ContentslistsavailableatScienceDirect
JournalofInternationalFinancialMarkets,Institutions&Money
journalhomepage:www.elsevier.com/locate/intfin
Doesstockmarketliquidityexplainrealeconomicactivity?NewevidencefromtwolargeEuropeanstockmarkets
NicholasApergisa,PanagiotisG.Artikisb,∗,DimitriosKyriazisc
abc
BusinessSchool,NorthumbriaUniversity,NewcastleuponTyneNE18ST,UK
DepartmentofBusinessAdministration,UniversityofPiraeus,80Karaoli&DimitriouStreet,18534Piraeus,Greece
DepartmentofBankingandFinancialManagement,UniversityofPiraeus,80Karaoli&DimitriouStreet,18534Piraeus,Greece
article
info
abstract
Articlehistory:
Received18December2013Accepted14May2015
Availableonline19May2015
Keywords:
StockmarketliquidityEconomicconditionsUKmarket
Germanymarket
Thispaperexaminestherelationshipbetweenstockmarketliquidity,whichproxiesfortheimplicitcostoftradingshares,withmacroeconomicconditions.WeprovideevidencethatstockmarketliquiditycontainsstrongandrobustinformationabouttheconditionoftheeconomyforboththeUKandGermanyinthepresenceofwell-establishedleadingindicators.Ourfindingsexemplifytheimportanceofsmallcapfirms’liquidityinexplainingthestateoftheeconomyandsupportthe“flight-to-qualityargument”.Finally,theempiricalfindingsshowthatthereisnotanydifferentialroleofliquidityinexplainingthecourseofmacroeconomicvariablesbetweenacapitalmarketandabank-orientedeconomy.
©2015ElsevierB.V.Allrightsreserved.
1.Introduction
Theexistenceofanilliquidityriskpremiumiswelldocumentedintheliterature,inthesensethatilliquidstockscommandhigherexpectedreturnsthanliquidstocks(e.g.,AmihudandMendelson,1986;Amihud,2002;Chordiaetal.,2005;KempfandMayston,2008;PastorandStambaugh,2003;AcharyaandPedersen,2005;Papavassiliou,2013).Theliquidityshockhypothesisarguesthatsuddendropsinassetmarketsliquiditycauseequitypricestofallandthepriceofliquidassetstorise(KiyotakiandMoore,2008).Moreover,inaworldwherefirmshavetocopewithfinancingconstraintsontheirinvestments,thisfallinequitypricesreducesthefundsforinvestmentsafirmcanraisebyissuingequityand/orusingequityascollateralinborrowing.Asaresult,investmentsfall,outputfollowsandarecessionstarts.Theliquidityshockhypothesishasreceivedwideattentionbecauseofitsimmediatepolicyimplications.Ifunexpectedfluctuationsinequityliquidityarethecauseofeconomicgrowth,thenagovernmentcanattenuatetheeconomicperformancebymakingthesupplyofliquidassetscountercyclical.Attheonsetofarecession,agovernmentcanuseliquidassetstobuyupsomeoftheilliquidequitytopreventequitypricesfromfallingprecipitously.Theincreaseinthesupplyofliquidassetsrelaxesfirms’financingconstraints,whilethestabilizationofequitypricesfurtherimprovesfirms’abilitytousetheequitymarkettofinancetheirinvestmentprojectswithlowercostofcapital,thus,increasingthereturnontheprojectstheyadopt.ThesepolicyimplicationsseemtoprovideajustificationforthelargeandrepeatedinjectionsofliquiditybytheUSFederalReserveSystemaswellasothercentralbanksovertherecessionaryperiod2008–2009.
Thegoalofthisstudyistoinvestigatetheinformationcontentofstockmarketliquidity,basedonfirm-leveldata,toexplainthecourseofeconomicactivity,aftercontrollingforanumberofequity(i.e.,marketriskpremium,stockmarket
∗Correspondingauthor.Tel.:+302104142200.
E-mailaddresses:Nicholas.apergis@northumbria.ac.uk(N.Apergis),partikis@unipi.gr(P.G.Artikis),dkyr@unipi.gr(D.Kyriazis).
http://dx.doi.org/10.1016/j.intfin.2015.05.002
1042-4431/©2015ElsevierB.V.Allrightsreserved.
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
43
volatility)andnon-equity(i.e.,housingstarts,termspread,short-terminterestrates,defaultspread)factors.Indoingso,weapplyalternativeliquidityproxiestodifferentindicatorsofeconomicactivity,whileweutilizeasampleofstocksoriginatingfromtwoofthelargestEuropeanstockmarkets,i.e.theLondonStockExchangeandtheDeutscheBörse,spanningtheperiod1994to2011and1997to2011,respectively.
Therationaleforexaminingwhetherstockmarketliquiditycanactasaleadingindicatorforeconomicactivityisthreefold.First,accordingtothe“flighttoquality”hypothesisputforwardbyLongstaff(2004),investorstendtoshifttheirportfoliostomoreliquidsecuritiesinturbulenttimesofeconomicactivity.Second,liquiditycanaffecteconomicactivitythroughcertaininvestmentchannels,sincealiquidsecondarymarketmayfacilitateinvestmentsinproductivelong-runprojects(Levine,1991).Third,BrunnermeierandPedersen(2009)showthatduringperiodsofeconomicdownturn,bothalackofassets’marketsliquidityandreducedfinancialintermediaries’fundingliquidityleadtoliquidityspirals.
TherelationshipbetweenstockmarketliquidityandeconomicactivityhasattractedlimitedattentionintheliteratureandcertainstudieshavefocusedeitheronUSdataoronsmallmarkets,suchasNorwayandSwitzerland.Beberetal.(2011)findthatanorderflowportfolio,basedoncross-sectormovements,canpredictthestateofthemacroeconomy.Inasimilarstudy,KaulandKayacetin(2009)showthattwoalternativeorderflowmeasurescanpredictGDPandindustrialproductiongrowth.Næsetal.(2011)usealternativeliquiditymeasures,bothfortheUSandNorway,anddocumentthatstockmarketliquiditycanserveasaleadingindicatorforthemacroeconomicvariables.Meichleetal.(2011)findthatstockmarketliquidityisthemainpredictorforeconomicactivityforSwitzerlandovertheperiod1990–2010.Morerecently,Florackisetal.(2014a)findthatstockmarketilliquiditycanbetterexplainandforecastthefutureUKGDPgrowththananyothervariableusuallyexamined(i.e.,termspread,short-terminterestratesandrealmoneysupply)andconfirmastatisticallysignificantnegativeassociationbetweenthesetwovariables.
Takingintoaccountthattheassociationbetweenstockmarketliquidityandmacrovariableshasattractedlimitedinterestintheliterature,furtherevidenceisneededintermsofmarketselection,methodologicalapproachesandthesampleperiod,inordertofullyunderstandthisassociation.Thepresentstudycontributestotheliteraturetowardsthisendinanumberofways.ItisclearfromtheabovediscussionthattherelationshipbetweenstockmarketliquidityandmacrovariableshasbeenexaminedmainlyinaUSsetting.Thus,weshedfurtherlightintheliteraturewiththeuse,forthefirsttime,ofdatafromtwolargeEuropeanstockmarkets,theUKandGermany.TheLondonStockExchange(LSE)andtheGermanstockexchange(DeutscheBörse)areselectedonthegroundsthatalthoughtheyaremajormarketsofgreatinternationalimportanceandinterest,rankingamongtheworld’slargestintermsofnumberoffirmslistedandtotalmarketcapitalization,theyhavealargerliquidityeffectandhavenotbeencross-examinedinthepreviousempiricalliterature.
Anothersignificantnoveltyofthepresentpaperisthatforthefirsttimeweprovideaninterestingcomparisonoftheinformationcontentofstockmarketliquidityforeconomicactivitybetweenacapitalmarketorientedeconomy(UK)andabankingorientedeconomy(Germany).Ithasbeenargued1thatthetypeofthefinancialsystem(i.e.,marketvs.bankbased)influenceseconomicgrowth,whileanumberofempiricalworksshowthatthedistinctionisirrelevant,atleastforthecaseofdevelopedandmaturemarkets(BeckandLevine,2002).Theissueexaminedinourstudyiswhetherweshouldexpectthatstockmarketliquiditycouldbehavedifferentlyinabank-basedsystem,suchasinGermany,thaninamarket-basedsystem,suchasintheUK,basedonthefactthatliquidityisexplicitlyusedasthemainexplanatoryvariableofthemacroeconomicenvironment.Stockmarketsprovidedirectfundingtoinvestors,whilebanksandotherfinancialinstitutions,asintermediaries,provideindirectfundingtothem.Therefore,wecouldarguethatstockmarketsprovideaneasierandquickertransmissionofliquiditytoinvestorsandtotherealeconomythanbankswhentheeconomyisthriving,butanequallyfasternegativeadjustmentofliquiditywhentheeconomyisplungingintorecession.
Intermsofmethodologicalapproaches,thepresentstudydifferentiatesfrompreviousworksintheareabyexaminingalternativeliquidityproxies.Tothisend,thepapermakesuseofalternativesdefinitionsofliquidityaswellastheInstru-mentalVariable(IV)methodologicalapproach,whichtakescaresofanyendogeneitybiasproblems.Thestudyfocusesonthesimplernon-sophisticatedliquidityproxies,which,however,aretheonesusedbypractitionersandinvestmentprofes-sionalsthatdonotrequirerestrictiveassumptionsasthemoresophisticatedproxiesdo.Moreover,thepresentstudydiffersfromthosebyNæsetal.(2011),Meichleetal.(2011)andFlorackisetal.(2014a)byexaminingtheinformationcontentoftwoliquiditymeasures,namely,theturnoverandthevolumeoftrading,toexplaineconomicactivityalongwiththerelativespreadandAmihud’silliquidityratiowhichhavebothbeenpreviouslyexamined.Weopttousedifferentliquidityproxiesinordertofullyexamineonhowvariousaspectsofliquidityaffecteconomicconditionsandtoproviderobustnesstoourresults.
Stiglitz(1985)andBhide(1993)claimthatstockmarketsdonotproducethesameimprovementinresourceallocationandcorporategovernanceasbanks.Thosewhofavourthemarket-basedsystemargueagainsttheroleofbanksforextractinginformationalrentsfromfirmsandreducingincentivestoundertakeriskyandinnovativebutprofitableprojects(e.g.,Rajan,1992;MorckandNakamura,1999).LaPortaetal.(2002)alsoargueagainsttheroleofstate-ownedbanksforhavingpoliticalgoalsintheprocessofsupplyingcredittorathertraditionallabourintensiveindustries,thantoinnovativeandtrulystrategicones.However,BootandThakor(1997)showthatbanksfacilitatebetterthegoalofeconomicgrowthinemergingfinancialsystemsandstockmarketsdobetterinmaturefinancialsystems.Inaddition,AllenandGale(2000)presentevidencethatbothbanksandmarketsprovidedifferentfinancialservices,whileeconomiesatdifferentstagesofeconomicdevelopmentrequiredifferentmixturesoffinancialservicestooperateeffectively.AsimilarfindingisprovidedbyTadesse(2002).BeckandLevine(2002)donotfindanyevidencethatthetypeoffinancialstructurereallymattersforindustrygrowthandtheefficientallocationofcapitalacrossindustries.
1
44
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
Forrobustnesspurposes,wehavealsoincludedforthefirsttimeanadditionalcontrolvariable,i.e.,housingstarts,apartfromthetermspread,marketriskpremium,stockmarketvolatility,short-terminterestrateanddefaultspread,alreadyusedinpreviousstudies.Theimportanceofhousingstartsinexplainingmacroeconomicconditionshasbeendocumentedextensivelyintheliterature(Green,1997;CoulsonandKim,2000;HuiandYiu,2003;Iacoviello,2003).Ahousingstartisgenerallycountedastheexcavationofthefoundationandindicatesadvancedemandinthehousingsector.Housingstarts,ontopofbeingaleadingindicatorofstrengthintheconstructionindustry,arealsoanimportantleadingeconomicindicator,duetotheirextensivespilloverbenefitstotheothersectors(i.e.,retail,manufacturing,utilities,labourmarkets),sincenewhomesneedtobeequippedandfurnishedfromscratch(Karamujic,2013).
Finally,asfarasthesampleperiodisconcerned,thepresentstudyisimplementedinaquiteuniqueandinterestingtimeframe,i.e.,1994–2011,sinceitcoversbothperiodsof“bull”and“bear”equitymarketsandperiodsofeconomicexpansion(1994–2002,2004–2008,2010–2011)andeconomicdownturn(2002–2004,2008–2010).Inparticular,thesecondperiodofrecessionwasquitesevere,followingtheburstoftherealestatebubbleintheUS,whichtriggeredtheglobalfinancialcrisisthatdrainedliquidityfromfinancialmarketsworldwide.Thus,ourempiricalmodelsaretestedacrossanumberofdifferenteconomicandstockmarketbackgroundsandtheimplicationsofourresultsmaybeofparticularinterestnotonlyforacademics,butalsoforinvestors(i.e.,retailandinstitutional),policymakersandregulators.
Toforeshadowtheresults,theyshowthatthereisastrongrelationshipbetweenstockmarketliquidityandthestateofthemacroeconomyinbothcountriesunderinvestigation.Whenthereisdrainageofstockmarketliquidityandtheimplicitcostsfortradingstocksincrease,theninvestorsshouldbeanticipatinglowerGDP,investmentsandconsumptionandhigherunemploymentrates.Finally,theliquidityofsmallstockcompaniesisfoundtohavealargerimpactonthemacrovariablesinvestigatedacrossbothcountries.ThisimportantresultcorroboratesthefindingsofAmihud(2002)andmorerecentlyofCakiciandTan(2014)andmayhaveseriousimplicationsforinvestorsandcentralbanks’policies,sincealargedropintheliquidityofsmallfirmsstocksgivesastrongsignalforthebeginningofarecessionaryperiod.Asinvestorsstartswitchingfromtheirpositionsonsmallcapstockstogovernmentbondsorlargecapstocks,centralbanksmayincreasepromptlythemoneysupplyaimingtostimulatetherealeconomy,avoidingplunginginadeepandprolongedrecession.
Theempiricalfindingsareexpectedtoshedfurtherlightontheroleofmarketliquidityinthegrowthprocess,especially,duringturbulentperiodsliketherecentrecession,sinceliquidityiscloselyassociatedwithboththemarketliquidityriskandthefundingliquidityrisk.Thefirsttypeofriskoccursasthemarketliquidityworsensandpotentialinvestorsneedtotrade,whilethesecondtypeistheriskwheretraderscannotfundtheirpositionsandareforcedtounwind.Thisperversesituationishavingsignificanteffectsontherealeconomy.
However,systematic(market)liquiditycanhaveseriousrepercussionsnotonlyforthefinancialsystem,butalsofortherealeconomy,sinceanydisruptionscanleadtofinancialcrises,whichdamagefinancialstability,resourcesallocationandhaveanegativeimpactontherealeconomy(Fergusonetal.,2007).Therefore,thepresenceofthisdownwardliquidityspiralrecommendsthatpolicymakershavetoimprovethefundingliquidityofinvestorsinthemarket,especiallythatofbankinginstitutions.
Therestofthepaperisorganizedasfollows.Section2presentsapertinentliteraturereview,whileSection3definestheliquidityproxiesalongwithanumberofothercontrolvariablesanddescribesthemethodologyemployed.Section4presentsanddiscussestheresultsobtainedfromtheempiricalanalysis.Finally,Section5concludesthepaper.
