ContentslistsavailableatScienceDirect
Journal
of
International
Financial
Markets,
Institutions
&
Money
j o u r n a l h o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / i n t f i n
Informed
trading,
trading
strategies
and
the
information
content
of
trading
volume:
Evidence
from
the
Taiwan
index
options
market
Wen-liang
G.
Hsieh
∗,
Huei-Ru
He
GraduateInstituteofFinance,NationalChiaoTungUniversity,1001UniversityRoad,HsinchuCity300, Taiwan,ROC
a
r
t
i
c
l
e
i
n
f
o
Articlehistory:Received30May2013 Accepted27March2014 Availableonline5April2014 JELclassification: G14 Keywords: Indexoptions Optionsvolume Informedtrading
Foreigninstitutionalinvestors Taiwan
a
b
s
t
r
a
c
t
Thispaperexaminesthepredictiveabilityofindexoption put-callvolumeonnext-dayindexmovementsintheTaiwanmarket. Wefindthatforeigninstitutionalinvestorsarethemostinformed traders,with theirpredictive abilitybeingmoreapparent in a downwardmarket.Whenengagingininformedtrading,foreign institutionalinvestorstend touseout-of-the-moneyoptionsto achievehighleverage,alongwithmedium-termoptionstoobtain largedeltaexposure andlow thetarisk,whilst also sacrificing liquiditybyforgoingtheuseofshort-termoptions.Thepredictive abilityofforeigninstitutionalinvestorsisfoundtobesignificantly enhancedondayswithimportantmacroeconomicnews,thereby indicatingtheirsuperiorinterpretativeabilityofpubliclyaccessible information.Basedupontheirlong-livedinformationaladvantage, foreigninstitutional investors willtend toengage in informed tradingusinglimitordersandmedium-sized tradesinorderto camouflagetheirinformation.
©2014ElsevierB.V.Allrightsreserved.
∗ Correspondingauthor.Tel.:+88635712121.
E-mailaddresses:wlh@faculty.nctu.edu.tw(W.-l.G.Hsieh),tina65762003@yahoo.com.tw(H.-R.He). http://dx.doi.org/10.1016/j.intfin.2014.03.012
1. Introduction
Wesetoutinthisstudytoexaminetheinformationcontentoftradingvolume,withtheoverallaim
ofidentifyingthepatternsofindexoptionsusageininformedtrading.Theextantliteratureprovides
evidenceontheinformativenessofoptionstrading,withsomestudiesidentifyingwhichtradersarein
possessionofprivatetradinginformation.Inthepresentstudy,wegoonestepfurthertoexaminethe
varioustradingdecisionsfacedbyinvestorswhopossesssuperiorinformation,aswellasthetrading
strategiesthattheychoosetoadopt.Inspecificterms,weinvestigatethecontractsinwhichinformed
traderstendtotrade(in-the-moneyversusout-of-the-money),thetypeofordersthattheyusefor
theirtradingactivities(marketversuslimitorders,smallversuslargeorders),thetypeofinformation
thattheyarelikelytoexploit(macroversusmicro,globalversusdomestic)andthemarketconditions
underwhichtheychoosetotrade.
It hasbeensuggestedinseveralofthepriorstudiesthattheoptionsmarketsattractinformed
tradingessentiallybecauseinvestorscanbenefitfromhighleverage(Black,1975),lowtransaction
costs(Flemingetal.,1996)andtheflexibilitytoengageinavarietyoftradingstrategiesthatare
unavailabletotheminthespotmarket.1Wheninvestorschoosetoengageininformedtradingwithin
theoptionsmarkets,thetradingprocessmaygeneraterichinformationcontentonfuturestockprices;
indeed,itiswelldocumentedthatthepricesofoptionsplayaleadingroleinpricediscovery,2and
thattheyareevencapableofpredictingfuturepricemovements;3thus,ourfocusinthepresentstudy
isontheinformationcontentofoptionstradingvolume.
Easleyetal.(1998)assertedthatobservedtransactionsplayanimportantroleinpricediscovery,
essentiallybecauseorderflowimbalancescanreflectthesignandmagnitudeofprivateinformation.
Theyproposedamodelwhichrevealedthatundercertaincircumstances,thesignedtradingvolume
ofoptionscontainedvaluableinformationonfutureequityprices.Caoetal.(2005)subsequently
wentontoprovidedirectevidenceofoptionsvolumeplayingastrongerinformationalroleduring
periodswheninformedtradingwasparticularlyintensive,withtheirfindingsrevealingthatduring
takeoverannouncementperiods,imbalancesincallvolumehadstrongpredictiveabilityonnext-day
stockreturns.PanandPoteshman(2006)furtherproposedthattheput-callratioofoptionstrading
volumeinitiatedbybuyersopeningnewpositionsshouldbetakenasaninformationvariable,since
thevolumeratiowasfoundtopredictstockreturnsforthenextfivedays,withbotheconomicand
statisticalsignificance.
Withineachoftheabovestudiesontheinformationalcontentofoptionstradingvolume,the
ten-dencyhasbeentofocusalmostexclusivelyonequityoptions,withmuchlessemphasisonindex
options.Asregardsoptionsonindividualstocks,itisalreadywellrecognizedthatcorporateinsiders
andproprietaryfirmtraders(i.e.,thosewhoareoftenfoundtopossessprivatefirm-specific
informa-tion),arelikelytoengageininformedtrading(CornellandSirri,1992;Harris,1993).However,with
regardtoindexoptions,itisfarlessclearwhichclassesoftraderpossesssuperiorinformation,and
indeed,exactlywheresuchinformationoriginatesfrom,essentiallybecauseitisunlikelythatany
investorwouldpossess‘private’informationattheaggregatemarketlevel.
However,Ahnetal.(2008)arguedthatsuperiorinformationprocessingskillsanddifferent
inter-pretationsofexactlythesamepublicinformationmaywellresultininformationasymmetry.They
notedthatforeigninstitutionalinvestorswerelikelytopossessaninformationaladvantageoverother
typesofinvestors,andindeed,theydemonstratedaparticularlylargeadverseselectioncomponent
inthebid-askspreadsinKOSPI200optionsforthosetradesthathadbeeninitiatedbysuchinvestors.
ChouandWang(2009)alsoidentifiedacleartendencyforstealthtradingamongstforeign
institu-tionalinvestorsandproprietaryfirmsintheindexfuturesmarketinTaiwan,afindingwhichimplied
thatcertainclassesoftradersmaywellpossessinformationaladvantagesrelatingtotheaggregate
markettrend.Nevertheless,giventheabovefactors,allofwhichprovidesupportforthelikelihoodof
1Optionstradingstrategiesunavailabletospotmarkettradersincludevolatilitytrading(Nietal.,2008),spreadtrading
(ChaputandEderington,2005)andunlimitedshortsales(FiglewskiandWebb,1993).
2ExamplesincludeFlemingetal.(1996),Easleyetal.(1998)andCaoetal.(2005). 3SeeChakravartyetal.(2004),PanandPoteshman(2006)andChanetal.(2009).
informedtradersusingindexoptions,thereissurprisinglylittledirectevidenceofinformationtrading
intheindexoptionsmarkets.
KangandPark(2008)reportednon-trivialinformationcontentfornetbuyingpressure(the
differ-encebetweenthenumberoftradesinitiatedbybuyersandsellers)inKOSPI200indexoptions,and
wentontosuggestthatthenetbuyingpressurecouldbeusedtopredictthenextfive-minutereturns
ontheindex.Chanetal.(2009)alsoexaminedoptionstradingbehaviorintheTaiwanindex,usingthe
valueoftheput-callratioasaninformationvariable,andfoundthatout-of-the-moneyoptionsledthe
stockindexbyupto90min.Furthermore,fromanexaminationofthepredictiveabilityoftheput-call
open-buyvolumeratiosonthenext-daymovementsintheTaiwanindex,Changetal.(2009)found
significantpredictiveabilityforcertaintypesofoptionstradedbyforeigninstitutionalinvestors.4
Theempiricalevidenceprovidedbytherecentstudiesreferredtoabovethereforeseemstosuggest
thattradingvolumeplaysamuchmoreaggressiveinformationalroleinindexoptionsthanhad
pre-viouslybeensuggestedwithintheearlierstudies.5Thepresentstudycanbeviewedasanextension
oftheChangetal.(2009)studyinwhichevidencewasfoundofpredictivepowerinTaiwanindex
optionsamongstforeigninstitutionalinvestors,particularlywithregardtotheirtradingpositionsin
near-the-moneyandmiddle-horizonoptions.
Ourprimaryaiminthepresentstudyistodeterminewhethertheinformationcontentinthe
optionstradingundertakenbyforeigninstitutionalinvestors isdependentuponmarketliquidity,
thedirectionoftheindexmovementsandtheimpactofmacroeconomicinformation,whilstalso
exploringthetypesoforders(marketversuslimitorders)andordersizethatarelikelytobeused
bysuchinvestorswhenengagingintheirinformedtrading.Asoundunderstandingofthesefactors
shouldhelpustodrawamorecompletepictureoftheinformationaladvantagespossessedbyforeign
institutionalinvestorsandthewaysinwhichtheyprofitfromtheirinformation.
