(1,1)isusuallysuf?cienttomodelvolatilityin?nancialtimeseries.3
430Z.Qiaoetal./Int.Fin.Markets,Inst.andMoney18(2008)425–437
Table1
DescriptivestatisticsforChinesestockindicesandtheHangSengindex
SHA
MeanMedianMaximumMinimumS.D.
SkewnessKurtosis
5.0795.1475.6434.1690.353?0.8703.182
SHB4.2724.2135.4543.0680.5070.0802.340
SZA3.7993.9094.4402.5720.468?1.1923.719
SZB2.8722.8703.9941.6800.568?0.1251.667
HK7.3277.3347.7516.7860.202?0.1572.272
Note:thesearedescriptivestatisticsofthelogarithmsofstockindices.SamplecoversJanuary1995throughDecember2005.Thetotalnumberofobservationsis574.Table2
LongmemorytestsoncointegrationresidualsResidualseries
Modi?edrangeoverS.D.(R/S)testTeststatistic
zHKSHAzSHBSHAzSHBHKzHKSZAzSZBSZAzSZBHK
2.0403.5513.6861.9513.9344.135
P-Value<0.05<0.01<0.01<0.05<0.01<0.01
Note:theresidualseriesareconstructedusingEq.(1)inthetext.Superscriptstandsforadependentvariableandsubscriptforanindependentvariable.
4.Empiricalresults
AsummaryofthebasicstatisticsofthenaturallogarithmvalueofpriceindicesisreportedinTable1.ByconductingtheADFandPPunitrootteststotesttheirstationaryproperty,theresultsindicatethatalloftheseseriesareI(1).5Thenextstepistoestimatethesixcointegrationresidualsz?tbasedonEq.(1)forthesixpairsofstockindices:HK–SHA,SHB–SHA,SHB–HK,HK–SZA,SZB–SZA,andSZB–HK.ThisisdonebyperformingaDOLSestimationwithlag
SHBSHBHKSZBSZBlengthp=2.TheresultingerrorseriesaredenotedbyzHKSHA,zSHA,zHK,zSZA,zSZAandzHK,
respectively,wherethesuperscriptstandsforthedependentvariable,andthesubscriptfortheindependentvariable.
WeuseLo’smodi?edR/Stest(Lo,1991)toexaminethelong-memorybehaviorinthesixresidualseries,andtheresultsarecontainedinTable2.Theevidenceindicatesthatallresid-ualserieshavelongmemory.This?ndingleadsustoapplytheARFIMAmodeltomodeleachofthesesixseries.ItmaybeseenfromTable3thatalloftheestimatedvaluesofdfallintotherange(0,0.5)andtheestimatedcoef?cientsoftheARtermsmeetthestationarycondi-tion,suggestingthatthecointegratingvariablesfollowlong-memorystationaryprocesses.We,
5
Theresultsareavailableonrequest.
Z.Qiaoetal./Int.Fin.Markets,Inst.andMoney18(2008)425–437
Table3
ARFIMA?tresults
ResidualseriesARFIMA(p,d,q)
dzHKSHA
zSHBSHA
zSHBHK
zHKSZA
zSZBSZA
zSZBHK0.093(0.046)**0.064(0.037)*0.100(0.039)***0.102(0.046)**0.171(0.041)***0.211(0.044)***AR(1)4310.967(0.015)***0.991(0.006)***0.988(0.008)***0.966(0.015)***0.982(0.010)***0.972(0.014)***Note:theresidualseriesareconstructedusingEq.(1)inthetext.Superscriptstandsforadependentvariableandsubscriptforanindependentvariable.Numbersinparenthesesarestandarderrors.SelectionofAR(lag)andMA(lag)terms,pandq,isbasedontheexaminationofACF,PACF,andBayesianInformationCriterion(BIC).***,**and*indicatesigni?canceatthe1,5and10%levels,respectively.
therefore,concludethatthesixpairsofstockmarketsarefractionallycointegratedwitheachother.