2.Literaturereview
Academicresearchhasextensivelyexaminedtherelationshipbetweenassetprices(e.g.,interestrates,termspreads,stockreturns,andexchangerates)andrealeconomicindicators.EstrellaandHardouvelis(1991)showthattheyieldcurvecanpredictfuturedevelopmentsinrealeconomicactivity,whileothershighlighttheroleofthetermspreads(i.e.,thedifferencebetweena10-yeargovernmentbondandanuncoveredshort-terminterestrate)inpredictingfutureturningpointsintheeconomy(EstrellaandMishkin,1998;RudebuschandWilliams,2009;Wright,2006).Therationaleofusingfinancialmarketvariablesasleadingindicatorsforeconomicactivityisthreefold:(a)investorsconveytheinformationaboutthefuturestateoftheeconomybytradingfinancialsecuritiesandchangingtheirrelativepriceovertime,basedonnewinformationarrivals(Beberetal.,2011),(b)theyareobservedandnotestimatedthroughatheoreticalmodel,and(c)theyareinstantlyandeasilyavailableatahighfrequencytoallmarketparticipantsandanalysts(Meichleetal.,2011).However,StockandWatson(2003),bycarryingouttheirempiricalstudyforsevenOECDcountriesspanningtheperiod1959–1999,concludethatthelinkbetweenfinancialmarketvariablesandrealeconomicactivityisnotuniformandstableacrossallcountriesandperiods.
Assetliquidity,whichistheabilitytosellaninvestmentinstantlyandatapriceclosetoitscurrentmarketprice,canbethoughtofasthechannelthroughwhichinformationaboutmacroeconomicvariablesisincorporatedintoassetprices.Anumberofresearchstudiesintheliteratureofmarketmicrostructurehavedocumentedapositivelinkbetweenasecurity’silliquidityanditsexpectedreturns,whichestablishesthepresenceofanilliquidityriskfactorandtheassociatedilliquid-ityriskpremium(AmihudandMendelson,1986;Amihud,2002;Jones,2002;PastorandStambaugh,2003;AcharyaandPedersen,2005;Guoetal.,2011).
Takingintoaccountononehandthatassetpricescanforecastrealeconomicindicatorsandontheotherthatassetliquiditycanexplainassetpriceschanges,thiscouldimplythatassetliquiditycontainsincrementalinformationaboutmacroeconomicconditions.Onepossibleexplanationcouldbetheroleofthe“flighttoliquidity”or“flighttoquality”,
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
45
putforwardbyLongstaff(2004)whoshowsthatinvestorsprefertoinvestinUSTreasurybondswhicharemoreliquidincomparisonwithRefcorp2bonds,althoughbothofthemcarrythesamecreditrisk.Infact,Longstaff(2004)discoversthatabout10%to15%ofT-Billpricescanbeattributedtotheirlargeliquiditypremia.LevineandZervos(1998)provideanexplanationbyshowinghowstockmarketliquidityaffectsrealeconomicactivityviacertaininvestmentchannels.Theyempiricallyestablishastatisticallysignificantpositiverelationshipbetweenstockmarketliquidityandcurrentandfutureratesofeconomicgrowthinseveralcountries,aftercontrollingforpoliticalandeconomicfactors.AdifferentexplanationisprovidedbyBrunnermeierandPedersen(2009)whodevelopamodelthatdescribesamutuallydependentrelationshipbetweenassetsmarketliquidityandfinancialintermediaries’fundingliquidity.Inparticular,themodelexplains,amongotherthings,howareducedfundingliquidityinperiodsoffinancialdownturnscanleadtoa“flighttoquality”,i.e.,thatfinancialintermediarieschangetheirliquidityprovisiontostockswithlowmarginrequirements.AsimilarpathhasbeenfollowedbyRöschandKaserer(2012)whoprovideevidenceconsistentwiththetheoreticalmodelofBrunnermeierandPedersen(2009).
This“flighttoquality”argumentmayalsobeassociatedwiththesizeoffirmsintermsofmarketvalue.Smallcapitalizationfirmssuffermorethanlargecapitalizationfirmsduringaneconomicdownturn,whiletheyprospermorewhentheeconomyisexpanding(Perez-QuirosandTimmermann,2000;Switzer,2010).Inaddition,smallcapstocksareusuallylessliquidthanlargecapstocks.Chordiaetal.(2004)witnessanincreaseinaggregatemarketliquidityovertime,whichhasbeenmorepronouncedforlargethanforsmallfirms.Duringarecession,investorsmoveoutmoreheavilyfromsmallcapstocksthatperformpoorlyandarelessliquidthanfromlargecapstocks.Chordiaetal.(2004)alsoestablishthataveragedailychangesinliquidityexertaheterogeneouseffectonstockreturns,dependingonthefirmsize,sincetheliquidityofsmallfirmsvariesmoreonadailybasisthanthatoflargefirms.Thus,theliquidityvariationofsmallcapstocksislargerthanthevariationoflargecapstocks.Næsetal.(2011)document,usingUSdata,thattheliquidityofsmallcapfirmsismoreinformativeaboutfuturemacrofundamentalsthantheliquidityoflargecapfirms.Theyattributethiseffecttothetendencyofinvestorstomoveoutfirstfromsmallcapstocks,eitherbecauseofchangingexpectationsabouteconomicenvironmentorduetoincreasedfundingliquidityconstraints.Theyconfirmthisbyexhibitingamuchlargerdropintradingvolumeofsmallcapstocksbeforearecession,thanthatobservedforlargefirmstocks.Inamorerecentstudy,CakiciandTan(2014),investigatingtheeffectofvalue3andmomentumvariablesin23developedinternationalmarkets,findthatvaluestockreturnsarelowerpriortoarecessionandthisisratherduetoadeteriorationoffundingliquidityandnotduetopoormarketliquidity.
Althoughtheliteratureprovidesanumberofdifferentexplanationsofwhystockmarketliquidityshouldberelatedtoeconomicgrowth,therelationshippersebetweenaggregatemarketliquidityandthefutureeconomicconditionshasattractedlessinterestfromacademicresearchers.First,anumberofstudies,closelyrelatedtoourwork,examinewhethermacroeconomicfactorsaffectstockmarketliquidity.Morespecifically,Fujimoto(2003)andSöderberg(2008)examinethein-sampleandout-ofsampleforecastingabilityofvariousmacroeconomicvariablesonliquidity,butdonotconsiderthatthisrelationshipcangointheoppositeway.LuandGlascock(2010)showthatthepricingcomponentofliquiditycanbesignificantlyaffectedbyanumberofmacroeconomicfactorsand,inparticular,thegrowthinindustrialproductionandwhentheeconomydigsdeeperintorecession.Inasomewhatdifferentapproach,GibsonandMougeot(2004)documentthatatime-varyingliquidityriskpremiumintheUSstockmarketcanbelinkedwitharecessionindex.
Inanotherstrandoftheliterature,therearestudiesthatexamineequity-marketorderflows,whicharecloselyrelatedtostockmarketliquidity.Specifically,KaulandKayacetin(2009)examinetheinformationcontentoftwodifferentmeasuresofaggregateequity-marketorderflowsforfuturemacroeconomicfundamentalsandexpectedstockmarketreturns.TheydiscoverthatbothcanpredictfuturegrowthratesofindustrialproductionandrealGDP,uptofourquartersahead,andthisresultisrobustevenaftercontrollingforvariablesassociatedwithcommonequitypricingfactors.Beberetal.(2011)addresshowtheissueoforderflowsmovementsbyinvestorsacrossequitysectorsisrelatedtocurrentandfutureeconomicconditions.Theirresultsshowthatlarge-sizedactiveorderflowsinthematerialssectorcanforecastanexpandingeconomy,whilelarge-sizedactiveorderflowsintoconsumerdiscretionary,financials,andtelecommunicationsforecastacontractingeconomy.
Inthelightoftherecentglobalfinancialcrisis,thereducedfundingliquidityandthedecreasedorderflowmovementsfrommarketmakersareaddressedbythreepapersthatdirectlyexploretheissueofstockmarketliquidityandtherealeconomy.Næsetal.(2011)employUSandNorwegianstockmarketdataanddisplaythatstockmarketliquidity(intermsofthetradingcostsofequities)canbeusedasapowerful“leadingindicator”oftherealeconomy,evenaftercontrollingforthepresenceofothervariables,whichareextensivelyusedinpreviousrelevantempiricalstudiesforpredictingbusinesscycles.TheirstudymakesuseofalargeUSdatasetovertheperiod1947–2008alongwithauniquedatasetforNorwayspanningtheperiod1990–2006.TheauthorsalsodiscoveranimportantrelationshipbetweenthesizeofthefirmsandtheinformationcontentofliquidityinpredictingGDPgrowth,afindingthatisconsistentwiththe“flight-to-quality”effect.
Meichleetal.(2011)claimthatforasmallopeneconomy,i.e.,theSwisseconomy,assetpricefactors,suchastermspreads,arenotentirelyappropriatetopredicteconomicgrowth,aslongastheyareseverelyaffectedbyexogenousfactors
RefcorpisagovernmentagencycreatedbytheFinancialInstitutionsReform,Recovery,andEnforcementActof1989(FIRREA).TheprincipalofthesebondsisfullycollateralizedbyTreasurybonds,whilefullpaymentofcouponsisguaranteedbytheTreasuryundertheprovisionsofFIRREA.3
Ithasbeenwellestablished(e.g.,FamaandFrench,1992,1996;Lakonishoketal.,1994)thatvaluestockswhicharedescribedasstockswithhighratiostofundamentals(e.g.,bookvalue,earningspershare,cashflows,etc)relativetostockpricetendtooverperformstockswithcorrespondinglylowratios.
2
46
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
(i.e.,co-movementsofinternationallong-terminterestrates).Theyalsofindthatoverthelasttwodecades(1990–2010)stockmarketliquidityisabetterpredictorofeconomicactivitythantermspreads.However,thispictureisreversediftheentiresampleperiod(1975–2010)isconsidered,withtermspreadsgainingpredictivepower(RudebuschandWilliams,2009).ThisfindingalsoconfirmstheresultsbyStockandWatson(2003)abouttheerraticandtimevaryingbehaviourofassetpricesaspredictorsofrealeconomicactivity.
Finally,Florackisetal.(2014a)inastudy,whichfocusesonlyontheUKmarketandconsidersonlymacroeconomicactivityintermsofGDPgrowth,examinetheexplanatorypowerofstockmarketliquidityinforecastingtherealUK.GDPgrowthovertheperiod1989–2012.Byusingstandardlinearandnonlinearmodels,theyfindastatisticallysignificantnegativerelationshipbetweenstockmarketilliquidityandfuturegrowthinGDPofUK,evenafterincludingtheusualexplanatoryvariables(e.g.,termspreads,short-terminterestratesandrealmoneysupply/divisia).Theyalsoshowthatthattheeffectofbothmarketilliquidityanddivisiamoneybecomesstrongerduringperiodsofilliquidmarketconditionsandpooreconomicgrowth.Furthermore,throughanout-of-sampleforecastinganalysistheydiscoverthataregimeswitchingmodelofilliquidvs.liquidmarketconditionspredictsUKgrowthinGDPbetterthananyothermodel,eventheonepublishedbytheBankofEngland’sinflationreport.
3.Methodology
3.1.Liquidityproxiesandsamplefirms
Asfarastheliquiditymeasuresareconcerned,therearenumerousindicatorsdevelopedintheliteraturethatattempttomeasurestockmarketliquidity.Thehighfrequencyliquiditymeasuresrequireintradaydataonbid/askquotes,orderflows,volumeoftradesetc.,whicharenotavailableforalongperiodoftime.Thus,weadoptlowfrequencyliquiditymeasuresthatcanbeestimatedwithdailydata,whichareavailableforlongertimeperiods.Furthermore,sinceliquidityisanunobservablecharacteristicofanassetmarket,whichcannotbecapturedinasinglemeasure,itisdesirabletoexaminetheissuewiththeuseofavarietyofliquiditymeasures.
Weusefouralternativeliquiditymeasures:(a)theAmihud(2002)illiquidityratio(ILR),(b)therelativespread(RS),(c)turnover(TUR)and(d)thevolumeoftrading(VTR).AccordingtoGoyenkoandUkhov(2009)andGoyenkoetal.(2009),thefirsttwoliquidityproxiescapturethespreadcostandthepriceimpactwhenestimatedwithdailydata.Therationaleforusingturnoverandvolumeoftradingistwofold.First,theyaresimpleandstraightforwardtocalculateanddonotrequirealargeamountofdataorrestrictiveassumptionsasthemoresophisticatedproxies,suchastheLesmondetal.(1999)andRoll(1984)liquiditymeasures.Second,theyarethemostcommonlyusedliquiditymeasuresbypractitionersandinvestmentprofessionalsandhavebeenpreviouslyusedinthepertinentliteratureinotheraspectsofliquidity,suchasinassetpricing.Amihud’s(2002)illiquidityratio(ILR)istheratioofabsolutestockreturnstomonetaryvolumeonadailybasis,displayinghowmuchpricesmoveforeachmonetaryunitoftrades.Thecostassociatedwithlargertradesismoreaccuratelycapturedinthepriceimpactofatrade.Hasbrouck(2009)showsthattheILRisthebestavailableprice-impactproxyconstructedfromdailydata.Moreover,Amihud(2002)showsthattheILRispositivelyandsignificantlyrelatedtoboththepriceimpactandthefixedcostcomponentestimatesdefinedbyBrennanandSubrahmanyam(1996).TheILRcapturesthesensitivityofpricestotradingvolumes,sinceitisameasureoftheelasticitydimensionofliquidity.TheILRiscalculatedas:
ILRi,T=
TRi,t1
DT
t=1
VOLi,t
(1)
whereDTisthenumberofobservationswithinatimewindowT,|Ri,t|istheabsolutereturnatdaytforstocki,andVOLi,tis
thetradingvolumeinmonetaryvaluesatdaytforstocki.TheILRessentiallyprovidesanilliquiditymeasure,sinceahighvalueindicateslowliquidity(i.e.,ahighpriceimpactoftrades).Moreover,ahighpriceimpactsuggeststhatthemarketdepthislowandasmallervolumeisneededtomovethatprice.
Amarketparticipantwhowishestofillhisorderimmediatelymustbewillingtopaytheaskpriceforabuyorderandcollectthebidpriceforasellorder.Thedifferencebetweenthetwopricesisthebid-askspread,whichreflectsthecostofimmediacy.Thus,thebid-askisaspreadcost,whichisobservedinbothdealerandlimitordermarkets.Inthepresentstudy,weestimateamarket-wideproportionalspreadmeasure,therelativebid/askspread(RS).Itisestimatedastheratioofthequotedspread(i.e.,thedifferencesbetweenthebestaskandbidquotes)overthemidpointprice(i.e.,theaveragesofthebestaskandbidquotes)onadailybasis:
RSi,T=
T1
ASK−PBIDPi,t
i,t
(2)
DT
t=1
ASK+PBID/2Pi,t
i,t
wherePASKi,tandPBIDi,taretheaskandbidprice,respectively,atdaytforstocki.TheRSprovidesarelativemeasureoftradingcostsandproxiesforapercentagetwo-waytransactioncost,i.e.,whatfractionofthepriceneedstobepaidto“cross”fromthebidtotheaskprice,orviceversa.SimilarlytothecaseoftheILR,theRSisanilliquiditymeasure,sinceahighspreadindicatesanilliquidmarketwheretheimplicitcostsoftradingarelarge.
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
47
Table1
Descriptivestatistics,serialautocorrelationsandcorrelationcoefficientsofliquidityproxies.
Mean
Median
Jarque–Berra
No.firms
No.Obs.