TheempiricalmodeladoptedforthepresentstudyfollowsthemethodologyofPanandPoteshman
(2006)inwhichthepredictiveabilityofoptionsvolumewasinvestigatedbyregressingthe
next-dayindexreturnsontheoptionput-callvolumeratio.We beginbyclassifyingthevolumeofall
optionstradingattributabletofourclassesoftraders,namely,individualinvestors,marketmakers,
foreigninstitutionalinvestorsanddomesticinstitutionalinvestors,andthenexaminethepredictive
abilityoftheput-callvolumeratioonnext-dayindexreturnsforeachoftheseclassesoftraders.Our
resultsclearlyindicatethatthetradingvolumeprovidedbyforeigninstitutionalinvestorscontains
richinformationrelatingtofuturechangesintheindex,whereastransactionsbyotherclassesof
tradersarelargelyfoundtobeuninformative.
Thisfindingisgenerallyconsistentwiththeresultsreportedontheaggressiveinformationalrole
playedbyforeigninstitutionalinvestorsinmanyoftheemergingderivativemarkets.Forexample,the
analysiscarriedoutbyChangetal.(2009)onindexoptionstradinginTaiwanshowedthattradingby
foreigninstitutionalinvestorsexhibitedsignificantpredictivepowerontheunderlyingindexreturns.
SimilarfindingswerealsoprovidedbyKangandPark(2008),Ahnetal.(2008)andLeeandWang
(2012)withregardtotradinginKOSPI200indexoptions.
Wealsodiscoverstrongerpredictivepowerinthetradingpositionsofforeigninstitutionalinvestors
indownwardmarketsvis-à-visupwardmarkets.ThisisconsistentwiththesuggestionofJohnson
andSo(2012)thattheshort-salesconstraintsintheequitymarkettendtohindertransactionsbased
uponbearishinformation;thus,informedagentswilltradeinoptionsmorefrequentlyonnegative
signalsthanpositivesignalsinordertoovercometheshort-salesrestrictionsontheunderlyingasset.
AlthoughtheaboveargumentpresentedbyJohnsonandSo(2012)wasbaseduponequityoptions,
4Severalotherstudieshaveexaminedinformationaltradinginindexoptions.Forexample,focusingontheinformation
providedbysplitorders,KimandRyu(2012)demonstratedthatsuchorderssubmittedbyinstitutionalinvestorsgenerally tendedtobemoreinformative,whilstChangetal.(2013)examinedthepre-openingtradingofindexoptionsandshowedthat information-motivatedtraderstendedtoconstructtheirpositionspriortothestockmarketopening,particularlywhenthey possessedbearishinformation.
5SchlagandStoll(2005)foundthatthepriceimpactofoptionsvolumeontheDAXindexwasonlyatemporaryphenomenon,
therebyimplyingthepresenceofaliquidityeffectasopposedtoaninformationeffect.PanandPoteshman(2006)alsoshowed thatindexoptionstradingvolumecontainednoinformationonfutureindexmovements,whereasequityoptionstrading volumehadstronginformationcontentonfuturestockpricemovements.
ourstudyprovidesevidenceofshort-salesconstraintshavingasimilareffectoninformedtradingin
indexoptions.
Ourresultsfurtherrevealthatwithimprovementsinoveralloptionstradingactivity(asmeasured
bytradingvolume),thetradingpositionsofforeigninstitutionalinvestorsprovideenhanced
predic-tivepower.Thefindingsuggeststhateitherinformedtradinginducesvolumeorthatinformedtraders
arecamouflagingtheirinformation-basedtradingthroughgreatervolume.Furtherevidence
consis-tentwithstealth-tradingtheoryisfoundwhenweanalyzesub-samplescategorizedbytradesizes,
withmedium-sizedtradescontainingricherinformationcontentonfuturechangesintheindex,as
comparedtoeitherlarge-orsmall-sizedtrades.
Asopposedtotradinginthemostliquidshort-termcontracts,foreigninstitutionalinvestorsprefer
toexploittheirinformationaladvantageusingmedium-maturityandout-of-the-moneyoptions.The
decisionsmadebyinformedtraderswithregardtocontractselectionsuggestthatsuchtradersare
willingtosacrificeliquidityforhighleverage,highdeltaandlowtheta.
Limitorderssubmittedbyforeigninstitutionalinvestorsarefoundtohavegreaterpredictiveability
onthenextdayindexvalue,relativetomarketorders.AccordingtoKanielandLiu(2006),informed
traderswilltendtochooselimitordersiftheyhaveasufficientlylonginformationhorizon.Sinceit
isdifficultforrivaltraderstocopysuchlong-livedinformation,informedtraderscanexecutetheir
tradingonsuchinformationusinglimitorders,whichmeansthattheyarepreparedtosacrifice
imme-diacyforalowerpriceimpact.Ourfindingsimplythatthesuperiorinformationpossessedbyforeign
institutionalinvestorsintheTaiwanoptionmarketsisalsolong-lived,perhapsderivedfromtheir
interpretationofpubliclyavailableinformation.Indeed,ourfurtheranalysisshowsthattheput-call
ratiosofforeigninstitutionalinvestorshavebetterpredictivepowerontheindexchangesondays
withimportantmacroeconomicnewsannouncements,ascomparedtodayswithnosignificantnews
announcements.
Finally,thepredictiveabilityofforeigninstitutionalinvestorsduring‘normal’tradingdaysisfound
tohavetotallydisappearedduringthefinancialtsunamiperiod,evenonthosedayswhenimportant
macroeconomicnewswasannounced.Furtheranalysisshowsthat,althoughswitchingtomore
defen-sivepositions,foreigninstitutionsfailedtoanticipatethelargemarketdownturnthuswereunableto
synthesizeprofitableoptionpositionspriortoadversemarketimpacts.Resultsalsoindicatethattheir
diminishedpredictabilitywerelessattributabletocapitalconstraintsorhedgingneeds.
Theremainderofthispaperisorganizedasfollows.Anintroductiontothetradingmechanism
forTaiwanindexoptionsisprovidedinSection2,followedinSection3byadescriptionofthedata
adoptedforthisstudy.Thepredictiveregressionmodelusedtoexaminetheinformationcontentof
theoptionsvolumeratioispresentedinSection4,followedinSection5bythepresentationand
discussionofourempiricalresults.Finally,theconclusionsdrawnfromthisstudyarepresentedin
Section6.
2. Institutionalbackground
WeempiricallyexaminetheTaiwanindexoptions(TXO)tradedontheTaiwanFuturesExchange
(TAIFEX)inordertoanalyzethetradingdecisionsofinformedtraders.TXOcontractsareactivelytraded
onaglobalscale,andindeed,accordingtotheannualreportoftheWorldFederationofExchanges,of
allindexoptionslistedworldwide,tradingfrequencyinTXOcontractswasrankedasthefourth(fifth)
largestintheworldin2007(2008).TheunderlyingspotindexoftheTXOistheTaiwanCapitalization
WeightedStockIndex(TWI),avalue-weightedindexcomprisingofvirtuallyallofthecommonstocks
listedontheTaiwanStockExchange(approximately700firms).
Atleastfiveexpirationmonthsarelistedonanytradingday–includingthespotmonth,thenext
twocalendarmonthsandthenexttwoquartermonths(March,June,SeptemberandDecember)–with
awiderangeofstrikepricesbeingavailableforeachcontractmonth.6Duringoursampleperiod,the
6Duringoursampleperiod,theTaiwanStockindexlevelrangedbetween7000and9000.Withinthisrangethestrikeprice
intervalis100indexpointsforthespotmonthandthenexttwocalendarmonths,whereasitis200indexpointsforthe additionaltwoquartermonths.
notionalvalueofeachoptionwasapproximatelyNT$400,000(US$16,000),withthesmallercontract
sizeattractingalargenumberofretailtraders,whocontributeover50percentofalltradingvolume.
TheTAIFEXoperatesafullyelectronictradingsystemwheretraderscansubmiteithermarket
orlimitorders,withthelimitordersalsobeingeithermarketableornon-marketable.Theregular
tradinghoursarefrom8:45a.m.to1:45p.m.Allorderssubmittedbefore8:45a.m.areexecutedbya
callmethoduponmarketopening,withordersbeingcontinuouslymatchedduringtheregulartrading
session.Incomingordersareautomaticallymatchedagainstexistinglimitordersontheoppositeside
ofthebook,followingstrictpriceandtimepriorityrules.
TheExchangecoordinatesdesignatedmarketmakersforTXOtransactionsinordertoensure
mar-ketliquidity,withallmarketmakersbeingobligatedtooffertwo-way(bidandask)quotesupon
receiptofaquoteinquiryfromothermarketparticipants.Thequoteenteredintothesystembya
marketmakerisafirmorder,whichwillentertheorderbooktocompetewithotherpublicorders
baseduponthesamepriceandtimepriorityrules.Afuturesproprietaryfirmcanapplytobeamarket
maker,whichenjoysdiscountsontransactionfeesifitsmarket-makingvolumeexceedsacertainlevel.
MarketmakersfailingtomeettheTAIFEXrequirementforliquidityprovisionwillbedisqualified.
TheTAIFEXdisclosesquotepricesanddepthsforthebestfivequotes,aswellastheresultsof
eachtradeexecuted(includingthetradepriceandvolume),withtheinformationbeingelectronically
disseminatedtothepublicinreal-time.Tradingisanonymousinthat,bothbeforeandafteratrade,
informationontheidentityofthosesubmittingordersandtheirtradecounterpartiesisunavailableto
thepublic.Afterthemarketclose,theTAIFEXpublishesitsdailytradingsummarystatistics,including
tradingvolumeinindividualcontracts,thevolumeofallfuturesproprietaryfirms,theput/callvolume
bytraderclassandthebuy/sellvolumebytraderclass.
