Index futures, spot volatility, and liquidity: Evidence from FTSE Xinhua A50 index futures. Yakup Eser Arisoy *

Similar documents
DOES TRADING VOLUME INFLUENCE GARCH EFFECTS? SOME EVIDENCE FROM THE GREEK MARKET WITH SPECIAL REFERENCE TO BANKING SECTOR

expressed here and the approaches suggested are of the author and not necessarily of NSEIL.

CALCULATION OF OMX TALLINN

Chapter 8: Regression with Lagged Explanatory Variables

Day Trading Index Research - He Ingeria and Sock Marke

Market Liquidity and the Impacts of the Computerized Trading System: Evidence from the Stock Exchange of Thailand

Return performance, leverage effect, and volatility spillover in Islamic stock indices evidence from DJIMI, FTSEGII and KLSI

Bid-ask Spread and Order Size in the Foreign Exchange Market: An Empirical Investigation

THE EFFECTS OF INTERNATIONAL ACCOUNTING STANDARDS ON STOCK MARKET VOLATILITY: THE CASE OF GREECE

CALENDAR ANOMALIES IN EMERGING BALKAN EQUITY MARKETS

SAMUELSON S HYPOTHESIS IN GREEK STOCK INDEX FUTURES MARKET

Investor sentiment of lottery stock evidence from the Taiwan stock market

A Note on the Impact of Options on Stock Return Volatility. Nicolas P.B. Bollen

Why Did the Demand for Cash Decrease Recently in Korea?

SPEC model selection algorithm for ARCH models: an options pricing evaluation framework

WHAT ARE OPTION CONTRACTS?

The Relationship between Stock Return Volatility and. Trading Volume: The case of The Philippines*

How Useful are the Various Volatility Estimators for Improving GARCH-based Volatility Forecasts? Evidence from the Nasdaq-100 Stock Index

A DCC Analysis of Two Stock Market Returns Volatility with an Oil Price Factor: An Evidence Study of Singapore and Thailand s Stock Markets

Measuring the Downside Risk of the Exchange-Traded Funds: Do the Volatility Estimators Matter?

GOOD NEWS, BAD NEWS AND GARCH EFFECTS IN STOCK RETURN DATA

Hedging with Forwards and Futures

Estimating Time-Varying Equity Risk Premium The Japanese Stock Market

Does Option Trading Have a Pervasive Impact on Underlying Stock Prices? *

Microstructure of Russian stock market and profitability of market making

THE IMPACT OF SHORT SALE RESTRICTIONS ON STOCK VOLATILITY: EVIDENCE FROM TAIWAN

Principal components of stock market dynamics. Methodology and applications in brief (to be updated ) Andrei Bouzaev, bouzaev@ya.

Ownership structure, liquidity, and trade informativeness

Morningstar Investor Return

The Greek financial crisis: growing imbalances and sovereign spreads. Heather D. Gibson, Stephan G. Hall and George S. Tavlas

The predictive power of volatility models: evidence from the ETF market

The impact of the trading systems development on bid-ask spreads

SURVEYING THE RELATIONSHIP BETWEEN STOCK MARKET MAKER AND LIQUIDITY IN TEHRAN STOCK EXCHANGE COMPANIES

Asymmetric Information, Perceived Risk and Trading Patterns: The Options Market

Measuring macroeconomic volatility Applications to export revenue data,

4. International Parity Conditions

Skewness and Kurtosis Adjusted Black-Scholes Model: A Note on Hedging Performance

Evidence from the Stock Market

Contrarian insider trading and earnings management around seasoned equity offerings; SEOs

Illiquidity and Pricing Biases in the Real Estate Market

THE NEW MARKET EFFECT ON RETURN AND VOLATILITY OF SPANISH STOCK SECTOR INDEXES

The Influence of Positive Feedback Trading on Return Autocorrelation: Evidence for the German Stock Market

Resiliency, the Neglected Dimension of Market Liquidity: Empirical Evidence from the New York Stock Exchange

Journal Of Business & Economics Research September 2005 Volume 3, Number 9

Markit Excess Return Credit Indices Guide for price based indices

Information Leadership in Advanced Asia-Pacific Stock Markets: Returns. and Volatility Spillover and the role of public information from the U.S.

Cointegration: The Engle and Granger approach

On Overnight Return Premiums of International Stock Markets

Vector Autoregressions (VARs): Operational Perspectives

Applied Econometrics and International Development Vol.7-1 (2007)

Flight-to-Liquidity and Global Equity Returns

Supplementary Appendix for Depression Babies: Do Macroeconomic Experiences Affect Risk-Taking?

Do Public Income Transfer to the Poorest affect Internal Inter-Regional Migration? Evidence for the Case of Brazilian Bolsa Família Program

Oil Price Fluctuations and Firm Performance in an Emerging Market: Assessing Volatility and Asymmetric Effect

Market Efficiency or Not? The Behaviour of China s Stock Prices in Response to the Announcement of Bonus Issues

Stock market returns and volatility in the BRVM

THE IMPACT OF CUBES ON THE MARKET QUALITY OF NASDAQ 100 INDEX FUTURES

Modelling and Forecasting Volatility of Gold Price with Other Precious Metals Prices by Univariate GARCH Models

Why does the correlation between stock and bond returns vary over time?