Havingveri?edthefeatureofthelong-termcointegrationrelationshipsineachpairofthestockindices,weproceedtoestimatetheFIVECM-BEKK(1,1),andtheresultsarepresentedinTable4.Theestimatedstatisticsallowustoanalyzetheshort-termadjustment,thelong-termequilibriumrelationship,andthespillovereffectsbetweeneachsegmentedChinesemarketandtheHKmarket.1(φ1)forHK–SHAmeasuresthemeanspillovereffectsSpeci?cally,theestimatedcoef?cientφ1221fromtheSHA(HK)markettotheHK(SHA)market;ARCH(1,2)andGARCH(1,2)(ARCH(2,1)andGARCH(2,1))measurethevolatilityspillovereffectsfromtheSHA(HK)markettotheHK(SHA)market,andparametersα1andα2indicateshort-termadjustmentstotheequilibriumoftheHKandSHAmarkets,respectively.
4.1.RelationshipsamongHongKong,ShanghaiA-andB-sharestockmarkets
1hasanegativesignandisstatisticallysigni?cant,butφ1isnotFromPanelAofTable4,φ1221signi?cant,indicatingthatthereisameanspillovereffectfromtheSHAmarkettotheHKmarket,
butthereverserelationshipdoesnothold.Thevalueoftheshort-termadjustmentparameterα1is?0.683,whichissigni?cantatthe1%level,suggestingthattheHKstockmarketmakesapartialadjustmentwhenitdriftsawayfromlong-runequilibrium.Themagnitudeofα1indicatesthat68.3%ofthediscrepancybetweenthetwostockmarketswouldbecorrectedineachweek,correspondingtoanadjustmentperiodof1.46weeks.α2bearsapositivesignbutisinsigni?cant,suggestingthatthecointegratingrelationshipbetweenthetwomarketsdoesnotrevealmovementsonthepartoftheSHAstockmarket.Forthevarianceequation,we?ndthatoff-diagonalARCH(1,2)andGARCH(1,2)termsaresigni?cant,whilethoseofARCH(2,1)andGARCH(2,1)arenot,indicatingtheexistenceofaunidirectionalvolatilityspillovereffectfromtheSHAtotheHKstockmarket.1WithrespecttotheestimatesoftheSHB–SHApairmarket,theevidenceshowsthatφ121isnot,indicatingaunidirectionalmeanspillovereffectfromtheSHAtoissigni?cantbutφ21theSHBstockmarket.Inotherwords,SHAleadsSHBinreturns.Thevalueoftheshort-term
Table4
EstimatesforFIVECM-BEKK(1,1)model
HK–SHASHB–SHASHB–HKHK–SZASZB–SZASZB–HK
PanelA:estimatedresults
c10.002(0.001)*?0.001(0.002)?0.001(0.002)0.002(0.001)*0.000(0.002)0.000(0.002)c20.000(0.001)0.000(0.001)0.001(0.001)?0.001(0.002)?0.002(0.001)0.002(0.001)φ1110.729(0.289)***0.885(0.440)**0.650(0.342)**0.676(0.282)**0.507(0.284)*0.460(0.183)***φ1?
φ121?0.217(0.074)***0.2740.497(0.300)(0.294)*?0.0390.009(0.201)(0.058)??0.1640.154(0.06)***(0.324)?0.0880.134(0.139)(0.172)??0.2060.024(0.148)(0.098)φ211?0.078(0.253)0.032(0.075)?0.188(0.203)0.055(0.049)0.066(0.074)0.025(0.098)0.019(0.078)α22
1?0.683(0.286)***?0.824(0.444)*?0.636(0.354)*?0.639(0.280)**?0.481(0.284)*?0.360(0.195)*α20.097(0.252)?0.263(0.302)?0.093(0.200)0.188(0.325)?0.065(0.143)?0.018(0.097)A(1,1)0.003(0.002)*0.014(0.002)***0.018(0.002)***0.002(0.002)0.013(0.002)***0.014(0.002)***A(2,1)0.008(0.003)***0.007(0.001)***0.000(0.002)0.010(0.006)*0.010(0.001)***?0.001(0.001)A(2,2)0.000(37.495)0.000(8.743)0.000(27.681)0.001(0.077)0.003(0.002)0.000(3.055)ARCH(1,1)0.252(0.033)***0.437(0.036)***0.420(0.036)***0.206(0.028)***0.436(0.031)***0.387(0.023)***ARCH(1,2)0.060(0.028)**0.132(0.051)***?0.146(0.060)***0.038(0.023)*0.185(0.040)***?0.282(0.049)***ARCH(2,1)0.010(0.055)0.061(0.023)***0.062(0.027