3. Data
Tick-by-tickdataonTXOoptionswasobtainedfromtheTAIFEXfor372tradingdaysbetween2
January2007and30June2008.Thedatasetincludesthreefilesonordersubmissions,tradeexecutions
andmarketmakerquotations.Theordersubmissionsfilerecordsthedate,time,traderID,abuy/sell
indicator,orderprice,ordersizeandthecontractcharacteristics(strikeprice,maturityandacall/put
indicator)foreverysubmittedorder.Thetradeexecutionsfilecontainsthetradeprice,volumeanda
keylinkingthetradebacktotheoriginalorderforeverymatchedtrade.
Bymatchingtheordersubmissionandtradeexecutionfiles,weareabletoconstructthecomplete
historyofeverytransactionandidentifythetradersoneachsideofthetransactions.Thetraders
areclassifiedasdomesticindividualinvestors,domesticinstitutionalinvestors,foreigninstitutional
investorsormarketmakers,andareassignedauniqueID.Thedatasetalsoindicateswhetherthe
transactioninvolvestheopeningofanewpositionortheclosure(offsetting)ofanexistingposition.
Thedetailedclassificationofthedataallowsustoaggregatetheoptionsvolumeattributabletothe
differentclassesoftraders(individualinvestors,domesticinstitutionalinvestors,foreigninstitutional
investorsandmarketmakers),thecharacteristicsofthecontracts(put/call,strikepriceandmaturity),
thetypeoforders(marketversuslimitorders)andtradingpositions(buyversussell,openversus
offsetting).
Furthermore,weareabletoidentifyeachofthepartiesinamatchedtradewithouthavingtorelyon
theLeeandReady(1991)algorithm,whichhelpsustoclearlydistinguishbetweenthemotivesbehind
eachtransaction.Thisisextremelyusefulforourexaminationofthewaysinwhichtheinformation
isimpoundedintosecuritypricesthroughselectionoftiming,contract,order,tradingpositionand
tradingstrategybythevariousclassesoftraders.7
Sincethepresentstudyfocusesonthepredictiveabilityofdirectionaltrades,thosetransactions
involvingmultiplepositions(suchasspread,straddleandstrangletrades,allofwhicharelargely
non-directionalinnature)areexcludedfromoursample.TheTAIFEXdatasetdistinguishesbetween
7Thedataiscollected,processedanddisseminatedelectronicallybytheTAIFEX.Thesamedatasetwasalsoemployedby
Changetal.(2009),Hanetal.(2009)andChouandWang(2009)toexaminevariousissuesinoptions/futurestradingonthe Taiwanindexderivativesmarkets.
Table1
Optionsvolume,bydifferentclassesoftradersandtradingpositions.
Variables Foreigninstitutions Marketmakers Domesticinstitutions Individualinvestors
PanelA:overalldistributionbytraderclasses
8.20 34.19 4.44 53.16
PanelB:distributionbycall/putandbuy/sell
Buycall 30.75 26.33 24.65 29.56
Sellcall 19.81 27.41 25.56 28.33
Buyput 30.44 23.01 25.66 20.66
Sellput 19.00 23.24 24.13 21.45
PanelC:distributionbyopen/closeandbuy/sell
Openbuy 55.62 21.82 24.69 33.48
Opensell 26.06 23.20 26.80 17.97
Closebuy 5.57 27.53 25.62 16.74
Closesell 12.75 27.45 22.89 31.81
Thistablereportsthedetailsofvolumedistributionwithinthefourclassesoftradersbetween2January2007and30June2008. ThedistributioninPanelAiscalculatedbydividingthetotaltradingvolumeforeachtraderclass(innumberofcontracts)by thevolumefortheoverallmarket,whilstthedistributionsinPanelsBandCarecalculatedbydividingthevolumeforeach categorybythevolumeofthecorrespondingclassoftraders.
plain-vanillaoption tradesandfourtypesofcombinationtrades,comprisingofstraddle,strangle, moneyspreadandcalendarspreadtrades.Thesecombinationtrades,whichaccountforlessthan2 percentofthetotaltradingrecords,aredulyexcludedfromouranalysis.
Asummaryoftradingvolume,bytraderclasses,buy/selltransactions,put/calloptionsandnewly openedpositionsversustheclosureofexistingpositions,isprovidedinTable1,withPanelAreporting
thedailytradingvolume(inpercentageterms)acrossthefourclassesoftradersandthetime-series
averagesover367tradingdays.Aswecanseefromthistable,tradingvolumedifferssignificantly
acrossthefourclassesoftraders,withindividualinvestorsbeingthemajorparticipantsintheTXO
market,accountingfor53.16percentofthetotalvolume.
Marketmakersarethesecondlargesttraderclass,accountingfor34.19percent,whilstforeign
institutionalinvestorsanddomesticinstitutionalinvestorsaretheleastactivetraders,respectively
accountingforonly8.20percentand4.44percent.Thepresenceofsuchlargenumbersofuninformed
retailtradersprovidesopportunitiesforothermoresophisticatedtraderstoexploittheirinformational
advantage.
Alltransactionswithineachclassoftradersarefurtherbrokendown,inPanelBofTable1,into
buy-call,sell-call,buy-putandsell-puttransactions,withtheproportionssummingverticallytounity
withineachtraderclass.Wefindthatforeigninstitutionalinvestorsholdsubstantiallylargerpositions
inlongcalls(30percent)thanshortcalls(20percent),whereasthelong-versusshort-callvolume
isfoundtobeevenlydistributedforotherclassesoftraders.Asimilarimbalanceisalsodiscerniblein
longputversusshortputvolumeforforeigninstitutionalinvestors,butnotfortheotherthreeclasses
oftraders.
Sincelongcalls(longputs)enjoygreaterpotentialprofitsthanshortputs(shortcalls)whenthe
underlyingindexmovesupward(downward),longpositionsaregenerallydeemedtobethemore
aggressivepositions (Panand Poteshman,2006).8 Thesubstantially largerlongoption volumein
thetradingaccountsofforeigninstitutionalinvestorsindicatesthatthesetraderstendtobemore
aggressivethanothertraders,intermsofdirectionaltrading.9
8PanandPoteshman(2006)notedthatthepredictiveabilityofoptionopen-buyvolumewasfoundtobebetterthanthatof
open-sellvolume,andindeed,theysuggestedthatinformationtradingwaslikelytobeimplementedusinglongcallsorputs, ratherthanshortputsorcalls.Thisisessentiallybecausetheworst-casescenarioinbuyinganoptionisthelossoftheoption premium,whereastheupsidegaincanbequitesubstantialiftheprivateinformationturnsouttobecorrect.Conversely,the best-casescenarioofsellinganoptionisretainingtheoptionpremium,whereasthedownsidelosscanbequitesubstantialif theprivatesignalturnsouttobeincorrect.
9Wewillexamineinlatersectionswhethertheaggressivetradingofforeigninstitutionalinvestorsisbasedupontheir
Detailsofthevolumedistributionforeachofthefourclassesoftradersinopen-buy,open-sell,
close-buyandclose-selltransactionsareprovidedinPanelCofTable1.Ofthefourtypesof
transac-tions,themostlikelytobepursuedbytradersinpossessionofsuperiorinformationonfutureprice
movementswouldpresumablybeopen-buytransactions,sinceopentradesareoftenusedtoestablish
newpositionsforspeculatingpurposesinresponsetonewinformationwithinthemarket.Conversely,
closetrades(includingbothclose-buyandclose-sell)areregardedasbeinglessaggressive,essentially
becausetradersmayexecutesuchtransactionsasameansofrealizingtheirgainsoracceptingtheir
losses.
Opentradesthereforecontainmoreinformationthanclosetrades,andwhenengaginginopen
trades,buyingoptionsprovideshigherpotentialprofitsthansellingoptions;therefore,traderswho
aggressivelytradeontheirinformationaladvantagewouldtendtoinitiatemorelongpositionsinorder
tomaximizetheirspeculativegains.Takingallofthesefactorstogether,sinceopen-buytransactions
maywellbeusedforinformedtrading,theyarelikelytohavemuchricherinformationcontentthan
othertypesoftransactions(PanandPoteshman,2006).
AsshowninPanelCofTable1,foreigninstitutionalinvestorsdevote55.62percentoftheirtrading
volumetoopen-buy transactions,asignificantlygreaterproportionthanthatofanyoftheother
threeclassesoftraders.Thissubstantiallylargerproportionofopen-buytradingindicatesthatforeign
institutionalinvestorsarelikelytopossesssuperiorinformationandaggressivelyuseopen-buyoption
transactionsinordertorealizetheirinformationaladvantage.10
4. Methodology
Theinformationcontentofoptionsisassessedthroughoutthisstudybythepredictiveabilityof
optionvolumeonfutureindexreturns,withourempiricalmodelfollowingthespecificationsofPan
andPoteshman(2006)toregressthenext-dayspotmarketreturnagainsttheoptionput-callvolume
ratio.Theoptionput-callvolumeratio,whichrepresentstheinformationpossessedbyoptiontraders
onfuturechangesinthedirectionoftheindex,isdefinedinthisstudyastheopen-buyputtrading
volume(innumberofcontracts)dividedbythesumofopen-buyputandopen-buycallvolume.
Changetal.(2009)subsequentlywentontosuggestthattheput-callratiocouldbeusedtoexamine
whetheraparticularclassoftraderspossessedsuperiorinformationoverothermarketparticipants,
withthecalculationbeingbasedupontransactionsfromasubsetoftraderclasses.Ourbasicregression
modelfollowsthissuggestiontoregressthenext-daystockindexreturnsontheput-callvolumeratios
calculatedforeachofthefourclassesoftradersshowninTable1.