Does Option Trading Have a Pervasive Impact on Underlying Stock Prices? *

The Economic Value of Volatility Timing Using a Range-based Volatility Model

Does informed trading occur in the options market? Some revealing clues

Florida State University Libraries

MODELING SPILLOVERS BETWEEN STOCK MARKET AND MONEY MARKET IN NIGERIA

I. Basic Concepts (Ch. 1-4)

Commission Costs, Illiquidity and Stock Returns

APPLICATION OF Q-MEASURE IN A REAL TIME FUZZY SYSTEM FOR MANAGING FINANCIAL ASSETS

An Econometric Analysis of Market Anomaly - Day of the Week Effect on a Small Emerging Market

An Empirical Comparison of Asset Pricing Models for the Tokyo Stock Exchange

The stock index futures hedge ratio with structural changes

An asymmetric process between initial margin requirements and volatility: New evidence from Japanese stock market

The Behavior of China s Stock Prices in Response to the Proposal and Approval of Bonus Issues

Nikkei Stock Average Volatility Index Real-time Version Index Guidebook

Option Put-Call Parity Relations When the Underlying Security Pays Dividends

ANOMALIES IN INDIAN STOCK MARKET AN EMPIRICAL EVIDENCE FROM SEASONALITY EFFECT ON BSEIT INDEX

Risk Modelling of Collateralised Lending

A closer look at Black Scholes option thetas

Price, Volume and Volatility Spillovers among New York, Tokyo and London Stock Markets

Usefulness of the Forward Curve in Forecasting Oil Prices

Can Individual Investors Use Technical Trading Rules to Beat the Asian Markets?

Chapter 6: Business Valuation (Income Approach)

Default Risk in Equity Returns

How To Calculate Price Elasiciy Per Capia Per Capi

Migration, Spillovers, and Trade Diversion: The Impact of Internationalization on Domestic Stock Market Activity

Empirical Investigation of Price Variability Mechanism of the Colombo All Share Price Index

Migration, Spillovers, and Trade Diversion: The Impact of Internationalization on Domestic Stock Market Activity

Relationship between Stock Returns and Trading Volume: Domestic and Cross-Country Evidence in Asian Stock Markets

Monetary Policy & Real Estate Investment Trusts *

The Relationship between Trading Volume, Returns and Volatility: Evidence from the Greek Futures Markets CHRISTOS FLOROS. Abstract

Does Stock Price Synchronicity Represent Firm-Specific Information? The International Evidence

Equities: Positions and Portfolio Returns

INTEREST RATE FUTURES AND THEIR OPTIONS: SOME PRICING APPROACHES

Duration and Convexity ( ) 20 = Bond B has a maturity of 5 years and also has a required rate of return of 10%. Its price is $613.

DO FUNDS FOLLOW POST-EARNINGS ANNOUNCEMENT DRIFT? RACT. Abstract

Volatility in Returns of Islamic and Commercial Banks in Pakistan

Causal Relationship between Macro-Economic Indicators and Stock Market in India

Appendix D Flexibility Factor/Margin of Choice Desktop Research

Predicting Implied Volatility in the Commodity Futures Options Markets

The Transmission of Pricing Information of Dually-Listed Stocks

Market Maker Inventories and Stock Prices

Transcription:

Index fuures, so volailiy, and liquidiy: Evidence from FTSE Xinhua A5 index fuures Yaku Eser Arisoy * Absrac This aer examines he imac of he inroducion of he FTSE Xinhua A5 index fuures conrac on he volailiy and liquidiy of is underlying. The resuls indicae a significan increase in so volailiy and liquidiy in he os-fuures eriod. The condiional volailiy esimaions sugges ha he change in volailiy is aribued o an increase in he rae of flow of informaion o he so marke, raher han seculaive rading. Afer conrolling for facors ha affec liquidiy, we find confirmaory evidence on he hyohesis ha inroducion of fuures rading induces migraion of uninformed raders from so marke o fuures marke. Overall, he resuls have imlicaions for financial regulaors and olicy-makers regarding he ineracion beween fuures and so markes, and inegraion of financial markes. JEL Codes: G5, G9 Keywords : fuures inroducion, so volailiy, so liquidiy, volailiy ersisence, informaion asymmery * Assisan Professor of Finance, IESEG School of Managemen.

. Inroducion Since he inroducion of he S&P 5 index fuures conracs, here has been a vas amoun of lieraure examining he imac of sock index fuures on is underlying so marke, and i has been one of he mos fruiful areas of emirical financial research. The majoriy of he effor is devoed o exlain he relaionshi beween fuures rading and so volailiy. On he oher side, here is he lieraure ha examines he effec of sock index fuures on so liquidiy, and informaion asymmery. 2 The aim of his sudy is o resen recen evidence on he lieraure regarding he relaionshi beween index fuures rading, so volailiy, and liquidiy, and he focus of ineres will be on he newly inroduced FTSE Xinhua A5 index fuures. The conribuions are execed o be hreefold. Firs of all, alhough here have been sudies on he imac of he inroducion of index fuures conracs on so volailiy and liquidiy searaely, his sudy reconciles hese wo srands of lieraure in one single seing. Second, as far as he auhor knows, his is he second sudy afer Lee and Ohk (992) ha examines he effec of fuures inroducion in a foreign exchange on he domesic so marke, and hird, i uses a unique samle ha has no been sudied before. The case of he inroducion of index fuures covering Chinese socks is worh invesigaing, because he Chinese financial markes disinguishes from oher develoed or emerging markes due o is unique exerience wih fuures rading. The firs sriking See Soll and Whaley (987), Edwards (988a, b), Grossman (988), Harris (989), Bechei and Robers (99), Bessembinder and Seguin (992), Anoniou and Holmes (995), Pericli and Koumos (997), Chang, Cheng and Pinegar (999), Gulen and Mayhew (2), and Yang, Balyea, and Leaham (25) for he imac of index fuures on volailiy. 2 See Gammill and Perold (989), Goron and Pennacchi (99), and Subrahmanyam (99) for he imac of index fuures on informaion asymmery, and Jegadeesh and Subrahmanyam (993) for he imac on liquidiy.