)**0.002(0.062)0.018(0.027)0.005(0.021)ARCH(2,2)0.277(0.032)***0.257(0.028)***0.226(0.042)***0.376(0.040)***0.422(0.034)***0.156(0.035)***GARCH(1,1)0.962(0.010)***0.864(0.019)***0.831(0.033)***0.977(0.007)***0.886(0.013)***0.880(0.015)***GARCH(1,2)?0.028(0.009)***?0.044(0.021)**0.085(0.035)***?0.024(0.009)***?0.103(0.015)***0.040(0.023)*GARCH(2,1)?0.009(0.015)?0.038(0.010)***?0.044(0.018)***?0.002(0.015)?0.024(0.012)**0.019(0.008)**GARCH(2,2)0.940(0.014)***0.958(0.010)***0.980(0.014)***0.897(0.019)***0.888(0.016)***0.980(0.007)***PanelB:modeldiagnosticstatistics
LB(10)-HK8.019NA6.8137.813NA7.366
LBS(10)-HK7.681NA5.68010.503NA6.768
LB(10)-SHA12.26110.712NANANANA
LBS(10)-SHA3.6948.143NANANANA
LB(10)-SHBNA9.61711.472NANANA
LBS(10)-SHBNA6.1078.649NANANA
LB(10)-SZANANANA10.32912.461NA
LBS(10)-SZANANANA6.5016.895NA
LB(10)-SZBNANANANA22.993***20.205**LBS(10)-SZBNANANANA10.48616.605*
Note:theestimatesarebasedonEqs.(3)and(4)inthetext.Thedependentvariableineachmodelismarkedinbold.The?rst-orderARCH(i,j)andGARCH(i,j)termsaretheelementsoftheARCHandGARCHcoef?cientmatricesA1andB1inEq.(4)Numbersinparenthesesarestandarderrors.***,**and*indicatesigni?canceatthe1,5and10%levels,respectively.LB(10)andLBS(10)aretheLjung-Boxstatisticsbasedonthelevelandthesquaredlevelofthetimeseriesuptothe10thlag.432Z.Qiaoetal./Int.Fin.Markets,Inst.andMoney18(2008)425–437
Z.Qiaoetal./Int.Fin.Markets,Inst.andMoney18(2008)425–437433
adjustmentparameterα1is?0.824,whichissigni?cantataboutthe6%level,suggestingthattheSHBmarketadjustswhenitdriftsawayfromlong-runequilibriumandtheadjustmentspeedisabout1.21weeks.Thenon-signi?canceoftheα2estimateindicatesthattheSHAmarketisnotboundbythecointegrationrelationship.Forthevarianceequation,thetestresultsshowthatalloff-diagonalARCHtermsandGARCHtermsaresigni?cantanddiscloseabi-directionalvolatilityspillovereffectbetweentheSHAandSHBstockmarkets,implyingstrongtransmissionofinformationbetweenthesetwostockmarkets.
ThethirdcolumnofTable4providestheestimatesfortheSHB–HKpairofstockreturns.From1norφ1issigni?cant,concludingthatthereisnospillovertheresults,we?ndthatneitherφ1221effectinthe?rstmomentbetweentheSHBandtheHKmarkets.Theadjustmentspeedcoef?cientα1issigni?cantlynegative,whilecoef?cientα2isinsigni?cant,implyingthattheHKmarketisnotboundbythecointegrationrelationship.Forthespillovereffectofvolatility,evidenceshowsthatalloff-diagonalARCHtermsandGARCHtermsaresigni?cant,whichissimilartothesituationfortheSHA–SHBpair.Thus,weconcludethattherealsoexistsstrongtransmissionofinformationbetweentheSHBandHKstockmarkets.
Tosumup,evidenceshowsthattherearebi-directionalvolatilityspillovereffectsbetweenSHB–SHAandbetweenSHB–HK,butonlyaunidirectionalvolatilityspillovereffectfromtheSHAtotheHKmarket.WeconcludethattheSHAmarketplaysthemostin?uentialroleamongthethreemarkets:itnotonlypassesreturnrealizationstotheSHBandHKmarkets,butitalsoleadsinthetransmissionoftheirvolatilities.Wealso?ndthatamongthethreepairsofstockmarkets,onlyonemarketisfoundtoadjusttoreturntoequilibrium:theHKmarketadjustsdisequilibriumconditionswiththeSHAmarket,whiletheSHBmarketadjustsinresponsetodisequilibriumwithboththeSHAandHKmarkets.