Thebasicmodelusedthroughoutthisstudyisasfollows:
Rt+1=˛i+ˇi,upXi,t×Dt+ˇi,downXi,t×(1−Dt)+εi,t (1)
Dt=
1, Rt>0;
0, otherwise
whereRt+1isthedailyclose-to-closereturnontheTWIspotindexatdatet+1,whichweconvert
intobasispointsbeforeperformingtheregressions;Xi,tistheinformationvariableproxiedbythe
open-buyput-callvolumeratios(calculatedfromthetransactionsoftraderclassiatdatet);andDt
isadummyvariablewhichisequalto1ifthecontemporaneous(dayt)marketreturnispositive;
otherwise0.
10WhenexaminingPanelCofTable1,wemayexpecttheopen-buy(open-sell)volumebeingequaltotheclose-sell
(close-buy)volume;however,anumberofpracticalissuesinterferewiththeequalityrelationship.Firstly,theclose-sellvolumeof marketmarkersismorethantheiropen-buyvolume,essentiallybecauseduringoursampleperiodsomemarketmakerswere disqualifiedwhilstotherswerenewly-licensed,whichalterstheidentityofthetradersandmakesitdifficulttopreciselymatch theopen-buyandclose-sellvolumewithinanyparticulartradertype.Secondly,theclose-sellvolumecouldbesmallerthanthe open-buyvolumewhensomeopen-buycontractsareheldtoexpirationandsettledforcash;insuchcases,therewillnotbea correspondingclose-selltransactionforanopen-buyposition.Thesameexplanationappliestothesmallerclose-buyvolume thanopen-sellvolume.
TheinformationvariableXi,tisdefinedas: Xi,t=
Pi,t Pi,t+Ci,t
(2)
wherePi,tandCi,t aretherespectivenumbersofcontractsfortheopen-buyputandcalltradesof
investorclassiatdatet.Iftheopen-buyvolumeofaparticularclassoftraderspredictsthesubsequent
movementsintheindex,thenwewouldexpecttoobservesignificantlynegativeˇcoefficients.
FiglewskiandWebb(1993)andDanielsenandSorescu(2001)suggestedthattheshort-sales
con-straintsinthestockmarketresultininvestorstradinginderivativesinresponsetobadnews.Ifthis
isthecase,thenoptionswouldbemoreinformativeonadaywhentherewasadeclineinthemarket
(whenspottradingismoresubjecttoshort-salesconstraints),ascomparedtoadaywhenthereisan
increaseinthemarket(whenfewerstocksarehinderedbyshort-salesconstraints).Weincludean
indicativevariable(Dt)toseparatetheupanddownmarketdays,andallowthisvariabletointeract
withtheput-callratios.Anegativelysignificantˇi,up(ˇi,down)wouldindicateatendencyforinformed
tradingtobetakingplaceinanupward(downward)market.11
Thebasicmodelcalculatesthevolumeratiousingonlyopen-buyvolume,becauseweexpectto
findthattheopen-buyvolumewillbemoreinformativethantheotherthreetypesofvolume.The
basicmodelisexpandedinEq.(3)byincludingadditionalthreevolumeratiosforopen-sell,close-buy
andclose-sellvolume,withthevolumeratiosbeingdefinedinaccordancewithEq.(2).Allofthese
volumeratiosaresettointeractwiththeupward/downwarddummyvariable.
Thisalternativeregressionmodelisspecifiedas:
Rt+1 =˛i+[ˇOBi,upX OB i,t +ˇ OS i,upX OS i,t +ˇ CB i,upX CB i,t+ˇ CS i,upX CS i,t]×Dt+[ˇ OB i,downX OB i,t +ˇ OS i,downX OS i,t
+ˇCBi,downXi,tCB+ˇCSi,downXCSi,t]×(1−Dt)+εi,t (3)
Dt=
1, Rt>0;
0, otherwise
whereXOB
i,t,Xi,tOS,Xi,tCBandXi,tCSrespectivelyrefertotheopen-buy,open-sell,close-buyandclose-sell put-callratios.
Speculatingonanupwardmovementintheindex,informedtraderswouldtendtopurchasecalls
(open-buycalls),sellputs(open-sellputs),closeexistingshortcallpositions(close-buycalls)orclose existinglongputpositions(close-sellputs).TheresultoftheirtradingwouldbetoraiseXOSandXCS,
whilstdepressingXOBandXCB.Similarly,traderswhohaveexpectationsofadeclineintheindexwould
tendtosellcalls(open-sellcalls),buyputs(open-buyputs),closeexistinglongcallpositions(close-sell
calls)orcloseexistingshortputpositions(close-buyputs).Theendresultoftheirtradingwouldbe
tolowerXOSandXCS,whilstraisingXOBandXCB.Ifcertainclassesoftradersweretoconsistentlymake
correctpredictions,wewouldthenobserveanegativecoefficientonbothˇOBandˇCB,andapositive
coefficientonbothˇOSandˇCS,astheinformationvariablesreflectingtheiropen-buy,close-buy,
open-sellandclose-selltransactions.
11Itshouldbenotedthatourdefinitionofup/downmarkets,whichisdependentontheindexreturnatdayt,differsfrom
theconventionalidentificationofbull/bearmarkets,basedonreturnsoveralongerperiodoftime.Ourmodelisdesignedto examinetheeffectofshort-salesconstraintsoninformedtradingintheoptionsmarket;toachievethisaim,theoveralleffect ofshort-salesconstraintscanbemorepreciselyidentifiedbydefiningup/downmarketsbasedontheindexreturnatdayt,as opposedtoreturnsoveralongerperiod.ThesamedefinitionwasadoptedbyChenandRhee(2010)toidentifytheroleofshort salesontheinformationefficiencyofstocksinupversusdownmarketsandHameedetal.(2010),whomodeledtheeffectof marketreturnsonthebid-askspread,conditionalonthedirectionoftheindexmovementondayt.
5. Empiricalresults
5.1. Identifyinginformedtraders
ThepredictiveregressionresultsofEqs.(1)and(3)foreachclassoftradersarepresentedinTable2,
withthefirstandsecondcolumnrespectivelyreportingthecoefficientvariablesandtheirexpected
signs(−or+)iftheput-callvolumeratiocorrectlypredictsthenextdayindexreturn.Thenumbers
showninboldtextindicatethatthecoefficientsarestatisticallysignificantandthattheirsignssupport
thepredictiveabilityoftheput-callratio.
Severalimportant findings arehighlighted, as follows.Firstly,of the fourdifferent classesof
investors,theoptiontradingpositionsofforeigninstitutionalinvestorsappeartoprovidethemost
accurateforecastingofthenext-dayspotindex,astheiropen-buyvolumeratioisfoundtobe
neg-ativelyassociatedwiththenext-dayindexreturninEq.(3),withstatisticalsignificanceatthe1per
centlevel.Thisindicatesatendencyforariseinthenext-dayindexasforeigninstitutionalinvestors
increasetheiropen-buycallsrelativetoopen-buyputs,andviceversa.Theclose-sellratioof
for-eigninstitutionalinvestorsalsocorrectlypredictsthenext-dayindexreturn,albeitwithmarginal
significance.
AsregardsEq.(1),whereonlytheopen-buyvolumeratioisincluded,ofthefourclassesoftraders,
onlytheopen-buypositionsofforeigninstitutionalinvestorsarefoundtohaveanysignificant
cor-relationwithnext-dayindexreturns.Furthermore,thepredictiveregressionforforeigninstitutional
investorsisfoundtoyieldthehighestR2levelsamongstalloftheregressionsunderthesamemodel,
whichtherebysuggeststhatthevolumeratiosofforeigninstitutionalinvestorsarecapableof
explain-ingagreaterproportionofthevariationinthenext-dayreturnsthanthevolumeratiosofallother
typesoftraders.12
Secondly,thepredictiveabilityofthevolumeratiosofforeigninstitutionalinvestorsisfoundto
prevailonlyinadownwardmarketbutcompletelyabsentinanupwardmarket.PanelAofTable2
showsthatthecoefficientontheopen-buyvolumeinadownwardmarket(ˇOB
down)isnegativeand
statisticallysignificant,whereasthecoefficientinanupwardmarket(ˇOB
up)isinsignificant.Such
asym-metryinpredictiveabilityisconsistentwithanalysesreportedinthepriorstudiesandmaywellreflect
theeffectofshort-salesconstraintsintheTaiwanspotmarket.13
Forexample,fromtheirexaminationofequityoptions,DanielsenandSorescu(2001)demonstrated
thattheshort-salesconstraintsimposedonthespotmarketmaywellresultintraderswithsuperior
informationbeingforcedtotradeontheirinformationintheequityoptionsmarkets.Ourevidence
furtherindicatesthattherestrictionsonshort-sellingstockscouldalsoresultinstrongerdemandfor
indexoptionsinadownwardmarket,essentiallyforthepurposeofhedgingagainstthedownsiderisk
andspeculatingonanupcomingdecline.Asinformedtradersrushtoopenhedgingorspeculative
positionsinindexoptions,theiroptionvolumeratiosbecomemoreinformative.