difference is he brief exerience ha China had wih fuures conracs beween 993 and 995. Bu he fuures rading had o be haled due o illiquid and seculaive rading. I would be worhwhile invesigaing he deerminans and consequences of his brief exosure and comare i wih he resuls obained here, bu unforunaely here is no daa available for ha eriod. The second difference comes from he origin of he sock exchange of inroducion. I was an inernaional sock exchange, Singaore Sock Exchange, which launched he FTSE Xinhua A5 index fuures, covering he 5 major ublicly comanies raded in mainland China. Thus, examining he imac of an index fuures conrac ha has is underlying in a differen counry is execed o yield ineresing resuls. The resuls are execed o shed some ligh on he inerdeendence and inegraion of financial markes. Furher, he resuls of he sudy are execed o give some reliminary informaion for Chinese regulaors, who are reared o launch heir second sock index fuures soon, and oher emerging marke regulaors who lan o launch index fuures conracs inernaionally due o fears of desabilizaion in domesic markes. Overall, he resuls can be summarized as follows. Firs, using a family of GARCH models and afer conrolling for he common characerisics ha drive he marke, we find ha here is a significan increase in os-fuures so volailiy, which is aribued o he increased seed and qualiy of informaion flowing o so markes. Second, he markes are more efficien in he sense of cauring differen effecs of recen informaion and incororaing hem in he rices. Furhermore, afer conrolling for facors ha affec liquidiy, we find ha even if he fuures inroducion has had a osiive imac on sreads, here has been a significan decrease in he average bid-ask sreads in he os-

fuures eriod due o an increase in volume, rice and volailiy. Overall, he resuls imly an increased rading volume, and more volaile, bu more efficien markes. The resuls have imlicaions for olicy-makers regarding he inegraion of inernaional financial markes, he ineracion beween fuures and so markes. The remainder of his aricle is organized as follows. Secion 2 resens he relaed lieraure regarding he imac of inroducion of index fuures on so volailiy and liquidiy. Secion 3 resens he daa and mehodology used in he analyses, and Secion 4 documens emirical findings. The final secion offers concluding remarks. 2. Relaed Lieraure The general aroach in he lieraure u o now has been o examine he effec of fuures rading on he volailiy and liquidiy of is underlying searaely. This secion follows his general aroach, and resens imoran findings in hose wo branches of lieraure. 2.. Lieraure on volailiy Alhough here have been many sudies done for he las wo decades, oday, he lieraure on he effec of he inroducion of index fuures on so volailiy is less han conclusive. The evidence, hus far, on he effec of index fuures on so volailiy can be summarized as follows: i) he fuures index desabilizes he so markes, and ii) he fuures index sabilizes he so markes.

The desabilizaion heory argues ha he inroducion of fuures rading increases so volailiy. For examle, Harris (989) documens marginal increases in he variances of S&P 5 socks afer rading in S&P 5 index fuures began. Lockwood and Linn (99) reor similar variance increases when index fuures began rading in 982. Brorsen (99) finds ha fuures rading ends o reduce auocorrelaions and increase he volailiy of index sock reurns. Lee and Ohk (992) documen ha he volailiy of sock reurns in Ausralia, Hong Kong, Jaan, he U.K., and he U.S. rose significanly, following he inroducion of index fuures. On he oher hand, Anoniou and Holmes (995), and Anoniou, Holmes, and Priesly (997) also documen increases in so volailiies afer he inroducion of index fuures, however his increase is aribued o an increase in he rae of flow of informaion o so markes. On he oher side Edwards (988a, b), Grossman (988), and Bechei and Robers (99) find ha S&P 5 index fuures have an insignican imac on cash marke volailiy. Schwer (99) mainains ha he growh in sock index fuures and oions rading has no caused increases in volailiy. Similar conclusions are reached by Beckei and Robers (99), Kamara, Miller and Siegel (992), Pericli and Koumos (997), Galloway and Miller (997), and Dara, Rahman and Zhong (22), who documen ha inroducion of sock index fuures has eiher decreased or no significanly increased he volailiy in so markes, confirming he sabilizaion heory.

2.2 Lieraure on Liquidiy The lieraure on he relaionshi beween index fuures rading and he liquidiy of he underlying also has differen conclusions regarding he imac of fuures inroducion on so liquidiy. For examle, Silber (985) argues ha as raders, marke makers use index fuures in order o hedge heir invenory securiies, hey will be able o decrease he risk and he cos on hese securiies, and hus roose lower sreads, increasing he liquidiy of he underlying socks. On he oher hand, Gammill and Perold (989), Harris (99), Goron and Pennacchi (993), and Subrahmanyam (99) sugges ha fuures markes will arac uninformed raders because he imac of firm-secific informaion asymmery is lower in fuures markes. As a consequence of his, few uninformed raders will rade in sock markes, hus fixed comonen cos of markemakers will increase, and herefore marke-makers will increase he sreads. Moreover marke-makers will have an increased robabiliy of making a ransacion wih informed raders on he sock markes. This will induce hem o increase he sread beween bid and ask rices, in order o comensae for he higher risk of rading wih informed raders. Furhermore, if he migraion of uninformed raders dominaes, rading volume in he underlying socks should decrease, and conversely, if he effec of addiional informed rading dominaes, i should cause an increase in socks rading volume. Jegadeesh and Subrahmanyam (993) emirically sudy he imac of he inroducion of he S&P 5 fuures conracs on he liquidiy of he socks in he so index. Afer conrolling for facors consiuing he bid-ask sread, hey examine he change in he roorional bid-ask sread of he socks before and afer he sock index

fuures inroducion, and documen ha he average sread of he socks increases afer fuures inroducion. The auhors conclude ha index fuures rading decreases he liquidiy by drawing away uninformed raders from so markes o fuures markes. The nex secion resens he daa and he mehodology for he ess conduced o measure he imac of he inroducion of FTSE Xinhua A5 fuures on so volailiy and liquidiy. 3. Mehodology and Daa This aricle follows mehodologies oulined by Anoniou and Holmes (995) for ess on volailiy, and Jegadeesh and Subrahmanyam (993) for ess on liquidiy. The general aroach is o examine he so rice volailiy and liquidiy before he onse of fuures rading, and hen o comare his wih so rice volailiy and liquidiy osfuures. The nex wo subsecions describe hese aroaches in deail. 3.. Mehodology on ess of volailiy In analyzing he effec of fuures inroducion on volailiy, he firs imoran ask is o conrol for exerior effecs ha are no due o fuures rading. In oher words, one should isolae he influences ha are due o oher facors (marke-wide facors), so ha he imac of fuures rading can be assessed more direcly and recisely. This is exremely imoran for he Chinese index fuures sudied here, because he FTSE Xinhua A5 index has quadruled since he inroducion of index fuures, and much of