4.2.RelationshipsamongHongKongandShenzhenA-andB-sharestockmarkets
Columns4–6ofTable4reporttheestimatedresultsofFIVECM-BEKKforHK–SZA,SZB–SZA,andSZB–HK.Ingeneral,we?ndthattherelationshipsamongthesethreemarketsareverysimilartothoseoftheircounterpartspresentedinthesub-sectionabove.FortheHK–SZApair,we?ndaunidirectionalmeanspilloverfromtheSZAmarkettotheHKmarket.Next,theestimatedparameterα1hasavalueof?0.639,whichisstatisticallysigni?cant,suggestingthattheHKmarketadjustsasitdivergesfromitslong-runequilibriumwiththeSZA.Thelengthoftheadjustmentis1.56weeks.Ontheotherhand,theestimateofα2isnotsigni?cant,indicatingthatthemovementoftheSHAstockmarketisnotgovernedbythecointegratingrelationshipbetweenthesetwomarkets.Bycheckingwiththevarianceequations,theresultsindicatethatthevolatilityspillovereffectisrunningonlyunidirectionallyfromtheSZAmarkettotheHKstockmarket.
AsweinspecttheSZB–SZApairrelationship,ourresultsshowthatthereisnomeanspillovereffectbetweenthesetwomarkets.Fortheshort-termadjustmentparameter,we?ndα1tobe?0.481andstatisticallysigni?cant.Thecomparablecoef?cient,α2,alsoshowsanegativesign;however,itisinsigni?cant,suggestingthattheSZBmarketmakestheadjustmentwhenitdeviatesfromalong-runequilibriumrelationshipandthespeedofadjustmentisabout2.08weeks.Incontrast,noevidenceindicatesthattheSZAmarketisboundbythecointegrationrelationship.Onthebasisoftheconditionalvarianceequa-tions,we?ndthattherearebi-directionalvolatilityspillovereffectsbetweentheSZAandSZBstockmarkets,implyingstrongtransmissionofinformationbetweenthetwostockmarkets.
434Z.Qiaoetal./Int.Fin.Markets,Inst.andMoney18(2008)425–437
1andφ1areinsigni?cant,thereFinally,weexaminetheSZB–HKpairofmarkets.Sincebothφ1221isnoevidencetoindicateanyspillovereffectinstockreturnsbetweentheSZBandHKmarkets.
Asfarastheadjustmentcoef?cientisconcerned,theestimatedα1is?0.360andstatisticallysigni?cant,whileα2isinsigni?cant,revealingthatthedisequilibriumbetweenthetwomarketswillbecorrectedonlybytheSZBmarket,andthecorrectionwilloccurwithin2.78weeks.Withrespecttothespillovereffectofvolatility,evidenceindicatestheexistenceofbi-directionalvolatilityspillovereffectsbetweentheSZBandHKmarkets.
ThemodeldiagnosticsreportedinPanelBofTable4listLjung-Boxtestsofwhitenoiseappliedtoboththestandardizedresidualsandthesquaredstandardizedresidualseries.ItdemonstratesthatnoneoftheLjung-BoxteststatisticsfortheHK–SHA,SHB–SHA,SHB–HK,andHK–SZApairsofmarketsissigni?cant,indicatingtheadequacyofthe?ttedmodelstosuccessfullycapturethedynamicsinthe?rsttwomomentsoftheindexreturnseries.However,thenullhypothesisoftheabsenceofjointsigni?cancefortheSZBreturnseriesoftheSZB–SZAandtheSZB–HKpairsisrejected,pointingtotheneedforfurtherinvestigationofthedynamicsoftheSZBmarket.6Inshort,interrelationshipsamongthesethreemarketsareverysimilartothoseamongtheSHA,SHB,andHKstockmarkets:therearebi-directionalvolatilityspillovereffectsbetweenSZB–SZAandbetweenSZB–HKandonlyaunidirectionalvolatilityspillovereffectfromtheSZAmarkettotheHKmarket.Similarly,we?ndthattheSZAmarketisthemostin?uentialamongthethreemarkets:itnotonlypassesreturnrealizationstotheHKmarket,butitalsoleadsthetransmissionofinformationaboutvolatility.Moreover,amongHK–SZA,SZB–SZA,andSZB–HK,onlyonesideoftheconnectedmarketsischaracterizedbyapartialadjustmentprocesstoreturntolong-runequilibrium.We?ndthattheHKmarketiscapableofadjustingitsdisequilibriumpositionstotheSZAmarketandtheSZBmarketadjustsinresponsetodisequilibriumwithboththeSZAandHKmarkets.7