Section1:UKliquidityproxiesPanelA:descriptivestatisticsILRRSTURVTR
PanelB:serialautocorrelations
0.1389470.0656210.464037477,129,289
0.0978430.0627430.473530485,855,808
2.56(0.72)5.39(0.49)3.18(0.55)3.28(0.53)
6,4219,3457,5488,356
25,68437,38030,19233,424
Lags1
4
12
ILRRSTURVTR
PanelC:correlationcoefficients
3.41(0.43)3.19(0.48)2.30(0.32)3.07(0.51)
3.93(0.29)3.74(0.37)2.58(0.24)3.30(0.39)
4.49(0.17)4.37(0.19)4.26(0.16)5.52(0.28)
RSTURVTR
Section2:GermanyliquidityproxiesPanelA:descriptivestatisticsILRRSTURVTR
PanelB:serialautocorrelations
ILR0.2330−0.3893−0.6569
RS
TUR
−0.0375−0.5648
0.5692
0.6749500.0448760.3580196940,187
0.4655970.0434630.3684416311,968
2.29(0.77)4.51(0.58)3.62(0.39)3.35(0.49)
3,6984,5214,0893,897
14,79218,08416,35615,588
Lags1
4
12
ILRRSTURVTR
PanelC:correlationcoefficients
3.03(0.50)3.14(0.56)2.61(0.31)2.68(0.62)
3.34(0.32)3.58(0.37)2.96(0.19)3.02(0.41)
3.85(0.20)3.83(0.24)4.19(0.12)5.17(0.22)
ILR
RS
TUR
RSTURVTR
0.2956−0.3277−0.3153
−0.1470−0.4425
0.3711
PanelsA,BandCinSection1showdescriptivestatistics,serialautocorrelationsandcorrelationcoefficientsrespectivelyfortheUKliquidityproxiesfrom1994to2011.TheliquiditymeasuresaretheAmihud’silliquidityratio(ILR),therelativebid–askspread(RS),theturnover(TUR)andthevolumetraded(VTR).Theliquidityproxiesarecalculatedforeachavailablestockineachquarter.PanelAshowsthemeanandmedianoftheliquiditymeasures,theJarque–Berrastatisticwithp-valuesinparenthesis,thenumberofsecuritiesusedandthetotalnumberofobservations(eachsecurityisobservedinseveralquarters).PanelBshowstheresultsoftheQtestforserialautocorrelationforlags1,4and12.PanelCshowscorrelationcoefficientsbetweentheliquidityproxies.Thecorrelationsarecalculatedacrossallstocksandtime;theliquiditymeasuresarecalculatedforeachavailablestockineachquarterandthecorrelationsarepairwisecorrelationsbetweentheseliquiditymeasures.Section2showscorrespondingstatisticsfortheGermanliquidityproxiesfrom1999through2011.
Turnover(TUR)isacommonmeasureofactivityandiscalculatedasthetotalnumberofsharestradedduringatimeinterval,relativetothenumberofoutstandingsharesinthesecurity.Weexpressturnoverinpercentageterms,thus,measuringthepercentageoftheissuedsharesthatchangehandsduringatradingwindow(e.g.,day,month,quarter).Finally,thevolumeoftrading(VTR)inmonetarytermsiscalculatedbymultiplyingthenumberofsharestradedbytheconcurrentstockprice.BoththeTURandVTRareliquiditymeasures,asopposedtotheILRandRS,whichareilliquiditymeasures.Allfourliquidityproxiesarefirstestimatedforeachstockandforeachquarterandthentheequallyweightedcross-sectionalaverageforeachquarterisobtained.
ThesampleconsistsofallstockslistedontheLondonStockExchangeandtheDeutscheBörse.Inordertocalculatetheliquidityproxies,dataondailystocksprices,returns,thetradingvolumeandtotalnumberofshareslistedforeachcompanyinthesamplearesourcedfromBloombergandonlystockswithavailabledataareincludedintothesample.ThetimeperiodofthestudyfortheUKspansfrom31/12/1994to31/12/2011andforGermanyfrom31/3/1999to31/12/2011.Table1providessomedescriptivestatisticsaboutthefourliquidityproxies,bothfortheUKandGermanyandthecorrelationcoefficientsamongthem.TheUKmarket,asexpected,isamoreliquidmarketthantheGermanstockmarket,sincethemean/medianvaluesoftheilliquidityproxies,namelytheILRandRS,fortheUKaremuchlowerfromthecorrespondingvaluesforGermany;asimilartrendcanalsobenoticedfortheliquidityproxies,namelytheTURandVTRmetrics,exhibitingmuchhighervaluesfortheUKthantheGermanstockmarket.
48
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
TheJarque–Berastatisticconfirmsacceptanceofthenullhypothesisofnormality,whiletheQtestforserialautocorre-lationforlags1,4and12hasbeenalsoimplementedforthefouralternativeliquidityindices.TheQ-statisticteststhenullhypothesisofnoautocorrelation.TheresultsforthelevelsoffourliquidityproxiesarealsoreportedinTable1alongwiththeircorrespondingp-values.Thefindingsprovidesupportiveevidenceoftheabsenceofanyautocorrelationatallthreelagstructuresunderstudy,implyingthatallfourseriesdisplaytheabsenceofaconsistentpatternofautocorrelation.Theresultsaresupportivetotheliteraturethatfindsminimalornilautocorrelationinfinancialseries,apieceofevidenceofmarketefficiency(Abrahametal.,2002;WorthingtonandHiggs,2004;Hamidetal.,2010;amongothers).
3.2.Macroandcontrolvariables
Tomeasureeconomicactivityweusethefollowingmacrovariables:a)realGDP(RGDP),realconsumption(RC),realinvestment(RI)andtheunemploymentrate(UnR).AllseriesaresourcedfromtheOECDdatabaseonaquarterlybasis,arechainedvolumeestimatesandseasonallyadjusted.
Inordertoaccountfortheotherfinancialvariablesthathavebeenidentifiedintheliteraturethatcanexplaineconomicactivity,weuseanumberofequityandnon-equitycontrolvariablesinourmodelspecifications.Specifically,weusehousingstarts(HS),thetermspread(TS),themarketriskpremium(MRP),stockmarketvolatility(V),short-terminterestrates(SR)andthedefaultspread(DS).ThecontrolvariablesaresourcedfromBloombergandThompsonReuters.Thetermspreadiscalculatedasthedifferencebetweentheyieldofthe10-yeargovernmentbondandtheyieldofthe3-monthTreasurybill,theshort-terminterestrateisproxiedbythe3-monthTreasurybillandthedefaultspreadisthedifferencebetweentheyieldofabondindexconsistingofcorporatebondsandtheyieldofthe30-yeargovernmentbond.ThemarketriskpremiumistheexcessreturnoftheFTSE100fortheUKandtheDAXforGermanyovertheyieldoftherespective3-monthTreasurybill.Thestockmarketvolatilityismeasuredasthecrosssectionaveragestandarddeviationofdailyreturnsofthesamplestocksoverthequarter.
Housingstarts(HS)isaleadingeconomicindicatorthatreflectsthenumberofprivatelyownednewhousesonwhichconstructionhasbeenstartedoveragivenperiod.Itincludesanumberofnewsingleormulti-familyhousesasdeterminedfromthenumberofpermitsissuedforconstructionofresidentialbuildings.Realestateinvestmentsareagoodmeasureofexpecteddemandforrealestate.SinceeconomicperformanceisreflectedontoGDP,ifthereisaleadingrelationshipbetweenGDPandrealestateinvestments,therealestatesectorwillbealeadingsectorofeconomicperformance.Therefore,realestateinvestmentsaregoodmeasuresforreflectingexpectedrealestatedemandandserveasgoodpredictorsofeconomicperformance.Green(1997)makesuseofcausalityteststoexaminetheeffectofhousingstartsonGDP.Hefindsthatresidentialinvestmentscause(lead)GDP,whileinvestmentsinequipmentandmachinerydonot.CoulsonandKim(2000)alsoconfirmthatGDP’sresponsetoashockinresidentialinvestmentisseveraltimesthemagnitudeofaresponsetoashockininvestmentinequipmentandmachinery.TheysuggestthatresidentialshocksexplainbyfarthevariationinGDPthandoesashockinequipmentandmachinery.HuiandYiu(2003)usecausalityteststoempiricallyexaminethemarketfundamentaldynamicsofprivateresidentialrealestateprices.TheyshowthatresidentialinvestmentsleadGDP,becauseGDPisregardedasoneofthemarketfundamentalsthataffectdemandforprivateresidentialrealestate,whileitisaffectedbysomemarketfundamentals.Iacoviello(2003)inhisstudyofconsumption,housingprices,andcollateralconstraints,alsofindsadirecteffectfromhousingpricestoconsumption,which,inturn,influencesthecourseofeconomicgrowth.
3.3.Modellinghypothesismethodology
Thebaselineempiricalmodelforanalyzingtherelationshipbetweeneconomicactivityandtheliquiditymeasuresisthefollowing:
yt+1=
LIQ
˛+ˇiXLIQt
HS
+ˇiHSt
TS
+ˇiTSt
MRP
+ˇiMRPt
V
+ˇiVt
SR
+ˇiSRt
DS
+ˇiDSt
y
+ˇi
pi=0
yt−i+et
(3)
whereyt+1isthegrowthrateofthemacrovariable(RGDPorRCorRIorUnR)ofinterest,XLIQtisavectoroftheliquiditymeasures(ILRorRSorTURorVTR),HStistheleadingindicatorofhousingstarts,TStisthetermspread,MRPtisthemarket
LIQ,ˇHS,riskpremium,Vtdenotesstockmarketvolatility,SRtistheshort-terminterestrate,DStisthedefaultspread,andˇ
yt−iˇTS,ˇMRP,ˇV,ˇSR,andˇDSarethecoefficientsestimatesoftheaboveregressionvariables,respectively.Finally,
displaysanumberoflagsforthedependentvariable.Thenumberoflagshasbeendeterminedthroughthefinalpredictionerror(FPE)criterion,recommendedbyAkaike(1969)anddefinedastheminimizationofFPEp=p2(T−p)−1(T+p),wherep2indicatestheestimatederrorvariance,tindicatesthenumberofobservations,andpisthesizeofthelaglength.Finally,thedynamicpropertiesofGDPserieshavebeenthesubjectofalongdebate(NelsonandPlosser,1982;KimandPerron,2009).Inparticular,anumberofstudiesindicatethatGDPdisplayshighlypersistentchangesacrosstime,whileCampbellandMankiw(1989)addressthisissuebyprovidingsupportiveevidencethatshockstooutputarelargelypermanentandthatsubstantialpersistenceofoutputshocksisanimportantfeatureofpost-wareconomictimeseries.Tothisend,wemakeuseoftheHodrickandPrescott(1997)toobtainthecyclicalcomponentofrealGDP.
Totesttheimpactofliquidityoneconomicgrowth,wemakeuseoftheInstrumentalVariable(IV)methodologicalapproach,whichtakescaresofanyendogeneitybiasissues.Eq.(3)cannotbeestimatedconsistentlyusingsimpleordinary
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
49
leastsquare(OLS)duetopotentialendogeneitybetweeneconomicgrowthandanyoftheexplanatoryvariables,sowemaybecapturingreversecausality.Usinginstrumentsisthebestoptiontotakecareofanypossibleproblemsofreversecausationrunningfromeconomicgrowthtotheexplanatoryvariables.Duetothelackofproperinstrumentsforthoseexplanatoryvariables,leadsusnottochoosetwo-stageleastsquares(2SLS).Instead,weuseweakmethodsofcontrollingendogeneity,byusinglagsoftheexplanatoryvariablesastheappropriateinstruments.
Next,inordertotakeintoaccounttheliquidityvariationbetweensmallandlargesizedfirms,wealsoconstructtwoversionsofeachliquidityvariable,oneforsmallcapfirmsandoneforlargecapfirms.Allsamplestocksaresortedonmarketcapitalizationandallocatedintosizequartiles.Thefourliquiditymeasuresaresubsequentlycalculatedforthe25%smallestfirms(LIQS)andthe25%largestfirms(LIQL).Then,werunthefollowingregressionmodelthatcapturestheinformationcontentofsmallandlargecapfirmsseparately:
yt+1=
LIQ
˛+ˇiSXLIQst
LIQ
+ˇiLXLIOLt
HS
+ˇiHSt
TS
+ˇiTSt
MRP
+ˇiMRPt
V
+ˇiVt
SR
+ˇiSRt
DS
+ˇiDSt
y
+ˇi
pi=0
yt−1+et
(4)
Theabovemodelisestimatedwithandwithoutliquiditymeasures.
Table2providessomedescriptivestatisticsaboutthefourliquidityproxiesforthesmallandlargecapfirmsandthecorrelationcoefficientsamongthem.Forbothcountries,asexpected,largecapfirmsexhibithigherliquiditylevelsintermsoflowerILRandRSandhigherTURandVTRaswell.Furthermore,thecorrelationcoefficientbetweentheILRandRSispositiveacrossallcases,sincebothproxiesmeasureilliquidity,whilethereisapositiverelationshipbetweenthetwoliquiditymeasuresTURandVTR.Onceagain,thestatisticsconfirmthepresenceofnormality(i.e.,throughtheJarque–Berastatistic)andtheabsenceofserialautocorrelationacrossthethreelagsaswellasacrossallfourliquidityvariables.
3.4.Causalitytests
Simsetal.(1990)andTodaandPhillips(1993)arguethatWaldteststatisticsfornon-causalityinanunrestrictedVARcouldhaveanonstandardlimitdistribution.TodaandYamamoto(1995)suggestanalternativeapproachtocausalitytesting.Thebasicideaistoartificiallyaugmentthecorrectorder,K,oftheVARbythemaximalorderofintegration,sayTmax.TheaugmentedVARisthenestimatedandWaldtestsforlinearornonlinearrestrictionsarecarriedoutonthefirstKcoefficientmatrix(Caporaleetal.,2004).
WeuseToda-Yamamotocausalitytestsfortestingstatisticalcausalitybetweenstockmarketliquiditymeasuresandvariousproxiesforeconomicgrowth.Thenullhypothesisofinterestis:H0:stockmarketliquiditydoesnotGranger-causeeconomicgrowth.TheWaldstatisticconvergesindistributiontoarandomvariablewithmdegreesoffreedom,regardlessofwhethertheprocessisstationary,I(1),I(2),possiblyaroundalineartrendorwhetheritisco-integrated.Inaddition,theirmethodologyrequiressomepretestinginordertodeterminethelaglengthoftheprocess.Simsetal.(1990)showthatlagselectionprocedures,commonlyemployedforstationaryVARs,whicharebasedontestingthesignificanceoflaggedvectorsbymeansoftheWald(orLMorLR)tests,arealsovalidforVARswithI(1)processes,whichmightexhibitcointegration.WeaugmentthebivariateVARbythemaximumorderofintegrationintheseries.InthiscasethevariablesturnouttobeI(1).Therefore,weaugmentthebivariateVARbyonelagandtestfornon-causalityzerorestrictionsontheparametersoftheoriginalVARbycarryingoutWaldtestsonthefirstKcoefficientmatrix(TodaandYamamoto,1995).
4.Resultsanddiscussion
4.1.Baselineestimates
Table3presentstheIVestimates.ThedependentvariableintheseestimatesiseitherrealGDP(RGDP),orrealconsumption(RC),orrealinvestment(RI),ortheunemploymentrate(UnR).Sections1–4reportstheresultsfromtheAmihudilliquidityratio(ILR),Relativespread(RS),Turnover(TUR)andVolumetraded(VTR),respectively,whiletheresultsfromtheUKandGermanyarereportedinPanelAandB,respectively.ThelastthreecolumnsshowtheR2withtheinclusion/exclusionoftheliquidityvariableandtheF-testwhichteststhenullhypothesisthatacceptstherestrictedmodel(whentheLiquidityvariableisexcluded)andisbasedonthestatisticF=(RU2−RR2)/(1−RU2)×df1/(df0−df1),withRU2beingtheunrestrictedR2,RR2beingtherestrictedR2,withdf1=(N−k)degreesoffreedomanddf0=(N−k0)degreesoffreedom,respectively.