his increase is aribued o facors ha are unrelaed o fuures inroducion, such as increased ineres by foreign and domesic invesors. To isolae he effec of hese marke wide movemens on so volailiy, we follow he mehodology oulined by Anoniou and Holmes (995), and include a roxy variable ha caures marke-wide movemens and ha is no relaed wih fuures conracs. The nex ask is o idenify he aroriae model o measure volailiy. Today, here is a wide lieraure ha documens he exisence of serial correlaion and heeroskedasiciy in sock reurns. Furhermore, he lieraure now agrees ha volailiy is ime varying. 3 In addiion, one has o disinguish he relaionshi beween informaion and volailiy. This relaion is imoran, because as assered by Ross (989) any change in he rae of informaion flow will have a direc affec on he volailiy of he underlying asse. Thus, a good model should ake ino accoun hese wo asecs, i.e ime-variaion in volailiy and he direc relaionshi beween informaion and volailiy. The naural candidae urns ou o be he GARCH family develoed by Engle (982), and exended by Bollerslev (986), and Engle and Bollerslev (986). The Generalized Auoregressive Condiional Heeroscedasiciy (GARCH) model differs from oher saisical models by modeling he condiional variance. In GARCH models, he condiional variance deends no only on he squared residuals of he mean equaion bu also on is own as values. A furher advanage of GARCH family models is ha hey allow for volailiy clusering and ersisence, which is observed in financial daa. Therefore, by using an aroriae GARCH model, while conrolling for ime- 3 For a deailed review on he ime-series roeries of volailiy and ARCH models, see Bollerslev, Chou, and Kroner (992).

varying roery of volailiy, one can esimae he changes in he informaion flow, i.e. he imac of recen and old news, on volailiy. The family of GARCH models ha forms he basis of our ess for he whole eriod is reresened by: R s = a ar, ε ψ ε ~ ( ) N,h () h q 2 = α α iε i β jh j γdf (2) i= j= for he GARCH(,q) model; R s = a ar a2h, ε ψ ε ~ ( ) N,h (3) h q 2 = α α iε i β jh j γdf (4) i= j= for he GARCH-M(,q) secificaion; R s = a ar, ε ψ ε ~ ( ) N,h (5) h q = i i j j i= j= 2 2 α α ε β h λε d γdf (6) for he TARCH(,q) secificaion, and finally; R s = a ar, ε ψ ε ~ ( ) N,h (7) ( h ) = α ( α i z i λz i ) β j log( h j ) γdf log (8) i= j= for he EGARCH(,q) secificaion. Equaions (), (3), (5), and (7) reresen he condiional mean equaions, and (2), (4), (6), and (8) reresen he condiional variance equaions for each model. R is he daily change in log rices of he FTSE Xinhua A5 index, and R is he daily change in q s

log rices of he marke roxy variable. h is he condiional variance of he error erm, ε, d is a dummy variable where d if ε < =, and o. w. z ε = is he sandardized h residual. DF is he dummy variable which akes on values re-fuures, and osfuures. DF is naurally omied when esimaing volailiy models in re-fuures and os-fuures eriods. Modeling condiional volailiy in one of he four forms resened above will hel us evaluae he following quesions: 2. If yes, wha is he relaionshi beween informaion and volailiy re- and osfuures?. Does he inroducion of fuures rading have an effec on volailiy? The answer o he firs quesion lies in he significance of he dummy coefficien, γ. Afer conrolling for marke-wide facors caured by R, if γ urns ou o be significan hen one can asser ha he inroducion of fuures rading has had an imac on so marke volailiy. The answer o he second quesion will be given by dividing he samle ino wo sub-samles: re-fuures, and os-fuures. By comaringα and β, refuures and os-fuures, i is sraighforward o evaluae he imac of he rae of informaion flow on volailiy. 3.2. Mehodology on ess of liquidiy The lieraure on liquidiy idenified hree facors ha consiue he bid-ask sread. These are he informaion asymmery, fixed coss, and invenory coss. Benson and Hagerman (974), and Soll (978) have emirically invesigaed he deerminans of

he bid-ask sread, and found ha a large orion of he variaion in bid-ask sread can be exlained by differences in rice level, reurn variance, and volume of ransacions. The inuiion behind why hese variables are relaed o bid-ask sreads is as follows: For a given number of shares raded, a high rice for a sock will imly higher dollar volumes, hus decreasing fixed coss, imlying lower sreads. Second, a high sock volailiy imlies higher invenory risk for risk-averse marke-makers, and greaer oenial rofis for informed raders, inducing an increase in sreads. Finally, higher rading volume enables he marke-maker o offse his invenory balances more flexibly, imlying lower sreads. Therefore, i is necessary o conrol for changes in hese facors before analyzing he imac of fuures rading on so bid-ask sreads. I is also imoran o noe ha his ar of he analysis deals wih individual socks ha consiues he index, raher he index iself. This is due o he fac ha here is no bid-ask daa for an index. Taking ino accoun he above re-idenified facors, we aly a log-linear regression model wih ooled cross-secional and ime-series daa suggesed by Jegadeesh and Subrahmanyam (993), i.e. LNSPRD i = a a LNPRC bdf ε, i i a LNVOL 2 i a LNVAR i =, K, N and =, 2. 3 i (3) In he above secificaion, LNSPRDi is he naural logarihm of he average quoed ercenage sread (he quoed sread as he ercenage of he rice level), and LNPRC, i LNVOL, and LNVAR are he naural logarihms of he average monh- i i