4.3.Analysesofdynamiccorrelations
Havingmodeledthelong-termequilibrium,short-termadjustment,andspillovereffectsacrossthesemarkets,itisofinteresttoanalyzetheeffectsofchangesin?nancialpolicyandeconomicconditionsonthedynamiccorrelationsbetweenthemarkets.Byvisualinspectionofthese?g-ures,weidentifytwointerestingpoints.8First,afterFebruary2001,thecorrelationsbetweentheSHA–SHBandSZA–SZBmarketsshowanupwardtrendovertime.ThispatternmaybeattributabletothemoreliberalgovernmentalpolicyallowingdomesticcitizenswhoinvestinA-sharemarketstoinvestinB-sharemarkets.Second,thetime-varyingcorrelationcoef?cientsshowdifferentpatternsofevolutionduringtheAsian?nancialcrisis,whichstartedinearlyJuly1997.Forexample,fromlate1997throughearly1998,we?ndthatthecorrelationsbetweenWetriedotherspeci?cationsforthemeanequationfortheSZBreturnseries,buttheresultsdonotimprove.Itcouldbeduetospuriouscorrelationswithsomemissingvariables.
7Itisofinteresttocomparethespeedsofadjustmentacrossdifferentmarkets.OurevidenceshowsthattheSHBmarkethasafasterspeedofadjustmentinresponsetodisequilibriumwiththeSHA(α1=?0.824)thanitdoeswiththeHKmarket(α1=?0.636).Similarly,theSZBmarketadjustsfasterinresponsetodisequilibriumwiththeSZA(α1=?0.481)thanitdoeswiththeHKmarket(α1=?0.360).OnepossiblereasonforthisphenomenonisthatallofthecompanieslistedontheSHBandSZBmarketsbelongtolocalChinesecompanies.Thus,theirpricediscoveryprocessrelativetoA-sharemarketsismoreef?cientthantheirpricediscoveryprocessrelativetotheHKstockmarket.
8Weinvestigatedthetime-varyingconditionalcorrelationcoef?cientsestimatedfromtheFIVECM-BEKKmodelforeachpairofmarkets.The?guresarenotshowntosavespace.However,theyareavailableuponrequest.6
Table5
EffectsofcrisisandpolicychangeonconditionalcorrelationacrossChinesesegmentedstockmarketsandHongKongstockmarket
EstimatesMarkets
HK–SHASHB–SHASHB–HKHK–SZASZB–SZASZB–HK
d00.013(0.010)0.437(0.010)***0.194(0.009)***0.024(0.009)***0.442(0.012)***0.194(0.010)***d10.145(0.042)***0.078(0.041)*0.023(0.036)0.172(0.038)***0.033(0.050)0.161(0.039)***d2?0.103(0.023)***?0.223(0.022)***0.220(0.020)***?0.097(0.021)***?0.224(0.027)***0.007(0.021)d30.168(0.015)***0.244(0.014)***?0.067(0.013)***0.122(0.013)***0.190(0.017)***0.024(0.014)*Note:theestimatesarebasedonEq.(5)inthetext.Numbersinparenthesesarestandarderrors.***,**and*indicatesigni?canceatthe1,5and10%level,respectively.Z.Qiaoetal./Int.Fin.Markets,Inst.andMoney18
(2008)
425–437
435
436Z.Qiaoetal./Int.Fin.Markets,Inst.andMoney18(2008)425–437
anyoftheA-sharemarketsandtheHKorB-sharemarketsdecrease.However,thecorrelationsbetweenanyoftheB-sharemarketsandtheHKmarketincrease.