Theoverallpictureoftheempiricalresultscanbecharacterizedbythreefacts:First,thevastmajorityoftheexplanatoryvariablesarehighlystatisticallysignificant(atthe1%significancelevel)withtheexpectedsignsremainingstable,irrespec-tivelyoftheliquiditymeasureorthemarketunderinvestigation.Second,theexplanatorypowerofthemodelisquitehigh,rangingfrom37%(ILRfortheUKwhenUnRisthedependentvariable)to64%(RCfortheUKwhenVTRisthedependentvariable).Third,theincrementalinformationalcontentofthevariablerepresentingtheliquidityproxy(ILR,RS,TURandVTR)canbeestablished,judgingfromtherejectionoftherestrictedmodel(i.e.,theoneinwhichtheliquidityvariableisexcluded)acrossallcases,implyingthesubstantialroleofliquidityinexplainingthebehaviourofeconomicactivity.Thiscan
50
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
Table2
Descriptivestatisticsoflargeandsmallcapsliquidityproxies.
Mean
Median
Jarque–Berra
Section1:UKlargecapsliquidityproxiesPanelA:descriptivestatisticsILR0.062320.023202.74(0.42)RS0.025800.023335.82(0.33)TUR0.50706
0.53318
3.71(0.45)VTR
1,506,398,386
1,580,166,192
6.01(0.14)
PanelB:serialautocorrelations
Lags1
4
12
ILR2.48(0.48)2.94(0.29)3.37(0.17)RS2.70(0.43)2.63(0.38)3.48(0.15)TUR2.64(0.50)2.59(0.35)3.30(0.21)VTR
2.39(0.56)
2.44(0.42)
2.96(0.29)
PanelC:correlationcoefficientsRS0.48108TUR−0.57706−0.30692VTR
−0.28798
−0.20230
0.66781
Section2:UKsmallcapsliquidityproxiesPanelA:descriptivestatisticsILR0.251420.195102.18(0.64)RS0.131040.122424.29(0.50)TUR0.511130.503403.36(0.48)VTR
72,728,154
49,659,454
3.09(0.57)
PanelB:serialautocorrelations
Lags1
4
12
ILR2.64(0.38)2.81(0.24)3.11(0.16)RS2.47(0.46)2.73(0.35)3.02(0.20)TUR2.36(0.52)2.88(0.21)3.24(0.11)VTR
2.29(0.57)
2.64(0.39)
2.89(0.28)
PanelC:correlationcoefficients
ILR
RS
TUR
RS0.01314
TUR
−0.34143−0.05217VTR−0.20001−0.30034
0.08426
Section3:GermanylargecapsliquidityproxiesPanelA:descriptivestatisticsILR13.65080
9.380153.74(0.30)RS
0.017440.016644.66(0.13)TUR
0.412490.390373.67(0.33)VTR
25,930,67122,062,306
3.84(0.28)
PanelB:serialautocorrelations
Lags
1
4
12
ILR2.59(0.42)2.85(0.28)3.36(0.14)RS2.84(0.36)3.01(0.23)3.40(0.12)TUR2.66(0.38)2.95(0.29)3.32(0.17)VTR
2.82(0.39)
3.24(0.19)
3.68(0.11)
PanelC:correlationcoefficients
ILR
RS
TUR
RS0.26203
TUR
−0.19579−0.39028
VTR
−0.06852−0.40239
0.46739
Section4:GermansmallcapsliquidityproxiesPanelA:descriptivestatistics
ILR
18.3646912.223802.90(0.34)RS
0.090500.080735.11(0.16)TUR
0.449210.401524.64(0.27)VTR
2,081,5232,605,287
3.59(0.30)
No.firms
No.Obs.
1,6056,4212,3369,3451,8877,5482,089
8,356
1,6056,4212,3369,3451,8877,5482,089
8,356
9243,6981,1304,5211,0224,089974
3897
9243,6981,1304,5211,0224,089974
3,897
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
51
Table2(Continued)
PanelB:serialautocorrelations
Lags1
4
12
ILRRSTURVTR
PanelC:correlationcoefficientsofGermansmallcapsproxies
2.48(0.49)2.67(0.41)2.51(0.46)2.39(0.55)
2.80(0.29)2.96(0.27)2.99(0.26)2.84(0.35)
3.38(0.16)3.29(0.17)3.42(0.14)3.30(0.20)
ILR
RS
TUR
RSTURVTR
0.32748−0.14907−0.25010
−0.01489−0.09497
0.04685
Section1showsdescriptivestatistics,serialautocorrelationsandcorrelationcoefficientsfortheUKlargecapsliquidityproxiesfrom1994to2011.TheliquiditymeasuresaretheAmihud’silliquidityratio(ILR),therelativebid–askspread(RS),theturnover(TUR)andthevolumetraded(VTR).Theliquidityproxiesarecalculatedforeachavailablestockineachquarter.PanelAshowsthemeanandmedianoftheliquiditymeasures,theJarque–Berrastatisticwithp-valuesinparenthesis,thenumberofsecuritiesusedandthetotalnumberofobservations(eachsecurityisobservedinseveralquarters).PanelBshowstheresultsoftheQtestforserialautocorrelationforlags1,4and12.PanelCshowscorrelationcoefficientsbetweentheliquidityproxies.Thecorrelationsarecalculatedacrossallstocksandtime;theliquiditymeasuresarecalculatedforeachavailablestockineachquarterandthecorrelationsarepairwisecorrelationsbetweentheseliquiditymeasures.Section2showscorrespondingstatisticsfortheUKsmallcapsliquidityproxies,whereasSections3and4showscorrespondingstatisticsfortheGermanlargecapsandsmallcapsliquidityproxiesrespectivelyfrom1999through2011.
beclearlydiscernedbyobservingthemuchhighervaluesofR2producedwhenweincludetheliquidityproxyincomparisonwiththecorrespondingvaluesofR2generatedwhenweexcludeit.
AfinalremarkcanbemadeonthebasisofthevaluesofR2ingeneral,whenwecompareourfindingsbetweenliquidityproxiesandmacrovariables.ThemodelsthatinvolvetheVTRhaveonaveragealargerR2fortheUKthanthemodelswiththeotherliquidityproxies,whereasforGermanythemodelswiththelargerR2aretheonesinvolvingtheRS.Finally,itshouldbenotedthatthebestfitofthedataforbothcountriesisfoundinmostcaseswhenRGDPisusedasthedependentvariable.
Inmoredetails,wecanclearlyseethattheILRandRSilliquidityproxieshaveastatisticallysignificantnegativelinkwithGDP,consumptionandinvestmentandasignificantpositiveassociationwiththeunemploymentrateinbothcountries.Thisbehaviourofthetwoilliquidityproxiesisexpectedsincetheirincreaseindicatesafallinmarketliquidityandconsequentlyitshouldbenegativelyrelatedwitheconomicgrowth.Whenweusetheothertwoliquidityproxies,namelytheTURandVTR,theirfactorloadingsshowastrongandstatisticalsignificantpositiverelationshipwithRGDP,consumptionandinvestmentandanegativerelationshipwiththeunemploymentrate.Thus,theresultsindicatethatinbothcountrieswhenbothTURandVTRincrease,implyingthatstocksaremoreliquidandmoreeasilytraded,weanticipatefavourablemacroeconomicconditionsintermsofhigherGDP,consumptionandinvestmentandalowerunemploymentrate.
TheHousingStarts(HS)controlvariableisdisplayedtobesignificantlypositivelyrelatedacrosstheRGDP,RC,RIeconomicvariablesandnegativelyrelatedwiththeUnRvariable.Thepositiverelationshipbetweenthehousingstartsvariableandtherealmacrovariables(i.e.,RGDP,RCandRI)isexpectedaccordingtotheoreticaljustifications.However,therelationshipbetweenhousingstartsandunemploymentcanbeambiguous.Recentworkintheareahasnotconcludedaboutthesignoftherelationshipbetweenhousingstartsandlabourmarketeffects.BlanchflowerandOswald(2013)establishthatrisesinhome-ownershipintheUSareaprecursortoeventualsharprisesinunemploymentduetoreasonsoflowerlevelsoflabourmobilityandtofewernewbusinesses.Theirevidencesuggeststhatthehousingmarketcanproducenegative‘externalities’uponthelabourmarket.
Theothernon-equityandequityorientedcontrolvariableshaveastatisticallysignificantrelationshipwitheconomicactivity.Specifically,thetermspread(TS)andMRP(marketriskpremium)arepositivelyrelatedwithRGDP,RCandRI,whilebeingnegativelyrelatedwithUnR.Ontheotherhand,marketvolatility(V),short-terminterestrates(SR)andthedefaultspread(DS)aresignificantandnegativelyrelatedwithRGDP,RCandRIandpositivelywithUnR.Theseresultsholdinbothcountriesandacrossallliquidityproxies,whiletheyareasexpectedbytheoreticalargumentsandconsistentwiththepreviousempiricalevidence(EstrellaandMishkin,1998;RudebuschandWilliams,2009;Wright,2006;Meichleetal.,2011;Næsetal.,2011).
Overall,theresultsshowthatalthoughtheequityandnon-equitycontrolvariablesprovedtobeusefulinexplainingeconomicactivity,theliquidityproxiesofferimportantincrementalinformationandcouldnotbeomittedfromtheanalysis.Marketliquidityisshowntobeastrongleadingindicatorforeconomicconditionsforbothcountriesunderquestion,eventhoughtheUKisacapitalmarket-basedeconomyandGermanyisabank-basedeconomy.
Thiscanbeexplainedbythefactthatinterbankmarketsforliquidityareindirectlyconnectedtooneanotherthroughvariouschannelswiththeliquidityofstockmarkets,asNyborgandÖstberg(2014)recentlypointedout.Theyshowthatanincreaseintightness(e.g.,higherinterestrates)intheinterbankmarketforliquidityleadstopull-back(i.e.,the‘liquiditypull-backhypothesis’),whichinvolvessellingfinancialassets(themostliquidones)eitherdirectlybybanksorthroughleveragedinvestorswhoselltheirmostliquidassetstheyown(i.e.,portfoliorebalancing)tomeettheincreasedmargins’requirementsthatbanksimposeonthem.Ahugedropinliquidityintheinterbankmarketwasthecaseinthelastfinancial
52Table3
IVEstimateswithliquidityproxies.
˛
ˇLIQˇHSˇTSˇMRPˇVˇSRˇDSˇ1y
R2Adj.
F-test
ex.LIQ
in.LIQ
Section1:Amihudilliquidityratio(ILR)PanelA:UK
RGDP
0.758
(3.47)*
Lags=8
RC
0.761
(3.02)*
Lags=9
RI
0.862
(3.25)*
Lags=8
UnR
0.773
(3.28)*
Lags=8
PanelB:Germany
RGDP
0.874
(3.49)*
Lags=9
RC
0.741
(4.35)*
Lags=9
RI
0.873
(4.36)*
Lags=10
UnR
0.742
(4.35)*
Lags=8
Section2:relativespread(RS)PanelA:UK
RGDP
0.764
(4.14)*
Lags=8
RC
0.769
(3.24)*
Lags=9
RI
0.674
(3.87)*
Lags=8
UnR
0.761
−4.0180.436(−5.62)*
(4.49)*
−3.3720.365(−5.64)*
(5.26)*
−2.3250.238(−5.51)*
(5.14)*
0.429−0.238(6.12)*
(−5.46)*
−0.4670.403(−5.46)*
(5.24)*
−0.4480.386(−5.73)*
(5.25)*
−0.2580.325(−5.36)*
(5.31)*
0.417−0.157(5.42)*
(−5.26)*
−0.3950.465(−5.49)*
(5.52)*
−0.4250.435(−6.26)*
(5.18)*
−0.2630.338(−5.26)*
(5.20)*
0.282
−0.204
0.3310.149(4.18)*
(5.27)*
0.2530.119(4.61)*
(5.36)*
0.3420.255(5.30)*
(5.26)*
−0.384−0.118(−5.30)*
(−5.26)*
0.2940.163(5.18)*
(5.30)*
0.2580.214(5.12)*
(4.28)*
0.4640.230(5.02)*
(4.96)*
−0.264−0.183(−4.25)*
(−4.21)*
0.3360.153(5.31)*
(5.42)*
0.2650.248(4.85)*
(5.42)*
0.3950.285(5.04)*
(5.32)*
−0.426
−0.178
−0.135−0.248(−4.48)*
(−5.26)*
−0.142−0.249(−5.04)*
(−5.93)*
−0.141−0.429(−5.23)*
(−5.06)*
0.1240.461(4.15)*
(6.48)*
−0.138−0.219(−5.28)*
(−5.03)*
−0.159−0.286(−5.36)*
(−6.27)*
−0.153−0.438(−5.47)*
(−5.39)*
0.1350.415(4.26)*
(5.74)*
−0.257−0.286(−5.38)*
(−5.11)*
−0.192−0.326(−5.41)*
(−6.09)*
−0.163−0.508(−5.16)*
(−6.19)*
0.147
0.406
−0.1850.439(−4.34)*
(6.48)*
−0.2320.486(−5.14)*
(6.17)*
−0.3450.476(−5.28)*
(6.15)*
0.227−0.387(5.39)*
(−6.14)*
−0.2130.485(−5.16)*
(6.39)*
−0.2550.548(−4.96)*
(5.81)*
−0.3310.482(−5.05)*
(5.86)*
0.232−0.411(4.87)*
(−6.15)*
−0.2280.484(−5.19)*
(6.05)*
−0.3280.537(−5.03)*
(5.42)*
−0.3250.517(−5.26)*
(6.27)*
0.285
−0.434
0.60
0.34
0.53
0.28
0.57
0.30
0.37
0.24
0.61
0.34
0.56
0.31
0.56
0.29
0.54
0.28
0.63
0.38
0.58
0.27
0.56
0.30
0.46
0.26
39.47[0.00]35.46[0.00]N.40.52[0.00]
Apergisetal.32.85/[0.00]
Int.Fin.33.92[0.00]Markets,37.58Inst.[0.00]and39.62[0.00]Money3836.64[0.00]
(2015)42–6430.57[0.00]17.93[0.00]28.48[0.00]
27.69
Table3(Continued)
˛
(4.35)*
Lags=8
PanelB:Germany
RGDP0.752(3.38)*
Lags=10
RC0.732(3.80)*
Lags=9
RI1.328(4.25)*
Lags=9
UnR0.815(4.09)*
Lags=10
Section3:Turnover(TUR)PanelA:UK
RGDP0.683(4.27)*
Lags=10
RC0.655(4.18)*Lags=8
RI0.729(4.14)*
Lags=9UnR
0.835(5.15)*
Lags=9
PanelB:Germany
RGDP0.652(4.11)*Lags=8
RC0.612(4.51)*
Lags=9
RI0.625(3.86)*
Lags=9
UnR
0.753
ˇLIQˇHS(5.36)*(−5.12)*−0.4520.418(−5.49)*
(5.26)*
−0.4140.392(−5.32)*
(5.26)*
−0.2550.346(−5.41)*
(5.37)*
0.336−0.243(5.27)*
(−5.14)*
0.3590.413(5.38)*
(4.85)*
0.4170.463(5.22)*(5.40)*
0.3250.409(4.81)*
(5.36)*
−0.246−0.216(−4.88)*
(−4.84)*
0.4160.464(5.25)*(5.38)*0.4060.438(5.71)*
(4.91)*
0.2390.364(5.21)*
(5.48)*
−0.325
−0.339
ˇTSˇMRP(−4.30)*(−4.26)*0.3270.259(4.40)*
(5.25)*
0.2580.237(4.63)*
(5.02)*
0.4860.285(5.26)*
(5.03)*
−0.228−0.152(−5.08)*
(−4.81)*
0.3140.163(5.02)*
(5.29)*
0.2840.228(5.37)*
(4.82)*
0.3520.287(5.04)*
(5.25)*
−0.405−0.224(−5.35)*
(−4.30)*
0.3520.227(5.19)*(5.38)*0.3500.265(5.26)*
(5.28)*
0.4360.249(5.38)*
(5.15)*
−0.349
−0.147
ˇVˇSR(5.42)*(5.53)*−0.146−0.327(−5.18)*
(−5.29)*
−0.124−0.313(−4.97)*
(−6.06)*
−0.145−0.479(−5.18)*
(−5.40)*
0.1360.411(5.13)*
(4.81)*
−0.255−0.351(−5.34)*
(−5.28)*
−0.218−0.353(−5.25)*
(−5.36)*
−0.169−0.451(−5.36)*
(−5.52)*
0.1620.433(5.17)*
(5.74)*
−0.161−0.356(−5.30)*(−5.18)*−0.162−0.325(−5.20)*
(−5.51)*
−0.142−0.424(−5.06)*
(−4.98)*
0.418
0.279
ˇDSˇ1y
(5.10)*(−5.71)*−0.2970.510(−5.49)*
(6.39)*
−0.2960.530(−5.35)*
(5.92)*
−0.3470.526(−5.31)*
(5.71)*
0.325−0.463(5.24)*
(−4.96)*
−0.3020.549(−5.14)*
(6.10)*
−0.2550.537(−4.81)*
(5.39)*
−0.3640.520(−5.14)*
(5.92)*
0.259−0.482(4.72)*
(−5.92)*
−0.3020.479(−5.27)*(5.83)*
−0.3030.486(−5.04)*
(5.82)*
−0.3580.486(−5.14)*
(6.15)*
0.549
−0.415
R2Adj.