end rices, average monhly rading volume, and he monhly reurn variance, resecively. The number of socks included in he regression is denoed as N, and = or 2 denoes he re- and os-fuures eriods. The dummy variable DFi akes on he value re-fuures, and os-fuures. The focus of ineres will be on he dummy coefficien, b, which indicaes how he bid-ask sread has changed afer he onse of fuures rading, and afer accouning for changes in oher sread deerminans. 3.3. Daa Chinese sock index fuures commenced heir rading on Seember 5 h 26. Traded on he Singaore sock exchange, he FTSE Xinhua A5 fuures is an indexbased conrac comrising of fify major A-share Chinese socks in erms of marke caializaion. The daa covers he eriod 9//25 o 9/3/27 for a oal of 484 rading days, which roughly corresonds o 2 monhs before and afer he inroducion. For ess on volailiy, we use he daily log reurns ln(p /P - ), where P,, and P - are he closing levels of he index, a and -, resecively. As saed in he revious secion, i is imoran o conrol for he marke-wide facors by incororaing a roxy variable in he mean equaion given by (). This roxy should be able o caure he general rend in he Chinese marke, no highly correlaed wih he FTSE Xinhua A5 index, and no associaed wih he fuures conrac. The firs roxy ha comes o mind is he SSE3 index which includes he 3 bigges comanies in erms of marke caializaions in he Shanghai and Shenzen Exchanges, bu unforunaely, his index inhibis many of he socks ha are included in he FTSE Xinhua A5 index, herefore

no saisfying he hird crieria. The case wih he FTSE Xinhua 6 index is similar. On he oher hand, he FTSE Xinhua Small Ca Index is comosed of comanies lised on he Shanghai and Shenzhen sock exchanges, which have a oo small marke caializaion o be lised on he FTSE Xinhua 6 Index. I saisfies he above hree crieria, and is herefore chosen as he roxy o conrol for marke-wide facors. 4 All daa is downloaded from Daasream, and afer excluding non-rading days, we end u wih 484 daily reurn observaions (242 re-fuures, and 242 os-fuures), which form he basis for ess on volailiy. Regarding he ess on liquidiy, he samle consiss of socks ha have been included in he FTSE Xinhua A5 index hroughou he whole samle eriod. Alying his crieria resuls in N = 28 socks ha form he basis for he ooled regression equaion given by (3). For each sock in he samle, he following daa were obained from Daasream: end-of-monh closing quoes for bid and ask rices daily and monhly closing rices daily number of shares raded The re-fuures (os-fuures) sread for a comany is hen simly he average of he monhly re-fuures (os-fuures) bid-ask sreads. Similarly, he re-fuures (osfuures) rice level of a sock is given by he average of he monhly re-fuures (osfuures) rice levels. The monhly rading volume is he cumulaive number of shares raded for ha comany for ha monh, and he associaed re-fuures (os-fuures) rading volume is calculaed by he average of he monhly re-fuures (os-fuures) 4 The correlaion of he FTSE Xinhua Small Ca reurns wih FTSE Xinhua A5 reurns is.7, and i does no conain any socks included in FTSE Xinhua A5 index.

rading volumes. Finally, he monhly reurn variance is esimaed by using daily log reurns and he re-fuures (os-fuures) reurn variance is calculaed by he average of he monhly re-fuures (os-fuures) reurn variances. 4. Emirical Findings 4. Imac on volailiy Before resening he resuls on he imac of fuures rading volailiy, we firs analyze he behavior of daily log reurns of FTSE Xinhua A5 index. Second, we focus on he dae of he inroducion of fuures rading, and examine wheher here has been a srucural change in he reurns of FTSE Xinhua A5 index. Finally, we analyze in deail he effec of fuures inroducion on volailiy. 4.. Descriive saisics As can be seen from Table, he sandard deviaion of daily reurns of he FTSE Xinhua A5 index is higher afer he inroducion of fuures rading. This is a reliminary indicaion ha fuures inroducion migh have led o a higher volailiy, bu we have o analyze his finding more in deail in order o come u wih a saisically meaningful exlanaion. Furhermore, here has been a significan increase in he mean daily reurns, and he samle exhibis negaive skewness and leokuric behavior. The highly significan Jarque-Bera saisics for all eriods sudied rejec he hyohesis ha he reurns of daily reurns of he FTSE Xinhua A5 index are normally disribued.

<<Inser Table here>> To furher analyze wheher he inroducion of fuures rading creaed a srucural change in he behavior of daily reurns of he FTSE Xinhua index, we conduc a cumulaive sum los (CUSUM) es suggesed by Taylor (2). The CUSUM es deecs he ossible oins of change in ime series daa. The CUSUM los, S i, are given by he following equaion, ( X X ) for i =,2, n Si = Si i K, where n is he number of observaions, X, X 2,, X n reresen daily log reurns of he FTSE Xinhua A5 index, X is he mean reurn, and S =. The CUSUM los give us an idea on how he FTSE Xinhua A5 reurn series behave around is mean. If he CUSUM char resemble an uward sloe during a eriod his indicaes ha he reurns in ha eriod end o be above he overall average, and similarly a segmen wih a downward sloe indicaes a eriod of ime where he reurns end o be below he overall average. Thus a sudden change in direcion of he CUSUM indicaes a shif in he value of he ime series which ends o be above he average insead of below or below insead of above. Thus, i shows a change of rend s value comared o he overall average. << Inser Figure here>> Figure los he CUSUM char wih he FTSE Xinhua A5 reurns from Augus 25 o Ocober 27. As can be observed from he CUSUM los, he srucure of FTSE Xinhua A5 reurns have changed beween 4/8/6 and 3//6. During his eriod