Inlightoftheseobservations,weexaminethetime-varyingcorrelationcoef?cientsinresponsetounusualmarketconditions,suchasa?nancialcrisisandchangesinregulation.Expressingthisnotioninaregressionmodel,wewrite:
ρij,t=d0+d1crisis1t+d2crisis2t+d3FPt+εt(5)whereρij,taretheconditionalcorrelationcoef?cientsbetweenMarketsiandj;crisis1tandcrisis2taredummyvariables,denotingtheearlystage(2July1997–15October1997)andtheeffectivestage(22October1997–28December1998)oftheAsian?nancialcrisis,respectively,9andFPtisadummytocapturetheimpactoftheremovaloftherestrictiononinvestmentinBshares(22February2001–28December2005).Thedummyvariablesaresettounitytoindicatethepresenceofaneffectandarezerootherwise.
Theestimatedcoef?cientsforEq.(5)arereportedinTable5.TheevidenceshowsthattheindicatoroftheeffectoftheAsian?nancialcrisis,d1,ispositiveforthesixpairsofmarkets.Thus,intheearlystageofthecrisis,2July1997–15October1997,theeffectwasnotfullyhittingthesesixmarkets.Theytendtobecorrelatedwitheachotheratabithigherlevel.However,fromHKandChineseinvestors’pointofview,thecrisistookeffecton20October1997.Thisledtothenegativeandhighlysigni?cantd2forHK–SHA,SHB–SHA,HK–SZA,andSZB–SZA.Incontrast,thecontagioneffectspreadthecrisistotheHKandtwoB-sharemarkets,whichledtoherding,asevidencedbyapositivecorrelationandsigni?cantd2ontheSHB-HKpair.TheevidencesuggeststhatthemarketsegmentationpolicyimposedbytheChineseauthorityisaneffectiveinstrumentforshieldingtheA-sharemarketsfromexternalturbulence.
Thecoef?cientoftheFPtvariable,d3,issigni?cantlypositiveforSHB–SHAandSZB–SZA,indicatingthatthecorrelationsbetweenA-andB-sharemarketsincreasedafterdomesticinvestorsinA-sharemarketswereallowedtopurchaseBshares.Interestingly,we?ndthatd3forHK–SHAandHK–SZAisalsopositiveandhighlysigni?cant,suggestingthatevendomesticinvestorsinA-sharemarketsarestillnotallowedtoinvestintheHKmarketandtheirparticipationinB-sharemarketstendstostimulateactivetransmissionofinformationbetweentheA-shareandtheHKmarkets.WeconcludethatthismorerelaxedpolicyonpurchasingBshareshelpedtoacceleratethemarketintegrationprocessofA-sharemarketswithinternational?nancialmarkets.Incontrast,we?ndthatd3forSHB-HKisnegativeandhighlysigni?cant,suggestingthattheparticipationofdomesticcitizensinSHBislessef?cientintransmittinginformationbetweentheSHBandtheHKmarketand,consequently,reducesthecorrelationbetweenthesetwomarkets.
5.Conclusions
Inthisstudy,weapplyarelativelynovelFIVECM-BEKKGARCHframeworktoexaminethelong-termequilibrium,short-termadjustment,andspillovereffectsamongsixpairsofstockmarkets,namely,HK–SHA,SHB–SHA,SHB–HK,HK–SZA,SZB–SZA,andSZB–HK.Our
AlthoughthecrisisoriginatedinThailandandthemarketdeclinedsharplyinJune1997,followedbythecollapseoftheIndonesianmarketinAugust,noseriousattentionwasgiventothesemarketsuntilthecrisishittheHongKongmarketinmid-October(betweenOctober20andOctober23theHangSengIndexdippedby23%).FromtheperspectiveofChinesestockinvestors,theHongKongmarketcrashinmid-Octoberwasadirectthreattotheirinvestments,sincetheportfolioperformanceintheB-sharemarketsisperceivedtobehighlycorrelatedwiththatofHongKong’smarketandShenzhenBsharesaremeasuredinHKdollars.9
Z.Qiaoetal./Int.Fin.Markets,Inst.andMoney18(2008)425–437437
FIVECMapproachisconsideredtobemoregeneralthanthetraditionalVECMapproach,sinceitcanmeasuretheeffectofthelongmemoryonthecointegratingrelationship,whichisimportantforrevealingthetruerelationshipsbetweentherelevantstockmarkets.Furthermore,augmentingtheFIVECMwithabivariateBEKKGARCHformulation,weinvestigatethemeanandvolatilityspillovereffectsacrossthesemarketssimultaneously.