Wecarryoutthetestsforthefollowingtwoimplicationsinordertofurtherinvestigatethe
short-saleshypothesis.Ifthevalueoftheinformationrelatingtotheoptionisenhancedwithstricterspot
short-salesrestrictions,then:(i)thepredictiveabilityoftheoptionvolumeratioshouldbemore
pronouncedwhenthereturnatt+1isnegativethanwhenitispositive;and(ii)thepredictiveability
shouldhavebeenweakenedafter12November2007,whentheTaiwanStockExchangeliftedthe
‘up-tick’ruleforaround150liquidstocks.Theresultsofthesetwotestsontheopen-buyvolumeratio
offoreigninstitutionalinvestorsarepresentedinTable3.14
12ThepredictiveregressionsusuallyyieldlowR2values.TheR2valuesinthepresentstudyarecomparabletothoseofChang
etal.(2009),whoreportedR2rangingfrom0.0016to0.0087forsimilarpredictiveregressionsofindexreturnsonput-call
ratios.
13Theshort-salesconstraintswithintheTaiwanstockmarketcompriseof:(i)thelimitedavailabilityofsharestoborrow;(ii)a
marginrequirementofabove110%oftheshareprice;(iii)therequirementforshortsellerstocovertheirshortpositionsduring awindowsurroundingex-dividenddays;and(iv)the‘up-tick’rulewherebyastockistemporarilybannedfromshort-sales transactionswhenthecurrentstockpricehasfallenbelowtheclosingpriceontheprevioustradingday.Althoughthe‘up-tick’ rulewasliftedforaround150ofthemostliquidstocksafter12November2007,itremainedinplaceforallotherstocks.
Theoverallsampleisdividedintotwosub-samplesinPanelA,withthefirstofthesesub-samples
comprisingofcases(days)forwhichRt+1>0,andthesecondcomprisingofcases(days)forwhich
Rt+1≤0.Theresultsofourregressionsarefoundtobeconsistentwiththeshort-saleshypothesis,
thatthepredictiveabilityofoptionswillbehigherinthosecaseswheretheindexreturnatt+1is
negative,ascomparedtowhenitispositive.Theˇdowncoefficientisfoundtobesignificantlynegative
fortheRt+1≤0sub-sample(withpredictiveability)whereasnostatisticalsignificanceisfoundinthis
coefficientfortheRt+1>0subsample.15
In PanelBofTable3,thesampleisdividedintotwosub-periods,pre-andpost-12November
2007,whentheTaiwanStockExchangerelaxedtheup-tickruleforaround150liquidstocks.We
allowtheindependentvariabletointeractwithanindicativevariable,S,whichisequalto1ifthe
observationoccurspriorto12November2007,otherwise0.Theresultslendfurthersupporttothe
short-saleshypothesis,withtheonlysignificantcoefficientwithacorrectpredictivesignbeingˇBeforedown,
thecoefficientforthevolumeratiopriortotherelaxationoftheup-tickruleforobservationsona
downmarketday.Aftereasingtheshort-salesconstraints,thepredictiveabilityofoptionsonadown
marketday,asshownbyˇAfterdown,isfoundtobestatisticallyinsignificant,albeitwiththecorrectsign.
Ourfindingsareconsistentwithanumberofthepriorstudieswheretheasymmetricinformation
efficiencyin upversusdownmarketsisattributedtoshort-salesconstraintsor shortselling. For
example,Chanet al.(2009)identifiedanenhancedleadingrole ofoptionsovertheequityindex
duringdownwardtrendperiods,whilstLeeandWang(2012)documentedastronginformationalrole
intheshort-sellingactivitiesofforeigninstitutionalinvestorswithintheKoreanstockmarket.Saffi
andSigurdsson(2011)alsoreportedlowerpriceefficiencyforthosestocksthatweresubjecttohigher
short-salesconstraints.
Our thirdimportantfindingisthatthepositionsofmarketmakerscanhardlybeclassifiedas
outcomesofinformedtrading.InPanelBofTable2,mostofthecoefficientsarefoundtohavethe
oppositesigntothatspecifiedinthefirstcolumn.Evenworse,marketmakers’open-buyand
open-sellratiosexhibitasignificantshiftinthewrongdirectioninthecaseofdownwardindexmovements.
Thisfindingmaylookquiteoddatfirstglance,giventhatmarket-makingfirmsinTaiwanareoperated
byprofessionalswithexpertiseandexperienceinoptionstrading;however,itmaybereconciled
byrecognizingtheprimaryroleofmarketmakersasliquidityproviders.Whilstprovidingliquidity
topotentialinformedtraders,thesemarketmakersareessentiallyleaningagainstthewind.Asa
consequence,theoppositeandsignificantregressioncoefficientsmerelyreflectthefactthattheyare
indeedfulfillingtheirobligationbytakinguppositionsontheoppositesideofinformedtrading.16
Fourthly,neitherdomesticinstitutionalinvestorsnorretailinvestorsarewellinformedwhen
trad-inginindexoptions;indeed,theopen-buyvolumeofdomesticinstitutionalinvestorsinadownward
markethasanegativecoefficientandap-valueof0.0935forregressionModel(3).Asidefromthis
marginalpredictiveability,tradingbydomesticinstitutionalinvestorsinanupwardmarketappears
toresultinshort-termlosses,sincetheirˇOS
up,ˇCBupandˇCSupcoefficientsexhibitoppositesignstoour
expectations.Itshouldbenotedthatthedefinitionof‘domesticinstitutionalinvestors’providedby
theTAIFEXincludesmutualfunds,banksandcorporations,whilstexcludingfuturesproprietaryfirms
(primarilycategorizedasmarketmakers).Sincetradinginoptionsbythesemarketparticipantsis
largelyforhedgingpurposes,theirpositionsshouldberelativelyuninformative.
Asregardsindividualtraders,allofthecoefficientsinthetworegressionsarefoundtobe
insignif-icant,despitesomesignsbeingconsistentwiththenext-dayindexreturns.Thisfindingissimilarto
thosereportedinseveralofthepriorstudiesinwhichitisnotedthatgiventheirdisadvantagesin
15Wealsoexamineanalternativemodel,wherethedummyvariableD
t(thepositive/negativeindicatoroftheindexreturn
ondayt)inEq.(1)isreplacedbyDt+1(thepositive/negativeindicatoroftheindexreturnondayt+1).Thecoefficientofˇdown
isfoundtobesignificantlynegative(−210.61,p-value<0.0001),whereasthecoefficientofˇupisfoundtobepositive(126.43,
p-value<0.0001).Theseresultsprovidesupportfortheshort-saleshypothesis,thatthepredictiveabilityofoptionswilltend tobehigherwhentheindexreturnondayt+1isnegativeascomparedtowhenitispositive.Thismodelmay,however,suffer fromtheproblemofendogeneity,sinceRt+1appearsonbothsidesoftheequation;therefore,thisresultisnottabulated.
16Theresulthereshouldnotbetakenasevidencethatmarketmakersincurshort-termlosses.Inordertoassessthenet
tradingprofits/lossesofanymarketmakers,weneedtotakeintoaccountthecompensationthatmarketmakersreceivefrom thebid-askspread.
Table2
Predictiveabilityofoptionput-callvolumeratios.
Variables Expectedsignifwithpredictiveability Eq.(3) Eq.(1)
Coefficient p-Value Coefficient p-Value
PanelA:foreigninstitutions
Intercept 31.22 0.3682 28.96 0.1766 ˇOB up − −22.50 0.7230 −48.48 0.2517 ˇOS up + −112.53 0.1213 – – ˇCB up − −14.23 0.7110 – – ˇCS up + 43.98 0.2985 – – ˇOB down − −176.63 *** 0.0097 −64.23* 0.0833 ˇOS down + 1.48 0.9844 – – ˇCB down − 39.28 0.3922 – – ˇCS down + 76.32 * 0.0836 – – R2 0.0401 0.0085
PanelB:marketmakers
Intercept −13.14 0.8030 −45.77 0.1987 ˇOB up − 113.32 0.3518 106.50 0.1865 ˇOS up + −19.25 0.8789 – – ˇCB up − −50.29 0.7250 – – ˇCS up + −3.09 0.9825 – – ˇOB down − 249.48 * 0.0749 83.70 0.2455 ˇOS down + −283.92 * 0.0770 – – ˇCB down − 177.06 0.3341 – – ˇCS down + −150.54 0.3355 – – R2 0.0165 0.0048
PanelC:domesticinstitutions
Intercept 35.74 0.4334 18.16 0.5558 ˇOB up − 40.81 0.5395 −29.14 0.6042 ˇOS up + −169.07* 0.0790 – – ˇCB up − 182.79** 0.0430 – – ˇCS up + −124.44* 0.0638 – – ˇOB down − –123.50 * 0.0935 –39.09 0.4411 ˇOS down + 61.67 0.5272 – – ˇCB down − 30.37 0.7389 – – ˇCS down + −2.33 0.9739 – – R2 0.0284 0.0020
PanelD:individualinvestors
Intercept 35.35 0.6645 −15.62 0.7083 ˇOB up − 156.90 0.3281 43.83 0.6740 ˇOS up + −3.68 0.9787 – – ˇCB up − −157.54 0.3059 – – ˇCS up + −48.00 0.7738 – – ˇOB down − −93.51 0.6308 28.26 0.8046 ˇOS down + 33.17 0.8307 – – ˇCB down − −78.98 0.6693 – – ˇCS down + 35.37 0.8067 – – R2 0.0064 0.0010
Thistablereportstheresultsofthefollowingpredictiveregressionsforeachofthefourclassesofinvestors.