he FTSE Xinhua A5 sock index fuures conracs have been launched on he Singaore Sock Exchange, he 5h Seember 26. Thus, he CUSUM los sugges reliminary evidence ha a srucural change in reurn behaviour migh have occurred due o he inroducion of index fuures. In order o be able o confirm wheher his change is aribued due o a change in so volailiy induced by he inroducion of fuures conrac, condiional volailiy esimaions for he whole eriod and for he re-fuures and os-fuures eriod are conduced. 4..2. GARCH esimaions To formally es he effecs of he inroducion of fuures rading on volailiy, we firs esimae a variey of GARCH(,q), GARCH-M(,q), TARCH(,q), and EGARCH(,q) secificaions equaions ( wih =, 2, 3, 4, 5, and q =, 2, 3, 4, 5) where he condiional mean equaions are given by (), (3), (5), and (7), and condiional variance equaions are given by (2), (4), (6), and (8), resecively. The unreored log-likelihood raios and F-saisics indicae ha GARCH(,) model reresens he bes secificaion for modeling condiional volailiy hroughou he samle eriod. Table 2 reors he esimaion resuls of he GARCH(,) model for he whole eriod, and for re-fuures, and os-fuures resecively. To es he effec of fuures inroducion on volailiy for he whole eriod, he dummy variable DF is inroduced ino he GARCH(,) rocess. The associaed -saisics and he -values give us he significance of he esimaed coefficiens.

<<Inser Table 2 here>> The firs observaion is ha he dummy variable (γ) for he whole eriod is osiive and significan. The dummy coefficien suggess ha he so volailiy has increased due o he inroducion of he fuures conracs. However, he analysis of he dummy variable does no le us know he exac source of his increase. The lieraure has idenified he increase in he amoun of informaion disseminaed, and he seed of informaion being imounded in rices as wo facors ha migh resul in increased volailiy afer fuures inroducion. To find ou wheher more informaion is being ransmied o he marke due o fuures inroducion, we comue he uncondiional variances beween wo eriods, given by he formula UV = α ( α ) β. The reasoning behind comaring uncondiional variances in re-fuures and os-fuures eriods is as follows. Ross (989) heoreically develos a no-arbirage condiion ha shows he relaionshi beween he rae of informaion disseminaed and he volailiy in he marke. The condiion for no arbirage imlies ha he variance of rice change will be equal o he rae (or variance) of informaion flow. The imlicaion of his is ha he volailiy of he asse rice will increase as he rae of informaion flow increases. If his is no he case, arbirage ooruniies will be available. I follows, herefore, ha if fuures increase he flow of informaion, hen in he absence of arbirage ooruniies he volailiy of he so rice mus change. By comuing he uncondiional variance re-fuures and osfuures, we find UV re =.948, and UV os =.492. The five-fold increase in he

uncondiional variance is consisen wih more informaion being ransmied ino he marke in he os-fuures eriod. Nex, we examine he source of he increase in he conen of informaion. In oher words, we examine wheher his increase in he rae of informaion being ransmied is due o recen news, or old news. The coefficien of he squared error erm, α, relaes o he changes in he so rice only on he revious day, which is aribuable o marke secific facors. Thus, α gives us informaion abou he seed of incororaion of recen news ino rices. On he oher hand, he coefficien of he lagged volailiy, β, relaes o he volailiy on he revious day, which in urn incororaes informaion abou he older days. Thus, β gives us informaion abou he seed of incororaion of old news ino rices. By comaring he coefficiens in Table 2, we observe ha α >, and, os α, re β <. Thus, we conclude ha he launch of fuures rading increased he seed, os β, re of recen news, and decreased he seed of old news being incororaed ino he rices. The resuls make sense, because he increase in he flow of informaion is execed o lead o a decrease in he uncerainy regarding revious news. In addiion, a significanly lower β (also called he ersisence arameer) indicaes ha volailiy is much less ersisen in he os-fuures eriod. The sum of α and β re-fuures is close o uniy (.96), whereas i decreases significanly o.57 os-fuures, indicaing a significan decrease in ersisence os-fuures. 5 5 Since he sum is close o uniy re-fuures, we also erformed ess of saionariy o see wheher here exiss a uni roo in he reurn series esecially before fuures inroducion. The unreored resuls rejec he exisence of a uni roo boh a he re-fuures and os-fuures eriods.

Furhermore, he saisically insignifican β migh indicae ha GARCH(,) secificaion migh no be he bes reresenaion of condiional volailiy in he osfuures eriod. Therefore, we esimae differen secificaions condiional volailiy ha migh caure he difference in increased volailiy and informaion in he os-fuures eriod. Table 3 resens he comarisons of differen condiional volailiy models in he os-fuures eriod. <<Inser Table 3 here>> Looking a he coefficiens, log-likelihood raios and Akaike and Schwarz informaion crieria, we conclude ha an EGARCH(,) is he mos arsimonious model ha exlains os-fuures volailiy among he four models considered. α, and β are saisically significan, he model selecion crieria are he lowes. Furhermore, wha is sriking is ha,γ, he coefficien for he sandardized residual is also saisically significan and negaive. This indicaes he exisence of he leverage effec in FTSE Xinhua A5 reurns during he os fuures eriod. In oher words, in he os-fuures eriod we observe no only he effec of increased disseminaion of recen news being imounded ino rices, bu also he asymmeric effec of news in he condiional volailiy. The EGARCH(,) model caures no only he effec of increased rae of news disseminaion on he FTSE Xinhua A5 reurns volailiy, bu also differeniaes he asymmeric effec beween good news and bad news. A saisically significan and

negaive coefficien indicaes ha bad news (associaed wih negaive sandardized residuals) have a much higher imac on volailiy han good news. 6 Overall, our findings on he effec of fuures inroducion on FTSE Xinhua A5 reurn volailiy can be summarized as follows. The osiive and significan dummy coefficien indicaes ha FTSE Xinhua reurn volailiy significanly increased afer he inroducion of fuures rading. GARCH(,) is found o be he bes secificaion o model condiional volailiy for he whole samle eriod. Furhermore, a deailed comarison of α, and β re-fuures, and os-fuures suggess ha inroducion of fuures rading on he Chinese markes imroved he qualiy and seed of he informaion being imounded in he rices. The resuls also indicae less ersisen volailiy osfuures. Finally, we find ha EGARCH(,) is he bes secificaion o model osfuures volailiy, which caures he asymmeric relaionshi beween news and volailiy. The resuls imly ha alhough he inroducion of fuures has an increasing effec on volailiy, fuures rading can be used as a ool o develo he efficiency of Chinese financial markes by increasing he qualiy and he flow of he informaion ino he so marke, and by beer differeniaing he effec of differen news (good and bad) on he marke. Finally, o check he robusness of he above resuls, we also erformed several analyses considering he imac of he inroducion of FTSE Xinhua A5 index fuures on he more general Shanghai and Shenzhen 3 (SSE3) index, and furher by using 6 This is in line wih he henomenon ha increased marke volailiy coincides wih downward marke moves, reored by French, Schwer, and Sambaugh (987), and Glosen, Jagannahan, and Runkle (993). Engle and Ng (993) also show ha volailiy is more associaed wih downward marke moves due o he leverage effec.