Ourequilibriumanalysesindicatethatallsixpairsofmarketsarefractionallycointegrated.TheHongKongstockmarketadjuststoreturntoequilibriumwiththetwoA-sharemarkets,whilethetwoB-sharemarketsadjusttoreturntoequilibriumwiththecorrespondingtwoA-sharemarketsandtheHongKongmarket.Thevolatilityspillovereffectshowsthattherearebi-directionalvolatilityspilloversbetweenthetwoA-sharemarketsandthetwoB-sharemarketsandbetweenthetwoB-sharemarketsandtheHongKongmarket.However,onlyunidirectionalvolatilityspillovereffectsfromthetwoA-sharemarketstotheHongKongmarketarepresent.Amongthealternativemarkets,we?ndthatthetwoA-sharemarketsaremostin?uentialinbothmeanandvolatilityspillovereffects.Furtherinvestigationofthedynamicpathofcorrelationcoef?cientssuggeststhatrelaxationofgovernmentrestrictionsonthepurchaseofBsharesbydomesticresidentsincreasesthecorrelationbetweenthetwoA-andthetwoB-sharemarkets,indicatingthatthedegreeofsegmentationhasbeenmoderatedandthatthetwoclassesofmarketshavetendedtograduallymerge.EvidencealsoshowsthatthisliberalpolicyacceleratedthemarketintegrationprocessoftheA-sharemarketswiththeHongKongstockmarket.Finally,we?ndthattheeffectsoftheAsiancrisisonthestock-returndynamiccorrelationsvaryacrossthesemarkets.References
Baillie,R.,1996.Longmemoryprocessesandfractionalintegrationineconometrics.JournalofEconometrics73,5–59.Cai,J.,Li,Y.,Qi,Y.,2006.Theday-of-weekeffect:newevidencefromtheChinesestockmarket.ChineseEconomy39,
71–88.
Chakravarty,S.,Sarkar,A.,Wu,L.,1998.Informationasymmetry,marketsegmentationandthepricingofcross-listed
shares:theoryandevidencefromChineseAandBshares.JournalofInternationalFinancialMarkets,InstitutionsandMoney8,325–356.
Cheung,Y.,Lai,K.,1995.Asearchforlongmemoryininternationalstockmarketreturns.JournalofInternationalMoney
andFinance14,597–615.
Chiang,T.C.,Tan,L.,Li,H.,inpress.Empiricalanalysisofdynamiccorrelationsofstockreturns:evidencefromChinese
A-shareandB-sharemarkets.QuantitativeFinance.
Ding,Z.,Granger,C.W.J.,Engle,R.F.,1993.Alongmemorypropertyofstockmarketreturnsandanewmodel.Journal
ofEmpiricalFinance1,82–106.
Engle,R.,Granger,C.,1987.Cointegrationanderrorcorrection:representation,estimationandtesting.Econometrica55,
251–276.
Engle,R.F.,Kroner,K.F.,1995.MultivariatesimultaneousgeneralizedARCH.EconometricTheory11,122–150.Kim,Y.,Shin,J.,2000.InteractionsamongChina-relatedstocks.Asian-Paci?cFinancialMarkets7,97–115.
Laurence,M.,Cai,F.,Qian,S.,1997.Weak-formef?ciencyandcausalitytestsinChinesestockmarkets.Multinational
FinanceJournal4,291–307.
Li,Y.,Yan,D.,Greco,J.,2006.MarketsegmentationandpricedifferentialsbetweenAsharesandHsharesintheChinese
stockmarkets.JournalofMultinationalFinancialManagement16,232–248.
Lo,A.W.,1991.Long-termmemoryinstockmarketprices.Econometrica59,1279–1313.
Saikkonen,P.,1991.Asymptoticallyef?cientestimationofcointegratingregressions.EconometricTheory7,1–21.Stock,J.H.,Watson,M.W.,1993.Asimpleestimatorofcointegratingvectorsinhigherorderintegratedsystems.Econo-
metrica61,783–820.
Wang,S.S.,Firth,M.,2004.Dobearsandbullsswimacrossoceans?Marketinformationtransmissionbetweengreater
Chinaandtherestoftheworld,JournalofInternationalFinancialMarkets.InstitutionsandMoney14,235–254.Yeh,Y.H.,Lee,T.,Pen,J.,2002.Stockreturnsandvolatilityundermarketsegmentation:thecaseofChineseAandB
shares.ReviewofQuantitativeFinanceandAccounting18,239–257.