Rt+1=˛i+ˇi,upXi,t×Dt+ˇi,downXi,t×(1−Dt)+εi,t (1)
Rt+1 =˛i+
ˇOB i,upX OB i,t+ˇ OS i,upX OS i,t+ˇ CB i,upX CB i,t+ˇ CS i,upX CS i,t ×Dt+ ˇOB i,downX OB i,t+ˇ OS i,downX OS i,t+ˇ CB i,downX CB i,t+ˇ CS i,downX CS i,t ×(1−Dt)+εi,t (3)whereRt+1isthedailyclose-to-closespotindexreturnondayt+1;XS(Xi,t=Pi,t/(Pi,t+Ci,t))aretheput-callvolumeratioscalculated
usingopen-buy(OB)andopen-sell(OS)volume,andclose-buy(CB)andclose-sell(CS)volume;andDisanindicativevariable whichisequalto1ifRt>0;otherwise0.Thesignsofthecoefficientssupportingthepredictiveabilityofthevolumeratioare
indicatedbyaplus(+)oraminus(−)inthefirstcolumn.Boldnumbersindicatecoefficientswithcorrectpredictivesignsand statisticalsignificance.
*Indicatesstatisticalsignificanceatthe10%level. **Indicatesstatisticalsignificanceatthe5%level. ***Indicatesstatisticalsignificanceatthe1%level.
Table3
Influenceofshort-salesrestrictionsonthepredictiveabilityofforeigninstitutionalinvestors.
Variables Coefficient p-Value
PanelA:predictiveabilityconditionalonreturndirectionondayt+1
a.SubsampleRt+1>0(169observations) Intercept 123.06*** <0.0001 ˇup 36.71 0.4778 ˇdown −10.17 0.8164 Intercept −117.89*** <0.0001 R2 0.0131 b.SubsampleRt +1≤0(198observations) ˇup −46.63 0.1458 ˇdown −73.09* 0.0811 Intercept 34.59 0.1137 R2 0.0112
PanelB:predictiveabilityconditionalonsub-periodsbeforeandafterstrictershort-salesrestrictions
ˇBefore up −78.78 0.1527 ˇBefore down −102.01 ** 0.0266 ˇAfterup −48.57 0.2607 ˇAfterdown −54.54 0.1601 R2 0.0143
PanelAreportstheresultsonthepredictiveabilityofforeigninstitutionalinvestors,ontwosub-samplescomprisingofcases whereRt+1>0andwhereRt+1≤0,baseduponthefollowingpredictiveregression:
Rt+1=˛+ˇupXt×Dt+ˇdownXt×(1−Dt)+εt,
whereRt+1isthedailyclose-to-closespotindexreturnondayt+1;Xt=(Pt/(Pt+Ct))istheput-callvolumeratiocalculatedusing
open-buy(OB)volumeofforeigninstitutionalinvestors;andDtisanindicativevariablewhichisequalto1ifRt>0;otherwise
0.Coefficientswithanegativesignareconsistentwiththepredictiveabilityofforeigninstitutionalinvestors.Boldnumbers indicatecoefficientswithcorrectpredictivesignsandstatisticalsignificance.
PanelBreportstheresultsconditionalonthetwosub-periodssubjecttodifferentshort-salesrestrictions,baseduponthe followingpredictiveregression:
Rt+1=˛+
ˇBeforeup Xt×Dt+ˇBeforedownXt×(1−Dt)×St+
ˇAfterup Xt×Dt+ˇAfterdownXt×(1−Dt)
×(1−St)+εt,
whereDtisanindicativevariablewhichisequalto1ifRt>0,otherwise0;Stisanindicativevariablewhichisequalto1if
theobservationoccurredpriorto12November2007(thedateonwhichtheTaiwanStockExchangeeasedtheup-tickrulefor around150liquidstocks),otherwise0.Allothervariablesareasdefinedpreviously.
*Indicatesstatisticalsignificanceatthe10%level. **Indicatesstatisticalsignificanceatthe5%level. ***Indicatesstatisticalsignificanceatthe1%level.
trading against professionals (in terms of capital, expertise and research resources), individual
investorscontributelittletopricediscoveryorthedisclosureofinformationonfutures(Frinoetal.,
2004;ChouandWang,2009),options(Ahnetal.,2008;Hanetal.,2009)orspotequities(Grinblatt andKeloharju,2000;Barberetal.,2009).
TheEqs.(1)and(3)regressionresultsindicatethattheopen-buyratioismoreinformativethan
theotherthreevolumeratios,whichisconsistentwiththesuggestionofPanandPoteshman(2006)
thatanopen-buystrategyrepresentsamoreaggressivestrategywhentradershavestrongviewson
futurepricemovements,therebyrevealingricherinformationcontentthanopen-sell,close-buyand
close-sellstrategies.Oursubsequentanalysisthereforefocusesentirelyonthepredictiveabilityofthe
open-buyvolumeratio.
OurregressionresultsusingEq.(1)showadeclineof64.23basispointsinthenext-dayindex
followingaoneunit increase(from0to1)intheopen-buyput-callratioofforeigninstitutional
investors.Translatingthisintoamoreplausiblevariationintheput-callratio,aonestandard
devia-tion(21percent)changeintheput-callratioisassociatedwitha−13.48basispointchangeinthe
Table4
Predictiveabilityoftheopen-buyratioconditionalonoptionmarketactivity.
Variables Foreigninstitutions Marketmakers Domesticinstitutions Individualinvestors Coefficient p-Value Coefficient p-Value Coefficient p-Value Coefficient p-Value
PanelA:lowaggregateoptionvolume(below33%)
Intercept 47.01 0.1944 −118.37* 0.0576 65.85 0.2433 4.02 0.9634 ˇup −80.67 0.2583 282.34** 0.0482 −114.83 0.2584 −3.04 0.9894
ˇdown −69.20 0.3085 275.32** 0.0448 −92.74 0.3155 13.23 0.9573
R2 0.0114 0.0351 0.0109 0.0004
%ofContracts 23.91 22.18 27.50 26.41
PanelB:mediumaggregateoptionvolume(between33%and67%)
Intercept −25.36 0.4875 46.70 0.4096 43.86 0.3274 −12.67 0.8372
ˇup 42.34 0.5408 −131.06 0.3035 −100.90 0.2226 4.71 0.9745
ˇdown 22.77 0.7120 −120.34 0.3217 −93.50 0.2224 5.21 0.9750
R2 0.0034 0.0089 0.0131 0.0000
%ofContracts 34.00 32.49 33.41 32.92
PanelC:highaggregateoptionvolume(above67%)
Intercept 52.98 0.1836 −77.17 0.2875 −61.47 0.3064 −51.69 0.4912 ˇup −88.19 0.2856 202.56 0.2130 146.36 0.1819 172.50 0.3589
ˇdown −120.54* 0.0680 122.57 0.3554 74.56 0.4396 102.26 0.6131
R2 0.0308 0.0161 0.0223 0.0146
%ofcontracts 42.09 45.33 39.09 40.67
Thistablereportsthepredictiveregressionresultsconditionalondailyoptionstradingvolumeandtraderclasses;the367 tradingdaysaresortedintothreeequal-sizedsub-samplesbasedonthedailyoptiontradingvolume.Theregressionmodel carriedoutforeachvolumetertileis:
Rt+1=˛i+ˇi,upOB,volumepercentileXi,tOB,volumepercentile×Dt+ˇi,downOB,volumepercentileXi,tOB,volumepercentile×(1−Dt)+εi,t
Dt=
1,Rt>0;
0,otherwise
,
whereRt+1isthedailyclose-to-closespotindexreturnondayt+1;XSaretheopen-buyput-callratios;andDisanindicative
variablewhichisequalto1ifRt>0;otherwise0.Theopen-buyput-callratioiscalculatedbydividingtheopen-buyputsbythe
sumoftheopen-buyputsandcalls.The‘%ofContracts’referstothepercentageofopen-buyvolumewithinthespecifictrader class.Boldnumbersindicatecoefficientswithcorrectpredictivesignsandcorrectstatisticalsignificance.
*Indicatesstatisticalsignificanceatthe10%level. **Indicatesstatisticalsignificanceatthe5%level.
changeof63.82points,the−13.48basispointsapproximatetoa−11.43changeintheindexlevel
(−13.48×10−3×8476),whichis18percentofthemediandailyindexmovement(11.43/63.82).
Whentakingintoconsiderationtheslopecoefficientof176.63basispointsinEq.(3),thissuggests
thataonestandarddeviationchangeintheput-callratioisassociatedwitha−31.42indexmovement,
whichaccountsfor49percentofthemediandailyindexchangeinoursampleperiod.Theseresults
indicatethattheput-callratiosofforeigninstitutionalinvestorspredictanon-trivialportionofthe
indexmovements,andthus,shouldbeconsideredtobeofeconomicsignificance.Theslopecoefficients
aresixtimesgreaterwhenweexamineasub-samplecontainingonlythosedayswithimportant
macroeconomicnewsevents(Table8).
5.2. Marketactivityandthepredictiveabilityofoptions
Wegooninthissectiontoexaminewhetherthereareanyvariationsinthepredictiveability
oftradingbyforeigninstitutionalinvestorsunderdifferentlevelsofoptionmarketactivity.Studies
havelongviewedtradingvolumeandinformationflowasinseparable;forexample,inthe‘sequential
arrivalofinformation’modelproposedbyCopeland(1976),thegradualarrivalanddisseminationof
newinformationresultsinmovementsinbothtradingvolumeandprice.17
17EvidenceofvolumestimulatedbyinformationwasfoundbyKimandVerrecchia(1991),whodemonstratedthattrading
Table5
Predictiveabilityoftheopen-buyratio,byoptionmarketcharacteristics.