in.LIQ
ex.LIQ
0.63
0.40
0.57
0.28
0.55
0.22
0.52
0.29
0.63
0.34
0.55
0.31
0.57
0.33
0.48
0.24
0.60
0.37
0.58
0.30
0.54
0.23
0.52
0.31
F-test
[0.00]
33.48[0.00]
32.78[0.00]N.37.85[0.00]Apergisetal.35.23/[0.00]
Int.Fin.33.81Markets,[0.00]Inst.29.82and[0.00]Money30.6238[0.00]28.09(2015)[0.00]
42–6428.26[0.00]32.47[0.00]
37.13[0.00]32.59
53Table3(Continued)
˛
ˇLIQˇHSˇTSˇMRPˇVˇSRˇDSˇ1y
R2Adj.
F-test
in.LIQ
ex.LIQ
(4.14)*
(−5.26)*(−5.16)*(−5.25)*(−5.16)*(5.07)*(5.72)*(5.24)*(−4.63)*[0.00]
Lags=8
Section4:Volume(VTR)PanelA:UK
RGDP0.7730.3630.4360.3290.209−0.262−0.330−0.2970.5360.61
0.38
30.06(4.25)*
(5.29)*
(5.14)*
(5.29)*
(5.24)*
(−5.28)*
(−5.03)*
(−5.25)*
(5.91)*
[0.00]Lags=10
RC0.7140.4420.4830.2750.246−0.240−0.321−0.2650.5270.64
0.30
32.94(4.06)*
(5.15)*
(5.39)*
(5.25)*
(5.25)*
(−5.24)*
(−5.10)*
(−5.13)*
(5.86)*
[0.00]
Lags=8
RI0.8190.3620.4340.3470.238−0.180−0.531−0.3140.5080.60
0.36
30.13(5.02)*
(5.09)*
(4.92)*
(5.26)*
(5.18)*
(−5.45)*
(−5.32)*
(−5.29)*
(5.89)*
[0.00]Lags=8
UnR
0.674−0.244−0.248−0.386−0.2490.1640.4260.285−0.5370.48
0.25
34.18(5.36)*
(5.02)*
(−5.25)*
(−5.35)*
(−5.12)*
(5.42)*
(5.30)*
(5.16)*
(−5.79)*
[0.00]
Lags=9
PanelB:Germany
RGDP0.7140.4150.4290.3590.264−0.148−0.329−0.2360.5150.55
0.33
32.79(4.28)*
(5.48)*
(4.83)*
(5.23)*
(5.26)*
(−4.85)*
(−5.04)*
(−4.87)*
(5.40)*
[0.00]Lags=9
RC0.6940.4370.4050.3260.284−0.146−0.329−0.2270.5230.58
0.31
33.76(3.48)*
(4.92)*
(4.81)*
(5.37)*
(5.14)*
(−5.25)*
(−5.29)*
(−5.19)*
(5.61)*
[0.00]Lags=9
RI0.6580.2480.4110.4030.215−0.130−0.416−0.3820.5460.53
0.26
37.94(4.29)*
(4.82)*
(5.20)*
(5.26)*
(4.88)*
(−4.82)*
(−4.67)*
(−5.38)*
(5.52)*
[0.00]Lags=9
UnR0.652−0.234−0.335−0.352−0.2360.1380.4280.242−0.5450.50
0.26
35.13(4.27)*
(−5.21)*
(−5.03)*
(−5.27)*
(−4.25)*
(5.25)*
(4.79)*
(5.18)*
(−5.39)*
[0.00]
Lags=8
Thetablereportstheresultswhenweregresseachmacrovariable(RealGDP,RealConsumption,RealInvestment,Unemploymentrate),againsttheAmihudilliquidityratio(ILR)inSection1,therelativespread(RS)inSection2,theturnover(TUR)inSection3,volumetraded(VTR)inSection4andsixcontrolvariables(HS,TS,MRP,V,SR,DS)forUK(PanelA)andGermany(PanelB).Thecross-sectionalliquiditymeasuresarecalculatedasequallyweightedaveragesacrossstocks.Figuresinparenthesesdenotet-statistics.Thecolumnlabelled“R2in.LIQ”givestheadjustedR2foramodelwiththeliquidityproxy(LIQ)andthecolumnlabeled“R2ex.LIQ”givestheadjustedR2foramodelwithouttheliquidityproxy(LIQ).F-testdenotesthetestonthenullhypothesisthatacceptstherestrictedmodel(whentheLiquidityvariableisexcluded)andisbasedonthestatisticF=(RU2−RR2)/(1−RU2)×df1/(df0−df1),withRU2beingtheunrestrictedR2,RR2beingtherestrictedR2,df1=(N−k)degreesoffreedomanddf0=(N−k0)degreesoffreedom.Lags=showsthenumberoflagsofthedependentvariablesusedastheinstrumentsintheregression.*
Denotesstatisticalsignificanceatthe1%significancelevel,whilefiguresinbracketsdenotep-values.
54N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64Table4
IVEstimateswithliquidityproxiesforbothsmall-andlarge-capfirms.
A
ˇLIQSˇLIQLˇHSˇTSˇMRPˇVˇSRˇDSˇy
R2adjusted
F-test
ex.LIQ
in.LIQ
Section1:Amihudilliquidityratio(ILR)PanelA:UK
RGDP
0.763(3.28)*
Lags=8
RC
0.722(3.81)*
Lags=10
RI0.784(3.28)*
Lags=9
UnR
0.613(3.93)*Lags=10
PanelB:Germany
RGDP
0.589
(3.41)*
Lags=9
RC
0.648(4.23)*
Lags=10
RI
0.855(4.26)*
Lags=9
UnR
0.614(4.25)*
Lags=10
Section2:relativespread(RS)PanelA:UK
RGDP
0.682
(3.86)*
Lags=10
RC
0.584
(4.38)*
Lags=9
RI
0.685
(3.28)*
Lags=8
UnR
0.709
−0.159
−0.126(−5.28)*(−5.19)*
−0.236
−0.215(−5.20)*(−4.87)*
−0.236
−0.214(−5.38)*
(−5.29)*
0.242
0.214(5.37)*
(4.96)*
−0.134−0.118(−4.86)*(−4.71)*
−0.264
−0.216(−5.27)*(−5.38)*−0.246
−0.214(−5.18)*(−5.09)*
0.239
0.226(5.36)*(5.14)*
−0.174−0.125(−5.28)*(−5.48)*−0.139−0.148(−5.01)*
(−5.26)*
−0.236−0.225(−4.79)*
(−5.19)*
0.139
0.126
0.3140.168(5.92)*
(5.23)*
0.3250.132(5.16)*
(5.27)*
0.2880.337(4.82)*
(5.41)*
−0.248−0.329(−5.36)*
(−5.02)*
0.3280.213(4.52)*
(4.83)*
0.3420.147(5.28)*(5.16)*
0.3570.250(5.39)*
(5.26)*
−0.328−0.314(−5.30)*
(−5.28)*
0.4420.173(5.39)*(5.14)*0.3390.142(5.13)*
(5.06)*
0.3380.329(−5.41)*
(4.59)*
0.236
−0.338
0.139−0.147(5.42)*
(−5.28)*
0.245−0.156(5.19)*
(−5.24)*
0.226−0.136(5.38)*
(−5.10)*
−0.2350.149(−5.41)*
(5.14)*
0.125−0.128(4.94)*
(−5.03)*
0.249−0.135(5.36)*
(−5.17)*
0.242−0.141(5.37)*
(−5.17)*
−0.2270.135(−5.28)*
(5.19)*
0.142−0.248(5.14)*
(−5.19)*
0.248−0.215(5.30)*
(−4.84)*
0.253−0.246(5.12)*
(−5.15)*
−0.225
0.142
−0.245−0.249(−5.02)*
(−4.86)*
−0.329−0.248(−5.16)*
(−4.82)*
−0.412−0.312(−5.65)*
(−5.01)*
0.3270.226(5.25)*
(5.78)*
−0.256−0.214(−4.87)*
(−4.53)*
−0.329−0.226(−5.18)*
(−5.35)*
−0.329−0.324(−5.36)*
(−5.11)*
0.3240.285(5.07)*
(5.75)*
−0.284−0.265(−5.40)*
(−5.21)*
−0.352−0.254(−5.18)*
(−5.36)*
−0.428−0.240(−5.81)*
(−4.94)*
0.352
0.263
0.4350.60
(6.06)*
0.4260.55
(6.61)*
0.5140.51
(6.13)*
−0.4290.44
(−6.11)*
0.4290.53
(5.37)*
0.4300.56
(5.82)*
0.5140.50
(5.27)*
−0.3790.52
(−5.49)*
0.4680.58
(5.41)*
0.4390.52
(5.71)*
0.5160.52
(6.05)*
0.462
0.45
0.29
35.48[0.00]0.26
38.37[0.00]0.26
36.29[0.00]0.22
36.71[0.00]
0.31
31.75[0.00]0.30
33.32[0.00]0.30
30.48[0.00]
0.25
36.73[0.00]
0.34
33.29[0.00]0.28
34.21[0.00]0.26
28.26[0.00]0.23
28.33
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–645556Table4(Continued)
A
ˇLIQSˇLIQLˇHSˇTSˇMRPˇVˇSRˇDSˇy
R2adjusted
F-test
(4.17)*
Lags=9
PanelB:Germany
RGDP0.685(3.16)*
Lags=8
RC0.652(4.19)*
Lags=8
RI0.840(4.27)*
Lags=10
UnR0.612(4.09)*Lags=9
Section3:turnover(TUR)PanelA:UK
RGDP0.736(3.95)*
Lags=9
RC0.714(4.21)*
Lags=9
RI0.724(4.25)*Lags=9
UnR
0.682(4.12)*
Lags=9
PanelB:GermanyRGDP0.583(4.32)*Lags=9
RC0.572(4.31)*
Lags=9
RI0.761(3.75)*
Lags=8
UnR
0.572
(5.11)*(5.30)*−0.168−0.155(−5.04)*
(−5.37)*
−0.225−0.159(−5.17)*
(−5.24)*
−0.268−0.224(−5.35)*
(−5.16)*
0.1980.157(5.10)*(5.13)*
0.1680.128(5.41)*
(5.04)*
0.2370.177(5.35)*
(5.21)*
0.1530.128(5.15)*(5.36)*−0.239−0.157(−5.46)*
(−5.39)*
0.1460.117(4.28)*(5.25)*0.2490.225(5.02)*
(5.01)*
0.1460.133(5.28)*
(5.19)*
−0.247
−0.227
(4.83)*(−4.88)*0.3530.245(5.01)*
(5.48)*
0.3280.158(4.92)*
(5.14)*
0.3300.268(5.17)*
(5.07)*
−0.329−0.325(−4.74)*
(−5.31)*
0.4280.217(5.00)*
(5.16)*
0.3410.135(5.38)*
(5.11)*
0.3210.325(5.19)*(5.26)*0.258−0.339(5.26)*
(−5.12)*
0.3740.285(5.06)*(4.96)*
0.4090.160(5.16)*
(5.08)*
0.3250.342(5.14)*
(5.21)*
0.345
−0.152
(−5.17)*(5.13)*0.146−0.247(5.15)*
(−5.20)*
0.244−0.262(5.15)*
(−5.07)*
0.255−0.152(5.21)*
(−5.29)*
−0.1620.149(−5.24)*
(5.20)*
0.236−0.253(5.14)*
(−4.88)*
0.249−0.224(5.26)*
(−5.19)*
0.239−0.238(5.36)*
(−5.51)*
−0.1600.145(−4.86)*
(5.24)*
0.238−0.229(4.87)*
(−5.14)*
0.225−0.234(5.24)*
(−5.26)*
0.249−0.237(4.86)*
(−5.02)*
0.158
0.416
(5.26)*(5.90)*−0.313−0.228(−5.25)*
(−5.19)*
−0.325−0.249(−5.16)*
(−5.04)*
−0.328−0.241(−5.82)*
(−4.80)*
0.3470.286(5.19)*
(6.25)*
−0.296−0.238(−4.74)*
(−5.29)*
−0.335−0.226(−4.92)*
(−5.15)*
−0.324−0.255(−5.29)*
(−5.07)*
0.3320.253(4.91)*
(5.85)*
−0.316−0.214(−4.78)*
(−5.24)*
−0.349−0.215(−5.26)*
(−5.26)*
−0.413−0.238(−5.63)*
(−4.86)*
0.238
0.485
in.LIQ
(5.97)*0.5070.60
(6.36)*
0.4180.54
(5.86)*
0.4810.52
(6.14)*
−0.4180.54
(−5.44)*
0.4290.61
(6.15)*
0.4360.50
(5.84)*
0.4720.52
(6.10)*
0.4460.47
(6.17)*
0.4290.55
(5.06)*
0.4060.52
(5.82)*
0.4150.57
(5.26)*
−0.252
0.58
ex.LIQ
[0.00]
0.31
36.18[0.00]0.29
38.33[0.00]0.27
31.08[0.00]0.24
35.89[0.00]
0.32
34.27[0.00]0.24
34.19[0.00]0.28
33.64[0.00]0.24
34.63[0.00]
0.32
29.78[0.00]0.30
36.15[0.00]
0.29
32.16[0.00]0.27
34.83
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64Table4(Continued)
A
ˇLIQSˇLIQLˇHSˇTSˇMRPˇVˇSRˇDSˇy
R2adjusted
F-test
in.LIQ
ex.LIQ
(4.11)*
(−4.92)*(−5.39)*(5.25)*(−5.19)*(5.32)*(5.74)*(5.29)*(5.82)*(−4.37)*[0.00]
Lags=10
Section4:Volume(VTR)PanelA:UK
RGDP0.6580.2270.1460.3500.2150.159−0.246−0.342−0.2360.4480.60
0.35
33.29(4.12)*
(5.25)*
(5.03)*
(5.27)*
(5.37)*
(5.36)*
(−5.10)*
(−5.84)*
(−5.15)*
(5.33)*
[0.00]Lags=9
RC0.5140.1390.1260.3140.1450.237−0.214−0.324−0.2360.4350.54
0.26
33.95(4.20)*
(5.05)*
(5.30)*
(5.08)*
(5.20)*
(5.31)*
(−5.02)*
(−4.84)*
(−4.79)*
(5.91)*
[0.00]Lags=10
RI0.6720.2360.2270.3840.2850.242−0.213−0.427−0.2250.4620.54
0.30
36.15(4.32)*
(5.13)*
(4.81)*
(5.25)*
(5.12)*
(5.36)*
(−5.21)*
(−5.89)*
(−5.26)*
(5.36)*
[0.00]
Lags=10
UnR
0.732−0.232−0.148−0.305−0.249−0.2360.2060.3920.2640.4270.48
0.26
35.19(4.10)*
(−5.14)*
(−5.13)*
(−5.20)*
(−5.31)*
(−5.12)*
(5.15)*
(5.11)*
(5.72)*
(5.24)*
[0.00]
Lags=10
PanelB:Germany
RGDP0.6070.1850.1550.4420.2480.174−0.310−0.326−0.2510.5120.59
0.33
34.16(3.78)*
(5.16)*
(5.26)*
(5.37)*
(5.05)*
(5.28)*
(−5.42)*
(−5.24)*
(−5.39)*
(6.08)*
[0.00]Lags=8
RC0.6250.1920.1490.3510.2090.254−0.158−0.339−0.2260.4360.57
0.33
33.85(3.91)*
(5.04)*
(5.24)*
(5.16)*
(5.25)*
(5.17)*
(−4.93)*
(−5.18)*
(−4.85)*
(5.94)*
[0.00]Lags=19
RI0.4830.2360.2180.3490.2470.240−0.247−0.429−0.2180.5460.61
0.29
36.42(3.39)*
(5.25)*
(5.26)*
(5.07)*
(5.12)*
(5.21)*
(−5.37)*
(−6.13)*
(−5.03)*
(5.72)*
[0.00]Lags=10
UnR0.606−0.223−0.158−0.327−0.239−0.4290.3050.4580.224−0.3370.55
0.27
31.64(4.15)*
(−5.19)*
(−4.84)*
(−5.13)*
(−5.36)*
(−5.17)*
(6.14)*
(5.92)*
(4.54)*
(−4.83)*
[0.00]
Lags=11
Thetablereportstheresultswhenweregresseachmacrovariable(RealGDP,RealConsumption,RealInvestment,Unemploymentrate),againsttheAmihudilliquidityratio(ILR)inSection1,therelativespread(RS)inSection2,theturnover(TUR)inSection3,Volumetraded(VTR)inSection4,forsmall-cap(LIQS)andlarge-cap(LIQL)firms,andsixcontrolvariables(HS,TS,MRP,V,SR,DS)forUK(PanelA)andGermany(PanelB).Thet-statisticsarereportedinparentheses.Thecolumnlabeled“R2in.LIQ”givestheadjustedR2foramodelwithbothilliquidityratiosandthecolumnlabeled“R2ex.LIQ”givestheadjustedR2foramodelwithoutanyilliquidityratios.F-testdenotesthetestofthenullhypothesisthatacceptstherestrictedmodel(whentheLiquidityvariableisexcluded)andisbasedonthestatisticF=(RU2−RR2)/(1−RU2)xdf1/(df0−df1),withRU2beingtheunrestrictedR2,RR2beingtherestrictedR2,df1=(N−k)degreesoffreedomanddf0=(N−k0)degreesoffreedom.Lags=showsthenumberoflagsofthedependentvariablesusedastheinstrumentsintheregression.*
Denotesstatisticalsignificanceatthe1%significancelevel,whilefiguresinbracketsdenotep-values.