he EGARCH mehodology. The resuls are similar o he ones drawn here. 7 Overall, he findings sugges ha he underlying Chinese so marke is more volaile, bu less ersisen in he os-fuures eriod. This increase in volailiy is viewed as a resul of increase in he flow of informaion o he so marke, rejecing he desabilizaion heory. 4.2. Imac on liquidiy This secion resens he resuls for he ess regarding he imac of he inroducion of FTSE Xinhua A5 index fuures on liquidiy. The measure of liquidiy is he average ercenage bid-ask sread of he socks, which were included in he index hroughou he whole samle eriod. We follow he mehodology oulined in he revious secion. 4.2. Tess on liquidiy As discussed reviously, he lieraure has idenified hree deerminans of he bidask sread: informaion asymmery, fixed coss, and invenory coss. Furhermore, emirical sudies show ha ha a large orion of he variaion in bid-ask sread can be exlained by differences in rice level, reurn variance, and volume of ransacions. Thus, i is imoran o see changes in hese arameers before examining he effec of inroducion of fuures rading on liquidiy formally. Table 4 resens he changes in he 7 The resuls are available uon reques.

average quoed sreads, average end-of-monh rice, average rading volume, and average esimaed variance, beween re-fuures and he os-fuures eriods. <<Inser Table 4 here>> As can be seen from he able, he resuls indicae a fory hree ercen decrease in he average quoed ercenage sreads from.8% o.62%, afer he inroducion of fuures conracs. Furhermore, here have been significan increases in he average endof-monh rices, average rading volumes, and average esimaed variances. According o heory, he increase in volume should lead o a decrease in fixed coss and he invenoryholding coss for he marke-makers, imlying o a decrease in he sreads. The increase in rices is also execed o decrease he fixed cos comonen for he marke makers. However he significan increase in reurn variance imlies more risk for he markemakers holding he invenory securiies, and hus hey are execed o increase he sreads due o a rise in volailiy. Alhough he decrease in average sreads is large and significan, i is remaure o draw conclusions on he effec of he inroducion of fuures conracs on he sread. The decrease in he bid-ask sread migh be due o changes in he hree facors oulined in he lieraure, or he inroducion of fuures rading, or boh. Thus, o exlore he effec of fuures rading on bid-ask sreads we conrol for he above-menioned variables and erform a log-linear regression model wih ooled cross secional ime series daa as shown in he mehodology ar. The resuls are resened in Table 5. <<Inser Table 5 here>>

As can be seen from Table 5, all he coefficiens are saisically significan. The hree variables oulined in he lieraure have a negaive imac on he logarihm of he average quoed ercenage sread of he rice level. The volume and he rice are in line wih he lieraure, which suggess a negaive relaion beween sreads and rices or volume. The negaive coefficien of he volailiy facor does no follow he resuls of Soll (978), and Jegadeesh and Subrahmanyam (993). Their heory redics ha a higher volailiy will imly more invenory risk for marke-makers inducing increase in sreads. However, on he oher hand, Admai and Pfleiderer (988) sugges a negaive relaionshi beween liquidiy and volailiy. They argue ha disguise of informaion raders among liquidiy raders lead o a negaive relaionshi beween liquidiy and volailiy variables. Our resuls are in line wih heirs, and migh indicae ha here exis informed raders who rade a imes of high liquidiy in order o disguise he informaion conen of heir rades. This migh in urn be a facor ha draws away uninformed raders from he marke. Regarding he effec of fuures rading, he osiive dummy coefficien indicaes ha he inroducion of he FTSE Xinhua A5 fuures conrac has had a negaive effec on he liquidiy of he 28 socks in he samle. This migh a firs sigh seem conradicory o he significan decrease a he sreads. However, he significan decrease in sreads is robably due o he fac ha here have been five-fold increases in rices, and wo-fold increases in volailiy and rading volume os-fuures. Therefore alhough he inroducion of fuures conracs had a reducing effec on he liquidiy of he consiuens of he index, he effecs of hese facors are more ronounced and dominae he negaive effec of fuures inroducion on liquidiy.

The difference in average sreads in he re- and os-fuures eriod is economically significan as well. The esimae from he ooled regression indicae ha he roorional sread has increased by more han 8% due o he inroducion of fuures markes. So, keeing all he oher facors fixed, he cos of urchasing, shares of sock for $ ha had a roorional sread of % in he re-fuures eriod would increase by $8.56, which is a significan amoun given he dollar size of he rade. In shor, heories redic a widening or a narrowing of he sread due o fuures inroducion. The osiive dummy coefficien we find means ha he sread has increased due o fuures rading confirming he move-away of uninformed raders from he so marke. This follow he conclusions of Gammill and Perold (989), Harris (99), Goron and Pennacchi (993), and Subrahmanyam(99) who sugges ha marke markers will have an increased robabiliy of making a ransacion wih informed raders on he sock markes. This will induce an increasing effec on he bid and ask sread, in order o comensae his higher risk. The resuls do no confirm he heory of Silber (995) who redics a decrease in sread due o he ossibiliy for he marke makers o hedge heir invenory orfolio wih he inroducion of fuures which give hem low-cos marke ooruniy. 5. Conclusion There have been many sudies done in he lieraure invesigaing he effec of fuures inroducion on so volailiy and liquidiy, however he resuls are less han conclusive. This aricle adds o he lieraure in wo dimensions: i) by examining a