Variables Foreigninstitutions Marketmakers Domesticinstitutions Individualinvestors Coefficient p-Value Coefficient p-Value Coefficient p-Value Coefficient p-Value
PanelA:moneyness
a.In-andnear-the-money
Intercept 17.56 0.2995 23.35 0.2032 –1.28 0.9390 –4.99 0.8076 ˇup −39.71 0.3262 −81.22 0.1528 3.51 0.9335 12.30 0.8482 ˇdown −40.05 0.1889 −47.71 0.1913 −2.82 0.9228 7.68 0.8727 R2 0.0048 0.0061 0.0001 0.0001 %ofcontracts 31.35 41.99 36.00 38.24 b.Out-of-the-money Intercept 24.16 0.2047 −35.73 0.0917 −14.21 0.5240 −25.79 0.2290 ˇup −36.62 0.2833 70.25 0.0733 22.02 0.5327 60.15 0.1891 ˇdown −54.04* 0.0943 65.21 0.1453 19.61 0.6000 59.37 0.3403 R2 0.0078 0.0088 0.0011 0.0048 %ofcontracts 50.40 43.61 44.82 42.07
A.3Deeplyout-of-the-money
Intercept −10.77 0.5164 −31.91 0.1400 −32.06 0.1152 −34.39* 0.0607 ˇup 22.52 0.3881 42.30 0.1455 45.45* 0.0955 55.99** 0.0798
ˇdown 3.13 0.9056 36.94 0.2778 39.49 0.2211 68.65 0.1160
R2 0.0031 0.0060 0.0083 0.0093
%ofcontracts 18.25 14.40 19.18 19.69
PanelB:timetomaturity
a.Short(lessthan30days)
Intercept 19.35 0.3391 −21.62 0.5322 14.23 0.6353 −12.74 0.7582 ˇup −27.20 0.4794 54.54 0.4893 −22.07 0.6925 38.01 0.7108 ˇdown −49.96 0.1565 31.82 0.6534 −34.13 0.4965 16.92 0.8798 R2 0.0066 0.0019 0.0018 0.0012 %ofcontracts 71.84 91.87 91.09 93.21 b.Medium(30–90days) Intercept 22.01 0.1525 9.05 0.6272 −23.67 0.2804 −9.11 0.5824 ˇup −38.32 0.2000 −25.06 0.4894 54.41* 0.0912 37.24 0.4601 ˇdown −51.65* 0.0517 −17.84 0.5774 27.40 0.4198 8.18 0.8981 R2 0.0106 0.0014 0.0136 0.0021 %ofcontracts 26.69 7.86 8.70 6.65
c.Long(morethan90days)
Intercept −8.87 0.6215 0.88 0.9617 5.18 0.9619 9.73 0.4447
ˇup 26.43 0.3807 2.98 0.9143 −4.31 0.9716 −33.68 0.3702
ˇdown 13.32 0.6481 −13.24 0.6082 5.76 0.9626 −45.61 0.3047
R2 0.0029 0.0021 0.0019 0.0036
%ofcontracts 1.47 0.27 0.21 0.13
Thistablereportstheresultsofthefollowingpredictiveregressionmodel,byoptioncharacteristics(moneynessormaturity) andtraderclasses:
Rt+1=˛i+ˇi,upOB,option leverageXi,tOB,optionleverage×Dt+ˇOB,optioni,down leverageXOB,optioni,t leverage×(1−Dt)+εi,t,
whereRt+1isthenext-dayspotindexreturn;XSaretheopen-buyput-callvolumeratioscalculatedforaspecifictype(moneyness
ormaturity)ofoptions;andDisanindicativevariablewhichisequalto1ifRt>0;otherwise0.Theopen-buyput-callratiois
calculatedbydividingtheopen-buyputvolumebythesumoftheopen-buyputandcallvolumeforthecorrespondingtypeof options.The‘%ofContracts’ineachpanelreferstotheopen-buyvolumeofthatcategoryasapercentageofthetotalopen-buy volumefortheentireclassoftraders.WedefineIn-andNear-the-Moneyoptionsasthosewithstrike-to-spotratiosbetween 0.98and1.02,Out-of-the-Moneycallsasthosewithstrike-to-spotratiosbetween1.02and1.07,andOut-of-the-Moneyputs asthosewithstrike-to-spotratiosbetween0.93and0.98.AllotheroptionsareclassifiedasDeeplyOut-of-the-Moneyoptions. Thoseoptionsthatexpirewithin30daysareclassifiedasshortmaturityoptions,thoseexpiringbetween30to90daysas mediummaturityoptions,andtheremainderaslongmaturityoptions.The‘%ofContracts’referstothepercentageofopen-buy volumewithinthespecifictraderclass.Boldnumbersindicatecoefficientswithcorrectpredictivesignsandcorrectstatistical significance.
*Indicatesstatisticalsignificanceatthe10%level. **Indicatesstatisticalsignificanceatthe5%level.
Analternativeexplanationoftheassociationbetweenvolumeandinformationwasalsooffered byKyle(1985)andAdmatiandPfeiderer(1988),whosuggestedthatinordertobothconcealtheir
informationandlowerthepriceimpact,informedinvestorswouldtendtotradewhenliquiditytrading
wasinplentifulsupply.18Boththeoriesimplythatthetimingofordersubmissionsbyinformedtraders
isrelatedtomarketactivity.
ToidentifywhetherinformedtradingontheTXOisdependentonmarketactivity,wesortthe367
tradingdaysintothreeequal-sizedsub-samplesbasedondailyoptiontradingvolume.Thepredictive
regressionEq.(1)isthenseparatelycarriedoutforeachvolumetertilesub-sample.Anegativeand
significantˇcoefficientwouldindicatethattheput-callvolumeratiocorrectlypredictsthe
next-dayindexreturn.AsshowninTable4,onlythetradingvolumeofforeigninstitutionalinvestorsin
thehighestvolumetertileisfoundtohaveasignificantˇcoefficientandasignconsistentwithits
predictiveability.InPanelCtheregressionR2 forforeigninstitutionalinvestorsisfoundtobethe
highestamongstalltraderclasses.Thisconcentrationofinformationtradingduringperiodsofhigh
volumeisconsistentwiththeimplicationsofthesequentialarrivalofinformationandtheintention
toconcealinformation.
Chordiaand Swaminathan(2000)foundthathigh-volumestocksexhibitedrapidresponsesto
market-wideinformation, whereaslow-volumestockstendedtorespondveryslowly;thus,they
assertedthattradingvolumeplayedasignificantroleinthedisseminationofmarket-wideinformation.
Inthepresentstudy,wepresentevidenceinsupportoftheirassertion,albeitintheindexoptions
market,wherepricesaredeterminedmainlybymarket-wideinformation.
Ourresultsalsoprovidesupportforthepropositionthattraderswilltrytoconcealtheirsuperior
informationbytradingwhenthereisgreaterliquidity-motivatedvolumewithinthemarket.
Empha-sizingthedifferencesbetweentradingbasedonprivateinformationandpublicinformationthatis
subjecttodifferentinterpretations,Bamberetal.(1999)reportedthattradingbasedontraders’own
interpretationsofpublicly-availableinformationtendedtobemoreintensivewhentradingvolume
washigher,essentiallybecauseahighvolumeofliquiditytradinghelpstocamouflagetheir
trans-actions.Ourfindingofthesignificantpredictiveabilityoftradingbyforeigninstitutionalinvestors
duringperiodsofhighvolumereinforcesthefindingsofBamberetal.(1999)thattradersactingon
theirdifferentialinterpretationelecttotradewhenvolumeishigher.
TheremainingfindingsinTable4arelargelyconsistentwiththeresultsreportedinTable2,with
thesignificantpredictiveabilityoftheoptionpositionsofforeigninstitutionalinvestorsonlybeing
discernibleinadownwardmarket.Inthemarketmakerregressions,thetwoˇcoefficientsinthe
lowestvolumetertilearefoundtobepositivewithstatisticalsignificance,therebyindicatingthatthe
obligationtoprovideliquiditywhenoverallliquidityislowexposesoptionmarketmakerstoadverse
selectionrisk.Theput-callratiosofindividualinvestorsandthevolumeratiosofdomesticinstitutional
investorsarefoundtohavelittleornoinformationcontentonfutureindexmovementsinanyofthe
volumetertiles.
5.3. Contractselectionbyinformedtraders
Inthissection,weexplorewhetherinformedtradersfavorcertaintypesofoptionswhenadopting
anopen-buystrategy.Specifically,weexaminethepredictiveabilityofopen-buyvolumeforoptions
atvariouslevelsofmoneynessandmaturity.
ItwasarguedinbothBlack(1975)andEasleyetal.(1998)thatinformedtradersoftenprefertotrade
inequityoptions,asopposedtotheunderlyingasset,essentiallybecausethehighleverageavailable
inoptionstradingraisesthepotentialprofitsfrominformedtrading.Accordingly,whenfacedwith
multipleseriesofoptions,informedtradersmaychooseout-of-the-moneyoptions,sincetheyprovide
alsoreportedabnormallyhightradingvolumeinequityoptionspriortotakeoverannouncements,indicatingthattraderswith superiorability,intermsoftheirinterpretationofinformation,tradeaheadofsuchannouncements.
18Chakravartyetal.(2004)foundthatoptionmarketsweremoreinformativewhenthetradingvolumewashighandthe
effectivespreadswerenarrower.Blauetal.(2009)providedempiricalevidencetoshowthatasymmetricinformationwas greaterduringperiodsofhighvolume,wheninformedtraderswereabletosubmittheirorderswithoutrevealingtheirprivate information.
higherleveragethaneitherin-the-moneyornear-the-moneyoptions.Theleverageconsiderationof
contractselectionisempiricallysupportedbyChakravartyetal.(2004),whofoundsignificantintraday
pricediscoveryforout-of-the-moneyoptionsascomparedtoat-the-moneyoptions,andKauletal.