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–645758
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
Table5
Waldtestsforsmall-andlarge-capfirms.
UKvariables
Germanyvariables
RGDP
RC
RI
UnR
RGDP
RC
RI
UnR
PanelA:H0:ˇLIQS−ˇLIQL=0Illiquidityratio(ILR)H0:ˇILRS−ˇILRL=0Relativespread(RS)H0:ˇRSS−ˇRSL=0Turnover(TUR)
H0:ˇTURS−ˇTURL=0Volume(VTR)
H0:ˇVTRS−ˇVTRL=0PanelB:H0:ˇLIQS−ˇLIQL>0Illiquidityratio(ILR)H0:ˇILRS−ˇILRL>0Relativespread(RS)H0:ˇRSS−ˇRSL>0Turnover(TUR)
H0:ˇTURS−ˇTURL>0Volume(VTR)
H0:ˇVTRS−ˇVTRL>0
(0.01)
(0.00)
(0.01)
(0.00)
(0.01)
(0.00)
(0.01)
(0.00)(0.00)(0.00)(0.00)
(0.00)
(0.01)
(0.00)
(0.00)
(0.00)
(0.01)
(0.00)
(0.01)
(0.01)
(0.00)
(0.00)
(0.01)
(0.00)
(0.00)
(0.00)
(0.00)
(0.01)
(0.00)
(0.00)
(0.01)
(0.01)
(0.23)
(0.24)
(0.20)
(0.16)
(0.26)
(0.22)
(0.21)
(0.19)(0.22)(0.20)(0.26)
(0.20)
(0.18)
(0.14)
(0.12)
(0.21)
(0.28)
(0.22)
(0.16)
(0.16)
(0.16)
(0.12)
(0.24)
(0.32)
(0.25)
(0.22)
(0.31)
(0.27)
(0.21)
(0.29)
(0.35)
(0.28)
ThetableshowsWaldtestsbetweenthecoefficientsoftheliquidityproxiesforbothsmall-andlarge-capfirms.ThetestisperformedforthewholesampleperiodfortheUKfrom1994to2011andforGermanyfrom1999to2011.Foreachmeasure,inPanelAwetestthenullhypothesisthatthedifferencebetweentheilliquiditymeasureforsmall-andlarge-capfirmsisequaltozeroandsecondinPanelBwetestthenullhypothesisthatthedifferencebetweentheilliquiditymeasureforsmall-andlarge-capfirmsisgreaterthanzero.Wereportp-valuesinparenthesesforeachtest.
Table6
CausalityresultsforUK&Germany(bothsmallandlargefirms).
UKvariables
Germanyvariables
RGDP
RC
RI
UnR
RGDP
RC
RI
UnR(+)17.81(0.00)*0.21(0.59)(+)23.18(0.00)*1.03(0.67)(−)18.14(0.00)*1.24(0.56)(−)19.85(0.00)*1.27(0.36)
Illiquidityratio
H0:ILRMACRO2-testp-Value
H0:MACROILR2-testp-Value
Relativespread(RS)H0:RSMACRO2-testp-Value
H0:MACRORS2-testp-ValueTurnover(TUR)
H0:TURMACRO2-testp-Value
H0:MACROTUR2-testp-ValueVolume(VTR)
H0:VTRMACRO2-testp-Value
H0:MACROVTR2-testp-Value
(−)19.74(0.00)*
(−)18.36(0.00)*
(−)17.34(0.00)*
(+)15.92(0.00)*
(−)18.28(0.00)*
(−)16.15(0.00)*
(−)24.66(0.00)*
0.36(0.58)
0.06(0.78)
0.09(0.78)
0.22(0.61)
0.58(0.47)
0.03(0.88)
0.00(0.94)
(−)16.59(0.00)*
(−)17.82(0.00)*
(−)20.53(0.00)*
(+)20.29(0.00)*
(−)15.24(0.00)*
(−)28.16(0.00)*
(−)20.53(0.00)*
0.57(0.49)
0.12(0.79)
0.00(0.99)
0.01(0.99)
2.04(0.31)
0.75(0.50)
0.84(0.71)
(+)18.64(0.00)*
(+)15.43(0.00)*
(+)19.61(0.00)*
(−)15.14(0.00)*
(+)20.47(0.00)*
(+)15.42(0.00)*
(+)22.36(0.00)*
15.52(0.00)*
22.71(0.00)*
1.05(0.45)
1.12(0.59)
18.93(0.00)*
24.22(0.00)*
0.00(0.99)
(+)18.49(0.00)*
(+)16.45(0.00)*
(+)20.73(0.00)*
(−)19.37(0.00)*
(+)17.21(0.00)*
(+)20.53(0.00)*
(+)22.90(0.00)*
17.54(0.00)*
22.62(0.00)*
0.38(0.64)
0.52(0.49)
19.82(0.00)*
16.35(0.00)*
0.00(0.99)
ThetableshowsGrangercausalitytestsbetweenMACRO:(a)realGDPgrowth,(b)realconsumption,(c)realinvestment,(d)theunemploymentrateand(a)theAmihudILR,(b)therelativespread(RS),(c)turnover(TUR)and(d)volumetraded(VTR).Thecross-sectionalliquiditymeasuresarecalculatedasequallyweightedaveragesacrossstocks.ThetestisperformedforthewholesampleperiodfortheUKfrom1994to2011andforGermanyfrom1999to2011.Foreachmeasure,wefirsttestthenullhypothesisthatmacrovariabledoesnotGrangercausemarketilliquidityandwhethermarketilliquiditydoesnotGrangercausethemacrovariable.Theresultsinparenthesesdenotethesignoftheassociationbetweenthevariablesunderinvestigation.WereportF-testvaluesandcorrespondingp-values(inparentheses)foreachtest.*
Denotesarejectionofthenullhypothesisofcausalityatthe1%level.
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
59
Table7
CausalityresultsforUK&Germany(smallfirms).
UKvariables
Germanyvariables
RGDP
RC
RI
UnR
RGDP
RC
RI
UnR(+)32.53(0.00)*6.36(0.12)(+)28.93(0.00)*15.93(0.00)*(−)25.52(0.00)*25.26(0.00)*(−)25.31(0.00)*23.09(0.00)*
Illiquidityratio
H0:ILRMACRO2-testp-Value
H0:MACROILR2-testp-Value
Relativespread(RS)H0:RSMACRO2-testp-Value
H0:MACRORS2-testp-ValueTurnover(TUR)
H0:TURMACRO2-testp-Value
H0:MACROTUR2-testp-ValueVolume(VTR)
H0:VTRMACRO2-testp-Value
H0:MACROVTR2-testp-Value
(−)37.61(0.00)*
(−)32.19(0.00)*
(−)25.62(0.00)*
(+)17.06(0.00)*
(−)29.17(0.00)*
(−)32.53(0.00)*
(−)30.88(0.00)*
7.16(0.11)
7.81(0.11)
6.42(0.14)
9.72(0.10)***
8.05(0.11)
4.91(0.18)
10.52(0.09)***
(−)30.15(0.00)*
(−)28.45(0.00)*
(−)27.55(0.00)*
(+)26.27(0.00)*
(−)34.02(0.00)*
(−)29.62(0.00)*
(−)25.16(0.00)*
16.13(0.00)*
12.35(0.02)**
13.52(0.01)*
15.41(0.01)*
17.54(0.00)*
15.31(0.01)*
11.37(0.03)**
(+)31.25(0.00)*
(+)24.18(0.00)*
(+)26.48(0.00)*
(−)23.92(0.00)*
(+)33.84(0.00)*
(+)24.02(0.00)*
(+)33.16(0.00)*
21.46(0.00)*
33.17(0.00)*
25.09(0.00)*
29.66(0.00)*
36.03(0.00)*
32.45(0.00)*
25.16(0.00)*
(+)31.16(0.00)*
(+)22.63(0.00)*
(+)29.71(0.00)*
(−)22.45(0.00)*
(+)26.15(0.00)*
(+)28.73(0.00)*
(+)26.26(0.00)*
34.32(0.00)*
35.16(0.00)*
27.94(0.00)*
25.86(0.00)*
33.15(0.00)*
27.38(0.00)*
25.17(0.00)*
ThetableshowsGrangercausalitytestsbetweenMACRO:(a)realGDPgrowth,(b)realconsumption,(c)realinvestment,(d)theunemploymentrateand(a)theAmihudILR,(b)therelativespread(RS),(c)turnover(TUR)and(d)volumetraded(VTR).Thecross-sectionalliquiditymeasuresarecalculatedasequallyweightedaveragesacrossthe25%smallcapfirms(LIQS).ThetestisperformedforthewholesampleperiodfortheUKfrom1994to2011andforGermanyfrom1999to2011.Foreachmeasure,wefirsttestthenullhypothesisthatmacrovariabledoesnotGrangercausemarketilliquidityandwhethermarketilliquiditydoesnotGrangercausethemacrovariable.Theresultsinparenthesesdenotethesignoftheassociationbetweenthevariablesunderinvestigation.WereportF-testvaluesandcorrespondingp-values(inparentheses)foreachtest.******,,Denotetherejectionofthenullhypothesisofcausalityatthe1%,5%,10%levels,respectively.
crisis(Cassolaetal.,2009).Insuchasituation,bankscannotattractadditionaldepositseasilyandwithoutasignificantcostofdeposits,i.e.,higherinterestrates.Moreover,AllenandCarletti(2008)emphasizethepositiveroleofbanksforbettermonitoringoflendingandtheirrisksharingroleinincompletemarketsandpointoutthatbanksarealsoexposedtodeposits’withdrawalsordry-upofliquidityintheshort-termcapitalmarkets;inthatsense,theycanspreadthecrisestotherealeconomy.AsNyborgandÖstberg(2014)document,theliquiditypull-backhypothesisisvalidevenforthepre-crisisperiod,whileitseffectissubstantiallystrongerfortheUSmarketoverthepre-crisisperiod,thanoverthecrisisperiodduetotheimmediateresponseoftheUSFedwithaloosemonetarypolicy(e.g.,TAFandQuantitativeEasingmeasures)whichhelpedthebankstocurtailtheirneedsforaliquiditypull-back.
Theresultsregardingthestrongexplanatorypoweroftheliquidityproxiesforeconomicactivityareverysimilartothosereportedbypreviousempiricalstudiesfollowingasimilarapproach(Meichleetal.,2011;Næsetal.,2011;Florackisetal.,2014a).Inparticular,ifwecompareourfindingswiththosebyNæsetal.(2011),weconcludethatwegetevenstrongerresults(i.e.,higherR2valuesinourcase),whichpointtothesamedirection.Thisconclusioncanalsobedocumentedbythefactthatweobtainstatisticalsignificanceconsistentlyacrossallindependentvariables(includingthecontrolvariables)andacrossallequationsused,withoutlosingtheincrementalinformationalcontentoftheilliquidity/liquidityproxies.Incontrast,inthestudyofNæsetal.(2011)whenthedependentvariableiseithertheinvestmentortheconsumptionvariable,whiletheyaddbothexcessmarketreturnsandvolatility,thecoefficientoftheILRproxyturnsouttobestatisticallyinsignificant,alongwithinsignificantcoefficientsofthevolatilitymeasure.