(2004),whoshowedthatinformedtradersspecificallychoosetotradeinoptionswithgoodliquidity
andhighleverage.
Weclassifyoptionsintothreecategoriesofleverage,basedupontheirstrike-to-spotratios.
In-the-moneyandnear-the-moneyoptionsarecallsandputswithstrike-to-spotratiosbetween0.98
and1.02,whilstout-of-the-moneyoptionsarecallswithstrike-to-spotratiosbetween1.02and1.07
andputswithstrike-to-spotratiosbetween0.93and0.98.Deeplyout-of-the-moneyoptionsarecalls
withstrike-to-spotratiosabove1.07andputswithstrike-to-spotratiosbelow0.93.Thepredictive
regressioninEq.(1)isthencarriedoutonceagainusingthevolumeratioconstructedfromoptionsin
eachcategoryofmoneyness.
AsshowninPanelAofTable5,theonlysignificantput-callvolumeratioisfoundinthe
out-of-the-moneyoptionstradingofforeigninstitutionalinvestorsduringadownwardmarket;although
in-the-moneyandnear-the-moneyoptionstradedbythisgroupconveycorrectinformationon
next-dayindexmovements,thecoefficientsarenotstatisticallysignificant.Noneoftheothertraderclasses
exhibitanysignificantpredictiveabilityinanycategoryofmoneyness.Thesefindingsareconsistent
withtheleveragehypothesis,thatinformedtradersprefertousehigh-leveragecontractsinorder
tocapitalizeontheirsuperiorinformation.Ourresultsalsoconfirmthefindingsofthestudiesof
Ahnetal.(2008),Changetal.(2009)andChanetal.(2009),eachofwhichreportedatendencyfor
out-of-the-moneyoptionstoleadtheequityindexinpricediscovery.
Asidefromleverage,liquidityisanadditionalandcriticalconsiderationintheselectionofwhich
contractstotradein.Informed tradershaveanincentivetotrade ina liquidmarket inorder to
concealtheirprivateinformationandminimizethemarketimpactcosts(Kyle,1985).Weuse
time-to-expirationasaproxyforoptionliquidity,classifyingtheoptionsintothreematurityranges:less
than30days,between30and90days,andlongerthan90days,wheretheshort-term(lessthan
30days)optionsarethemostliquidandthelong-term(longerthan90days)optionsaretheleast
liquid.
Our predictive regressionsusingthe volumeratioconstructed fromoptionsin each maturity
categoryarepresentedinPanelBofTable5.Surprisingly,theonlyvariablewithasigncorrectly
andsignificantlypredictingthenext-dayindexreturnistheopen-buyratioofforeigninstitutional
investorsinmedium-termoptions,andnot,asmightbeexpected,theshort-termoptionswiththe
bestliquidity.Thisfindingis,nevertheless,consistentwiththosereportedbyHanetal.(2009)and
Changetal.(2009),bothofwhichundertookanalysesintoTXOoptionsandfoundbetterinformation
contentinmedium-maturityoptionsascomparedtoshort-maturityoptions.
Thisresultobviouslygivesrisetothequestionofwhyforeigninstitutionalinvestorsdonotchoose
toconcentratetheirinformedtradinginshort-termcontracts,wherethereissufficientliquidity,and
theanswermayrelatetothetime-decaynatureofoptions.Theclosertheexpirationdate,thelarger
thetheta,andthus,themoretheoptionvalueisdiminishedwitheachpassingday.Theverynatureof
increasingvaluedecayovertimeisdetrimentaltoholdingshort-termlongoptionpositions.Informed
traders,whotendtoholdlongpositions(recallourfindingsinTables2and4),canmitigatetheextent
ofthevaluedecaybyselectingcontractswithlongertime-to-expiration.
Anadditionalmeritoftheuseoflonger-termoptionsisthatout-of-the-moneyoptionshavegreater
deltaswhentheoptionshavelongertime-to-expiration(Bakshietal.,2000).Thus,inordertogain
themaximumpossiblebenefitfromtheirinformationaladvantage,foreigninstitutionalinvestorswill
tendtoselectthemedium-termcontracts,withhighdeltaexposure,amongtheout-of-the-money
options.Thiscontractselectionofinformedtradersisalsorevealedbytheirproportionalholdingof
optionsacrossmaturities.
Asshowninthe‘%ofcontracts’inPanelBofTable5,withtheexceptionofforeigninstitutional
investors,alltradersdevotedover90percentoftheirvolumetoshort-termoptions.Incontrast,foreign
institutionalinvestorsarefoundtohavetradedmoreinmedium-termoptions(26.69percent)than
othertraders(lessthan9percent).Theregressionresultsthereforesuggestthatthepreferencefor
medium-termoptionsamongstforeigninstitutionalinvestorscouldbemotivatedbytheopportunity
Insummary,wefindthattheout-of-the-moneyandmedium-maturityoptionstradedbyforeign
institutionalinvestors havericherinformationcontent,andthatthedecisionsmadebyinformed
traderswithregardtotheircontractselectionreflecttheirwillingnesstosacrificeliquidityforhigh
leverage,highdeltaandlowtheta.
5.4. Stealthtradingintheindexoptionsmarkets
Inthissection,wegoontoexploretheinformation contentofoptionstradingusingdifferent
tradesizesinanattempttoprovidedirectevidenceof‘stealthtrading’intheindexoptionsmarket.
Thestealth-tradinghypothesisproposedbyBarclayetal.(1993)suggeststhatinformedtraderswill
oftenfragmenttheirlargeordersinordertoreducetheimpactonpricesandslowdowntheprocess
ofdisclosureoftheirvaluableinformation.Asaresult,itisthemedium-sizedtradeswhichtendto
havetherichestinformationcontentandwhicharemostlikelytomoveprices.Althoughahandfulof
studieshaveprovidedsupportforthishypothesisinthestockmarkets,thereislessevidencewithin
thederivativemarkets.19
FollowingAnandandChakravarty(2007),wedefinesmall-sizedoptionstradesasthose
transac-tionscomprisingof1–4contracts,medium-sizedtradesasthoserangingbetween5and99contracts,
andlarge-sizedtradeasthoseinvolving100contractsormore.Wethengoontore-calculatethe
put-callratiosforeachclassoftraders,bytrade-sizegroups,andundertakethepredictiveregression
inEq.(1)foreachofthesetrade-sizegroups.TheresultsarereportedinTable6.
Theˇdowncoefficientintheregressiononforeigninstitutionalinvestorsinadownwardmarketis
negativeforallthreetradesizes;however,itisonlyinthemedium-sizedregression(PanelB)thatthe
coefficientisfoundtobestatisticallysignificant.Thesignificantinformationalroleofthe
medium-sizedtradesmadebyforeigninstitutionalinvestorsisconsistentwiththestealthtradinghypothesis.
Indeed,ChouandWang(2009)notedfrequentorder-splittingbehavioramongstforeigninstitutional
investorsintheTaiwanfuturesmarket.
Itwouldseemthatwhenengagingininformedtrading,foreigninstitutionalinvestorsinTaiwan
tendtosplittheirlargeordersintomedium-sizedorders,withtheirmedium-sizedtradesultimately
providingthebestpredictionsonfutureindexmovements.Ourfindingsonindexoptionsare
consis-tentwiththosereportedonequityoptionsbyAnandandChakravarty(2007),whereinformedtraders
werealsofoundtoprefermedium-sizedtrades.
AsshowninTable6,approximately34percentofthetransactionsmadebyretailtradersare
small-sizedtrades,withthesetradessignificantlypredictingthewrongdirectionofchangesintheindex.
AccordingtoEasleyandO’Hara(1987),smalltradesarelikelytobeattributabletonoisetraders;we
alsoshowthatsmalltradesbyretailinvestorshaveverylittleinformationcontent,andindeed,wefind
thatsuchtradestendtosufferfromimmediateshort-termlosses.Ourresultsonindividualtraders
areconsistentwiththeevidenceontheTaiwanstockmarketprovidedbyBarberetal.(2009),where
retailinvestorswerefoundtosuffersubstantialtradinglosses.
5.5. Choiceofordertypesbyinformedtraders
Anothertradingdecisionwhichhastobemadebyinformedtradersintheoptionsmarketisthe
choicebetweenmarketordersandlimitorders.Underthetraditionalview,theassumptionisthat
informedtraderswillplacemarketordersonly,essentiallybecausetheimmediacyoftheseorders
allowssuchinformedtraderstotakeuptheirpositionsbeforetheirinformationleaksout(Rock,1990;
19Asregardsevidenceonthestockmarkets,Barclayetal.(1993)foundthatmedium-sizedtradesaccountedforanestimated
92.8percentofthecumulativepricechangeduringpre-tenderofferannouncementperiods;Chakravarty(2001)notedthat informativetradesontheNYSEwerealmostentirelyattributabletothemedium-sizedtradesinitiatedbyinstitutions;and AlexanderandPeterson(2007)providedfurtherevidenceofincreasedclusteringofmedium-sizedtradesontheNYSEand Nasdaq,whichtendedtohavegreaterpriceimpactthanlargeroundedtrades.Thefirstevidenceonderivativesmarketswas presentedbyAnandandChakravarty(2007),whodocumentedapreferenceamongstinformedtradersformedium-sizedtrades inequityoptions.