Finally,Table4showstheestimationresultsfortheUKandGermany,whentheliquidityratioproxy(ILR,RS,TURandVTR)isappliedforbothsmallandlargecapfirms.ThelastthreecolumnsinTable4showthevaluesoftheR2again,undertwooptions:withtheliquidityproxyofbothsmallandlargecapfirmsandwithouttheliquidityproxyforbothfirmsandtheF-test,whichteststhenullhypothesisthatacceptstherestrictedmodel(whenbothliquidityvariablesareexcluded).Wecanobservethatthecoefficientacrossallliquidityproxies(ILR,RS,TURandVTR)regardingsmallfirmsishigherthanthecorrespondingliquiditycoefficientforlargefirms.Thisdifferenceremainsconsistentacrossalldifferentapplicationsofdependentvariables(RGDP,RC,RIandUnR)andinbothcountries.Thisfindingimpliesthattheimpactoftheliquidityproxyoneconomicactivityismorepronouncedinthecaseofsmallcapfirms,asitwasexpected.Forbothcountriesandacrossall
60
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
Table8
CausalityresultsforUK&Germany(largefirms).
UKvariables
Germanyvariables
RGDP
RC
RI
UnR
RGDP
RC
RI
UnR(+)24.84(0.00)*0.54(0.58)(+)30.43(0.00)*1.16(0.64)(−)25.94(0.00)*1.38(0.40)(−)25.19(0.00)*1.16(0.45)
Illiquidityratio
H0:ILRMACRO2-testp-Value
H0:MACROILR2-testp-Value
Relativespread(RS)H0:RSMACRO2-testp-Value
H0:MACRORS2-testp-ValueTurnover(TUR)
H0:TURMACRO2-testp-Value
H0:MACROTUR2-testp-ValueVolume(VTR)
H0:VTRMACRO2-testp-Value
H0:MACROVTR2-testp-Value
(−)28.59(0.00)*
(−)30.26(0.00)*
(−)23.92(0.00)*
(+)25.14(0.00)*
(−)23.82(0.00)*
(−)22.61(0.00)*
(−)32.19(0.00)*
1.17(0.42)
0.36(0.75)
0.58(0.69)
0.41(0.63)
1.45(0.33)
0.24(0.81)
0.25(0.85)
(−)27.25(0.00)*
(−)22.17(0.00)*
(−)28.03(0.00)*
(+)30.48(0.00)*
(−)26.13(0.00)*
(−)28.27(0.00)*
(−)26.92(0.00)*
2.24(0.30)
0.38(0.72)
0.24(0.79)
0.52(0.83)
2.26(0.20)
1.24(0.33)
1.39(0.73)
(+)31.26(0.00)*
(+)23.29(0.00)*
(+)21.08(0.00)*
(−)22.36(0.00)*
(+)34.22(0.00)*
(+)23.19(0.00)*
(+)32.14(0.00)*
2.03(0.29)
1.46(0.48)
1.15(0.50)
1.29(0.48)
2.12(0.30)
1.24(0.49)
0.37(0.78)
(+)28.01(0.00)*
(+)22.36(0.00)*
(+)29.32(0.00)*
(−)23.16(0.00)*
(+)28.13(0.00)*
(+)27.14(0.00)*
(+)30.35(0.00)*
1.25(0.39)
1.72(0.35)
0.58(0.40)
0.68(0.43)
1.09(0.32)
0.71(0.41)
0.28(0.76)
ThetableshowsGrangercausalitytestsbetweenMACRO:(a)realGDPgrowth,(b)realconsumption,(c)realinvestment,(d)theunemploymentrateand(a)theAmihudILR,(b)therelativespread(RS),(c)turnover(TUR)and(d)volumetraded(VTR).Thecross-sectionalliquiditymeasuresarecalculatedasequallyweightedaveragesacrossthe25%largecapfirms(LIQL).ThetestisperformedforthewholesampleperiodfortheUKfrom1994to2011andforGermanyfrom1999to2011.Foreachmeasure,wefirsttestthenullhypothesisthatmacrovariabledoesnotGrangercausemarketilliquidityandwhethermarketilliquiditydoesnotGrangercausethemacrovariable.Theresultsinparenthesesdenotethesignoftheassociationbetweenthevariablesunderinvestigation.WereportF-testvaluesandcorrespondingp-values(inparentheses)foreachtest.*
Denotesarejectionofthenullhypothesisofcausalityatthe1%level.
macrovariables,whenweincludeboththeliquidityproxiestheR2turnsouttobehigherthanthatobtainedbyexcludingthem.Forexample,inSection1(ILR)intheequationofRGDPfortheUK,theR2inthefirstcaseisupto60%,whileinthelattercaseitgoesdownto29%.AsimilareffectisfoundinthecaseofGermany(sameSection1,ILR),whentheR2goesupfrom31%to53%,whenILRisincluded.
Toenhancetherobustnessofthesefindings,wefurthertestwhetherthereisadifferencebetweenthecoefficientsoftheliquidityproxiesforsmall-capandlarge-capfirmsbyperformingtheWaldtestsinTable5.Foreachmeasure,wefirsttestthenullhypothesisthatthedifferencebetweentheliquidityproxyforsmall-andlarge-capfirmsisequaltozero(PanelA)andthentestthenullhypothesisthatthedifferencebetweentheilliquiditymeasureforsmall-andlarge-capfirmsisgreaterthanzero(PanelB).Theresultsobtainedindicatethatinbothcountriesandacrossallmacrovariables,theliquiditycoefficientofthesmall-capfirmsisstatisticallysignificantdifferentfromtheliquiditycoefficientofthelarge-capfirmsandthattheliquiditycoefficientofthesmall-capfirmsisstatisticallysignificantlargerthattheliquiditycoefficientofthelarge-capfirms,indicatingthelargereffectofsmall-capfirmliquidityinexplainingeconomicactivity.Thisfindingisinagreementwithanumberofpreviousstudies(e.g.,Amihud,2002;Næsetal.,2011;CakiciandTan,2014).Thus,itisimportantforinvestorsandcentralbankauthoritiestowatchcloselyforalargedropintheliquidityofsmallfirmsstocks,aslongasitgivesastrongsignalforthebeginningofarecessionaryperiod.Asinvestorsmaystartswitchingfromtheirpositionsonsmallcapstockstogovernmentbondsorlargecapstocks,centralbanksmayincreasepromptlythemoneysupplyaimingtostimulatetherealeconomyandhencetoavoidplunginginadeepandprolongedrecession.
4.2.Causalitytests
Thecausalityresults,reportedinTable6,displaythatinboththeUKandtheGermanstockmarketsthereexistsaone-waycausalityrunningfromboththeilliquidityratio(ILR)andtherelativespread(RS)toallfouralternativemacroeconomicvariables(MACRO).Thesefindingsarecompatiblewiththeforward-lookingnatureofliquidityrelatedtothestockmarketasdocumentedbyFama(1991).Inaddition,itrecognizestheimportanceofmarketliquidityasacomponentofthefinancialsystemintheprocessofeconomicgrowth,gainingfurthersupportbytheargumentputforwardbyLevineandZervos(1998).
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
61
Table9
CausalityresultsforUK&Germanybetweensmallandlargefirms’liquiditymeasures.
UK
Germany(+)19.64*(0.00)(+)3.92(0.16)(+)20.75*(0.00)(+)3.38(0.24)(+)29.07*(0.00)(+)3.75*(0.18)(+)36.83*(0.00)(+)3.62(0.20)
Illiquidityratio
H0:ILRSILRL2-testp-Value
H0:ILRLILRS2-testp-Value
Relativespread(RS)H0:RSSRSL2-testp-Value
H0:RSLRSS2-testp-ValueTurnover(TUR)
H0:TURSTURL2-testp-Value
(+)24.93*(0.00)(+)3.51(0.19)
(+)23.63*(0.00)(+)2.74(0.35)
(+)26.93*(0.00)
H0:TURLTURS2-testp-ValueVolume(VTR)
H0:VTRSVTRL2-testp-Value
H0:VTRLVTRS2-testp-Value
(+)2.16(0.41)
(+)30.17*(0.00)(+)2.27(0.37)
ThetableshowsGrangercausalitytestsbetweenthe(a)AmihudILRofsmallandlargefirms,(b)relativespread(RS)ofsmallandlargefirms,(c)turnover(TUR)ofsmallandlargefirms,and(d)volumetraded(VTR)ofsmallandlargefirms.Thecross-sectionalliquiditymeasuresarecalculatedasequallyweightedaveragesacrossstocks.ThetestisperformedforthewholesampleperiodfortheUKfrom1994to2011andforGermanyfrom1999to2011.Foreachmeasure,wefirsttestthenullhypothesisthattheliquiditymeasureforsmallfirmsdoesnotGrangercauseliquidityforlargefirmsandwhetherliquidityforlargefirmsdoesnotGrangercauseliquidityforsmallfirms.Theresultsinparenthesesdenotethesignoftheassociationbetweenthevariablesunderinvestigation.WereportF-testvaluesandcorrespondingp-values(inparentheses)foreachtest.*
Denotesarejectionofthenullhypothesisofcausalityatthe1%level.
Inparticular,marketliquidityisimportanttorestoretheconfidenceofinvestorsinthevalueofinformationassociatedwithtrading.Therefore,investors,inthecaseofincreasesinmarketliquidity,areencouragedtoinvestmoreheavily,increasingtheflowofcapital,leadingtofurtherincreasesineconomicactivity.
Bycontrast,fortheturnoverandthetradingvolumeliquidityindicesthisone-waycausalityholdsonlyinthecasesofinvestmentandtheunemploymentrateinbothcountries.Thepresenceoftwo-waycausalitydenotesthatthegrowthprocessitselfnurturesthegrowthoffirmsand,consequently,increasesthereturnstothosebusinesses;asaresult,highergrowthlevelsprovideaboostbothtothestockmarketandtotheliquiditymeasuresassociatedwithit.
OurfindingstendtobeconsistentwiththefindingsbyNæsetal.(2011)whodocumentthatforthewholesampleperiodthereisfullsupportforGrangerone-waycausalityfrommarketilliquidity/liquidityproxies(ILR,implicitspreadestimatorofRoll,theLOTmeasure)toGDP,whilenoreversecausalityfromGDPtotheseproxiesispresent.
TheseresultsarealsoconsistentwiththeargumentputforwardbyLevineandZervos(1998)andRousseauandWachtel(2000),accordingtowhich,marketliquidityisrelatedtoeconomicgrowth.Asinvestorsareencouragedbyhighmarketliquiditytoinvestinequitiesandincreasetheflowofcapital,suchmoveswouldefficientlyallocateresourcesand,hence,enhanceeconomicgrowthinthelongrun.Thecausalityempiricalfindingsdisplaythatstockmarketscaneffectivelymobilizefundsthathavebeennotfullyabsorbedbyfinancialintermediariesintoproductiveinvestmentsand,hence,spureconomicgrowth.
ThecausalitytestswerealsoperformedseparatelyforsmallandlargesizedcompaniesandtheresultsarereportedinTables7and8,respectively.Ourinitialfindingsconcerningtheonewaycausalityrunningfromtheilliquidityratio(ILR)andtherelativespread(RS)toallfourmacroeconomicvariablesarevalidatedfromtheanalysisofsmallcapfirmsonlyforthecaseofILR.
Weconsiderthatthisfindingmaybejustifiedonthebasisoftheargumentthatsmallfirmsaremorevulnerabletonegativemacroeconomicshocksthanlargefirmsandthustheirliquidityisfirstlyaffectedbyadversemacroeconomicconditions,eveniftheirdropinliquidityhasmorepredictivepowerfortheoncomingrecessionthanthatofthelargefirms.ThisisconsistentwiththefindingsprovidedbyGertlerandGilchrist(1994)whoreportthatsmallfirmsaremoresensitivethanlargefirmstonegativechangesinthemacroeconomicenvironment(e.g.,theirsalesdeclinefasterfollowingacontractioninbanklendingaswellasafallinGDP).Thisimpactisstrongerforsmallfirmsandisconsummatedthroughboththebank
62
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
lendingandborrowingchannels,asKhwajaandMian(2008)showexaminingthecaseofabankliquiditycrisisinPakistanin1998.Therefore,smallfirmswouldbethefirstwhowillexperiencetheeffectsofacreditcrunchandacontractionaryeconomy.
Lookingatthecausalitytestsresultsobtainedforthesampleofonlylargefirms(Table8)wecanobservethataclearone-waycausalitylinkexistsforallliquidityandmacroeconomicvariablesexaminedacrossboththeUKandGermany.Thus,itseemsthattheone-waycausality,whichisprevalentinthesampleoflargefirms,dominatesthemarginaltwo-waycausalityrelationshipobservedinthesampleofthesmallfirms.
Finally,asrobustnesstestsweperformedGrangercausalitytestsbetweensmallandlargecapliquidityproxies(whenbothvariableshavebeenexplicitlyusedintheregressions)acrossallfourliquiditymeasuresandtheresultsarereportedinTable9.Wefocusonlyonthecausalityresultsacrossthetwotypesoffirmsandthefindingsdocumentthatacrossbotheconomiesandacrossliquiditymeasures,thereexistsaunivariatecausalityrunningfromtheliquidityofsmallfirmstothatoflargefirms.Inotherwords,liquidityinnovationsinsmallfirmsareinformativeinpredictingliquidityshiftsinlargefirms,whiletheotherwayarounddoesnotseemtohold.Thisasymmetrysuggeststhatincaseofstressfulmacroeconomicenvironmentsthereexistsrelativelyamoreactiveroleofthesmallcapsectorinpropagatingmarket-wideshocks.Transac-tionsinresponsetoeconomiceventsoccurfirstinsmallstocks,andthenspilloverovertolargestocks.Inthatcase,marketparticipants/investorschoosetoexploitanyinformationaladvantagetheyareholdinginthesmall-capsector,whichmaydriveliquidity/orilliquiditymeasuresandmaycauselargefirms’liquiditytolagtothatofsmallfirms.
5.Conclusions
Thepresentresearchstudiedtheinformationcontentofstockmarketliquidityoneconomicconditionsintwohighlycapitaldevelopedcountries,theUKandGermany.Theempiricalresultssuggestedthatstockmarketliquidityandeconomicindicatorswerestronglyassociatedacrossbotheconomies,eventhoughtheUKisacapitalmarket-basedeconomyandGermanyabank-basedeconomy.Thesefindingshighlightthatthemechanismofchannellingfundstotheeconomy(viamarketsorviabanks)doesnotmatter,atleastformaturemarkets,whilethereisnotanydifferentialroleofliquidityinexplainingthecourseofmacroeconomicvariables.However,theliquidityofsmall-capfirmsprovedtobemoreimportantthanlarge-capfirmliquidity,inbothcountries,revealingthatinvestorsreallocatetheirportfoliosfromilliquidsmall-capriskystockstolessriskylarge-capstocks(the“flighttoquality”effect)withhigherliquidity,whentheexpectationsforthefuturestateoftheeconomychange.
Ourresultscouldhelppolicymakersandcorporatemanagersimproveresourceallocation,sincetheybothcanbeconfi-dentenoughinmakinguseofstockmarketliquiditytoreachdecisionsthatrelyheavilyuponeconomicactivity.Inaddition,providingmorestrengthtotheliquidityprofileofacapitalmarketisexpectedtoreducetheriskassociatedwithinvestments,asthiswillallowsaverstoacquireequitiesandsellthemreallyfastandwithoutincurringhighcostsiftheyneedtomakechangesintheirportfolios.Anenhancedmodeofstockmarketliquiditywillalsoimprovecapitalallocations,thus,leadingtofurtherinvestments.Furthermore,firmswithilliquidmarketsfortheirequitytendtobemoreexposedtoanumberofexternaltothemshocks.Therefore,byprovidingliquiditytothestockmarketisexpectedtosubstantiallyreduceliquidityrisksfacedbyinvestorsand,moreover,thiswillleadtoequitycapitalcostreductionsinfuturefundraisingandfutureeconomicgrowth(Levine,1991;LernerandSchoar,2004).
Whatshouldpolicymakersdoinordertopushstockmarketliquidity?Theyshouldremoveanypotentialimpediments,suchastax,legalandregulatorybarrierstostockmarketdevelopment.Duringacrisiseventwecannotignoretheroleofmonetaryauthoritiesintheprocessofimplementinganefficientmonetarypolicywithrespecttomaintainingstablefinancialmarkets,giventhatduringacrisis,stockmarketsarehighlyilliquid.Therefore,theroleofthecentralbanktomaintaintherequiredliquiditylevelsisamajoronebyprovidingliquiditytothosemarketsanddirectlytolargeinvestorsthatholdlongpositionsinstocks.However,asarecentstudybyFlorackisetal.(2014b)showed,traditionaltoolsofmonetarypolicy(e.g.,cutsininterestrates)maynotbeeffective,atleastintheUK,sincetheywereperceivedbythemarketparticipantsasasignalofdeterioratingeconomicprospectsandinfactaggravatedthe“flighttosafety”effect,tradingeitherfromstockstogovernmentbondsorfromlessliquid(e.g.,smallcap)stockstomoreliquidones(largecapstocks).
Theneedforfurtherresearchisobviousbyincludingmorecountrieswithdevelopedcapitalmarkets,and/orcountrieswithlessdevelopedcapitalmarketsaroundtheglobeinordertoobtainmoreevidenceontheactualrelationshipbetweenstockmarketliquidityandeconomicconditions.Moreover,futureresearchattemptscouldidentifythosefactorsthataffectdirectlystockmarketliquidityand,therefore,indirectlymacroeconomicactivity,suchas,legal,regulatory,accounting,tax,political,andmacroeconomicfactors.Forexampleoneofthesefactorsmaybetheanalysts’forecasterrorsandfutureresearchcanexamineitsrelationshipwithbothmarketliquidityandthemacroeconomicenvironment.
References
Abraham,A.,Seyyed,F.,Alsakran,S.,2002.TestingtherandombehaviorandefficiencyoftheGulfstockmarkets.Financ.Rev.37(3),469–480.Acharya,V.V.,Pedersen,L.H.,2005.Assetpricingwithliquidityrisk.J.Financ.Econ.77(2),375–410.Akaike,H.,1969.Fittingautoregressivemodelsforprediction.Ann.Inst.Stat.Math.21(2),243–247.
Allen,F.,Carletti,E.,2008.Therolesofbanksinfinancialsystems.In:Berger,A.,Molyneux,P.,Wilson,J.(Eds.),TheOxfordHandbookofBanking.Oxford
UniversityPress.
Allen,F.,Gale,D.,2000.ComparingFinancialSystems.MITPress,Cambridge,MA.
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
63
Amihud,Y.,2002.Illiquidityandstockreturns:cross-sectionandtime-serieseffects.J.Financ.Markets5(1),31–56.Amihud,Y.,Mendelson,H.,1986.Assetpricingandthebid-askspread.J.Financ.Econ.17(2),223–249.
Beber,A.,Brandt,M.W.,Kavajecz,K.A.,2011.Whatdoesequitysectororderflowtellusabouttheeconomy?Rev.Financ.Stud.24(11),3688–3730.Beck,T.,Levine,R.,2002.Industrygrowthandcapitalallocation:doeshavingamarket-orbankbasedsystemmatter?J.Financ.Econ.64,147–180.Bhide,A.,1993.Thehiddencostsofstockmarketliquidity.J.Finan.Econ.34,1–51.
Blanchflower,D.G.,Oswald,A.J.,2013.Doeshighhome-ownershipimpairthelabormarket?NBERWorkingPaper,No.19079.NationalBureauofEconomic
Research.
Boot,A.,Thakor,A.,1997.Financialsystemarchitecture.Rev.Financ.Stud.10,693–733.
Brennan,M.,Subrahmanyam,A.,1996.Marketmicrostructureandassetpricing:onthecompensationforilliquidityinstockreturns.J.Financ.Econ.41(3),
441–464.
Brunnermeier,M.K.,Pedersen,L.H.,2009.Marketliquidityandfundingliquidity.Rev.Financ.Stud.22(6),2201–2238.
Cakici,N.,Tan,S.,2014.Size,value,andmomentumindevelopedcountryequityreturns:macroeconomicandliquidityexposures.J.Int.MoneyFinance
44,179–209.
Campbell,Y.J.,Mankiw,G.N.,1989.Internationalevidenceonthepersistenceofeconomicfluctuations.J.MonetaryEcon.23(2),319–333.
Caporale,G.M.,Howells,P.G.A.,Soliman,A.M.,2004.Stockmarketdevelopmentandeconomicgrowth:thecausallinkage.J.Econ.Dev.29(1),33–50.
Cassola,N.,Holthausen,C.,LoDuca,M.,2009.The2007/2009turmoil:achallengefortheintegrationoftheeuroareamoneymarket?EuropeanCentral
BankWorkingPaper.
Chordia,T.,Sarkar,A.,Subrahmanyam,A.,2004.Liquiditydynamicsacrosssmallandlargefirms.EconomicNotesbyBancaMontedeiPaschidiSienaSpA.
33,1,pp.111–143.
Chordia,T.,Sarkar,A.,Subrahmanyam,A.,2005.Anempiricalanalysisofstockandbondmarketliquidity.Rev.Financ.Stud.18(1),85–129.Coulson,N.E.,Kim,M.S.,2000.Residentialinvestment,non-residentialinvestmentandGDP.RealEstateEcon.28(2),233–247.Estrella,A.,Hardouvelis,G.A.,1991.Thetermstructureasapredictorofrealeconomicactivity.J.Finance46(2),555–576.
Estrella,A.,Mishkin,F.S.,1998.PredictingUSrecessions:financialvariablesasleadingindicators.Rev.Econ.Stat.80(1),45–61.Fama,E.F.,1991.Efficientcapitalmarkets:II.J.Finance46(5),1575–1617.
Fama,E.F.,French,K.R.,1992.Thecross-sectionofexpectedstockreturns.J.Finance47,427–465.
Fama,E.F.,French,K.R.,1996.Multifactorexplanationsofassetpricinganomalies.J.Finance51,55–84.
Ferguson,R.W.,Hartmann,P.,Panetta,F.,Portes,R.,2007.Internationalfinancialstability.In:GenevaReportsontheWorldEconomy.InternationalCentre
forMonetaryandBankingStudies.
Florackis,C.,Giorgioni,G.,Kostakis,A.,Milas,C.,2014a.Onstockmarketilliquidityandreal-timeGDPgrowth.J.Int.MoneyFinance44,210–229.
Florackis,C.,Kontonikas,A.,Kostakis,A.,2014b.Stockmarketliquidityandmacro-liquidityshocks:evidencefromthe2007–2009financialcrisis.J.Int.
MoneyFinance44,97–117.
Fujimoto,A.,2003.Macroeconomicsourcesofsystematicliquidity.In:WorkingPaper.YaleUniversity.
Gertler,M.,Gilchrist,S.,1994.Monetarypolicy,businesscycles,andthebehaviorofsmallmanufacturingfirms.Q.J.Econ.109(2),309–340.
Gibson,R.,Mougeot,N.,2004.Thepricingofsystematicliquidityrisk:empiricalevidencefromtheU.S.stockmarket.J.Bank.Finance28(1),157–178.Goyenko,R.Y.,Holden,C.W.,Trzcinka,C.A.,2009.Doliquiditymeasuresmeasureliquidity?J.Financ.Econ.92(2),153–181.
Goyenko,R.Y.,Ukhov,A.D.,2009.Stockandbondmarketliquidity:along-runempiricalanalysis.J.Financ.Quant.Anal.44(1),189–212.
Green,R.K.,1997.Followtheleader:howchangesinresidentialandnon-residentialinvestmentpredictchangesinGDP.RealEstateEcon.25(2),253–270.Guo,H.,Mortalb,S.,Savickasc,R.,Woodd,R.,2011.Uncoveringtherelationbetweenaggregatestockilliquidityandexpectedexcessmarketreturns.In:
Workingpaper.UniversityofCincinnati.
Hamid,K.,Suleman,M.T.,Shah,S.Z.A.,Akash,R.S.I.A.,2010.Testingtheweakformofefficientmarkethypothesis:empiricalevidencefromAsia–Pacific
markets.Int.Res.J.FinanceEcon.58(2),121–133.
Hasbrouck,J.,2009.TradingcostsandreturnsforUSequities:estimatingeffectivecostsfromdailydata.J.Finance64(3),1445–1477.Hodrick,R.J.,Prescott,E.C.,1997.PostwarUSbusinesscycles:anempiricalinvestigation.J.Money,Credit,Bank.24(1),1–16.
Hui,E.C.M.,Yiu,C.Y.,2003.Marketdynamicsofprivateresidentialrealestateprice–anempiricaltestinHongKong.J.Financ.Manage.PropertyConstr.8
(3),155–165.
Iacoviello,M.,2003.Consumption,housepricesandcollateralconstraints:astructuraleconomicanalysis.J.House.Econ.13(4),304–320.Jones,C.,2002.Acenturyofstockmarketliquidityandtradingcosts.In:WorkingPaper.ColumbiaUniversity.
Karamujic,H.M.,2013.Buildingapprovalsasaleadingindicatorofpropertysectorinvestment.Int.J.Bank.Finance9,44–58.
Kaul,A.,Kayacetin,V.,2009.Forecastingeconomicfundamentalsandstockreturnswithequitymarketorderflows:macroinformationinamicromeasure?
In:WorkingPaper.UniversityofAlberta.
Kempf,A.,Mayston,D.,2008.Liquiditycommonalitybeyondbestprices.J.Financ.Res.31(1),25–40.
Khwaja,A.I.,Mian,A.,2008.Tracingtheimpactofbankliquidityshocks:evidencefromanemergingmarket.Am.Econ.Rev.98(4),1413–1442.
Kim,D.,Perron,P.,2009.Unitroottestsallowingforabreakinthetrendfunctionunderboththenullandthealternativehypotheses.J.Econ.148(1),1–13.Kiyotaki,N.,Moore,J.H.,2008.Liquidity,BusinessCycles,andMonetaryPolicy.In:WorkingPaper.PrincetonUniversity.LaPorta,R.,Lopez-de-Silanes,F.,Shleifer,A.,2002.Governmentownershipofcommercialbanks.J.Finance57,265–301.Lakonishok,J.,Shleifer,A.,Vishny,R.,1994.Contrarianinvestment,extrapolation,andrisk.J.Finance49,1541–1578.Lerner,J.,Schoar,A.,2004.Theilliquiditypuzzle:theoryandevidencefromprivateequity.J.Financ.Econ.72(1),3–40.
Lesmond,D.A.,Ogden,J.P.,Trzcinka,C.A.,1999.Anewestimateoftransactioncosts.Rev.ofFinanc.Stud.12(5),1113–1141.Levine,R.,1991.Stockmarkets,growth,andtaxpolicy.J.Finance46(4),1445–1465.
Levine,R.,Zervos,S.,1998.Stockmarkets,banks,andeconomicgrowth.Am.Econ.Rev.88(3),537–558.Longstaff,F.A.,2004.Theflight-to-liquiditypremiuminU.S.TreasuryBondprices.J.Bus.77(3),511–526.
Lu,R.,Glascock,J.,2010.Macroeconomiceffectsonstockliquidity.In:WorkingPaper.UniversityofCincinnati.
Meichle,M.,Ranaldo,A.,Zanetti,A.,2011.Dofinancialvariableshelppredictthestateofthebusinesscycleinsmallopeneconomies?Evidencefrom
Switzerland.Financ.MarketsPortfolioManage.25(4),435–453.
Morck,R.,Nakamura,M.,1999.BanksandcorporatecontrolinJapan.J.Finance54,314–340.
Næs,R.,Skjeltorp,J.A.,Ødegaard,B.A.,2011.Stockmarketliquidityandthebusinesscycle.J.Finance66(1),139–176.
Nelson,C.R.,Plosser,C.I.,1982.Trendsandrandomwalksinmacroeconomictimeseries:someevidenceandimplications.J.MonetaryEcon.10(1),139–162.Nyborg,K.G.,Östberg,P.,2014.Moneyandliquidityinfinancialmarkets.J.Financ.Econ.112,30–52.
Papavassiliou,V.,2013.Anewmethodforestimatingliquidityrisk:insightsfromaliquidity-adjustedCAPMframework.J.Int.Financ.MarketsInst.Money
24,184–197.
Pastor,L.,Stambaugh,R.F.,2003.Liquidityriskandexpectedstockreturns.J.Pol.Econ.111(3),642–685.
Perez-Quiros,G.,Timmermann,A.,2000.Firmsizeandcyclicalvariationsinstockreturns.J.Finance55,1229–1262.Rajan,R.,1992.Insidersandoutsiders:thechoicebetweeninformedandarm’s-lengthdebt.J.Finance47,1367–1400.
Roll,R.,1984.Asimpleimplicitmeasureoftheeffectivebid-askspreadinanefficientmarket.J.Finance39(4),1127–1139.
Rousseau,P.L.,Wachtel,P.,2000.Equitymarketsandgrowth:cross-countryevidenceontimingandoutcomes,1980–1995.J.Bank.Finance24(12),
1933–1957.
Rudebusch,G.D.,Williams,J.C.,2009.Forecastingrecessions:thepuzzleoftheenduringpoweroftheyieldcurve.J.Bus.Econ.Stat.27(4),492–503.Rösch,C.G.,Kaserer,C.,2012.Theroleofliquiditycommonalityandflight-to-quality.In:WorkingPaper.TechnicalUniversityofMunich.Sims,C.A.,Stock,J.H.,Wallace,M.W.,1990.Inferenceinlineartimeseriesmodelswithsomeunitroots.Econometrica58(1),113–144.
64
N.Apergisetal./Int.Fin.Markets,Inst.andMoney38(2015)42–64
Stiglitz,J.E.,1985.Creditmarketsandthecontrolofcapital.J.MoneyCreditBank.17,133–152.
Stock,J.,Watson,W.,2003.Forecastingoutputandinflation:theroleofassetprices.J.Econ.Lit.41(3),788–829.
Switzer,L.N.,2010.Thebehaviourofsmallcapvs.largecapstocksinrecessionsandrecoveries:empiricalevidencefortheUnitedStatesandCanada.North
Am.J.Econ.Finance21(3),332–346.
Söderberg,J.,2008.Domacroeconomicvariablesforecastchangesinliquidity?Anout-of-samplestudyontheorder-drivenstockmarketsinScandinavia.
In:WorkingPaper.VaexjoeUniversity.
Toda,H.Y.,Phillips,P.C.B.,1993.Vectorautoregressionandcausality.Econometrica61(6),1367–1393.
Toda,H.Y.,Yamamoto,T.,1995.Statisticalinferenceinvectorautoregressionswithpossiblyintegratedprocesses.J.Econom.66(1),225–250.Tadesse,S.,2002.Financialarchitectureandeconomicperformance:internationalevidence.J.Financ.Intermediation11,429–454.
Worthington,A.,Higgs,H.,2004.RandomwalksandmarketefficiencyinEuropeanequitymarkets.GlobalJ.FinanceEcon.12(1),59–78.Wright,J.H.,2006.Theyieldcurveandpredictingrecessions.FinanceandEconomicsDiscussionSeries,FederalReserveBoard.
因篇幅问题不能全部显示,请点此查看更多更全内容