reviously unsudied fuures conrac, he FTSE Xinhua A5 index fuures, and ii) by examining he effec of an inernaionally raded fuures conrac on he volailiy and liquidiy of is domesic underlying. The resuls indicae ha volailiy of he FTSE Xinhua A5 index has increased significanly afer he inroducion of sock index fuures by he Singaore Sock Exchange. Furher analyses imly ha his increase is due o he fac ha more informaion is imounded o rices afer he inroducion of fuures rading. Moreover, he seed and he naure of informaion also differ beween re-fuures and os-fuures eriods. More secifically, recen news is imounded more raidly ino rices osfuures, and os-fuures marke is more efficien in he sense ha i can differeniae good news from bad news, a henomenon no observed re-fuures. Furhermore, he incororaion of old news has significanly reduced in he os-fuures eriod, indicaing ha he marke is less ersisen o changes in volailiy. Regarding liquidiy, alhough he average quoed ercenage sreads have decreased os-fuures, afer conrolling for he effecs of rice, volume and volailiy, we find evidence ha fuures inroducion has a negaive effec on he liquidiy of is underlying. The resuls suor he heory ha inroducion of fuures rading draw away uninformed raders from sock markes. Overall, he inroducion of FTSE A5 fuures conracs imlies a more volaile, bu less ersisen, more efficien, and less liquid so markes. The findings resened in his aricle migh be ineresing for regulaors and olicy-makers who lan o launch he rading of sock index fuures domesically or inernaionally.

References Admai, A.R., Pfleiderer, P., 988. A heory of inraday aerns: Volume and rice variabiliy. Review of Financial Sudies,, 3-4. Anoniou, A., Holmes, P., 995. Fuures rading, informaion, and so rice volailiy: Evidence for he FTSE- sock index fuures conrac using GARCH. Journal of Banking and Finance, 9, 7-29. Anoniou, A., Holmes, P., Priesly, R., 998. The effecs of sock index fuures rading on sock index volailiy: An analysis of he asymmeric resonse of volailiy o news. Journal of Fuures Markes, 8, 5-66 Beckei, S., Robers, D.J., 99. Will increased regulaion of sock index fuures reduce sock marke volailiy? Federal Reserve Bank of Kansas Ciy, 33 46. Nov/Dec. Benson, G., Hagerman, R., 974. Deerminans of bid-ask sreads in he over-hecouner marke. Journal of Financial Economics,, 353-364 Bessembinder, H., Seguin, P.J., 992. Fuures rading aciviy and sock rice volailiy. Journal of Finance, 47, 25 234. Bollerslev, T., 986. Generalized auoregressive heeroscedasiciy. Journal of Economerics, 3, 37-327 Bollerslev, T., Chou, R.Y., Kroner, K.F., 992. ARCH modeling in finance: a review of he heory and emirical evidence. Journal of Economerics 52, 5 59. Brorsen, B.W., 99. Fuures rading, ransacions coss, and sock marke volailiy. Journal of Fuures Markes, 53-63. Chang, E.C., Cheng, J.W. and Pinegar, J.M., 999. Does fuures rading increase sock marke volailiy? The case of he Nikkei sock index fuures markes. Journal of Banking and Finance, 23, 727-753

Coeland, T.E, Galai, D., 983. Informaion effecs on he bid-ask sread. Journal of Finance, 38, 457-469. Darra, A.F., Rahman, S., 995. Has fuures rading aciviy caused sock rice volailiy? Journal of Fuures Markes, 5, 537 557. Darra, A.F., Rahman, S., Zhong, M., 22. On he role of fuures rading in cash marke flucuaions: Pereraor of volailiy or vicim of regre? Journal of Financial Research, 25, 43 444. Edwards, F.R., 988a. Does fuures rading increase sock marke volailiy? Financial Analyss Journal, 63 69. Edwards, F.R., 988b. Fuures rading and cash marke volailiy: sock index and ineres rae fuures. Journal of Fuures Markes 8, 42 439. Engle, R.F., 982, Auoregressive condiional heeroscedasiciy wih esimaes of he variance of UK inflaion. Economerica, 5, 987-8. Engle, R.F, Bollerslev, T., 986. Modeling he ersisence of condiional variances. Economeric Reviews. 5, -5 Engle, R.F., Ng, V.K., 993. Measuring and esing he imac of news on volailiy. Journal of Finance, 48,749-778. French, K., Schwer, W., Sambaugh, R., 987. Execed sock reurns and volailiy. Journal of Financial Economics, 9, 3-29. Galloway, T.M., Miller, M.J., 997. Index fuures rading and sock reurn volailiy: Evidence from he inroducion of Mid-Ca 4 index fuures, The Financial Review. 32, 845-866. Gamill, J.F., Perold, A.F., 989. The changing characer of sock marke liquidiy. Journal of Porfolio Managemen. 6, 3-8. Glosen, L.R., Milgrom, P.R., 985. Bid, ask and ransacion rices in a secialis marke wih heerogeneously informed raders. Journal of Financial Economics. 4, 7-

Glosen, L.R., Jagannahan, R., Runkle, D.E., 993. On he relaion beween he execed value, and he volailiy of nominal excess reurn on socks. Journal of Finance, 48, 779-8. Goron,G., Pennacchi, G., 993. Securiy baskes and index-linked securiies. Journal of Business. 66, -27. Gulen, H., Mayhew, S., 2. Sock index fuures rading and volailiy in inernaional equiy markes. Journal of Fuures Markes. 2, 66 85. Harris, L., 989. S&P 5 cash sock rice volailiies. Journal of Finance. 44,55-75. Ho, T.S.Y., and Soll, H.R., 98. Oimal dealer ricing under ransacions and reurn uncerainy. Journal of Financial Economics. 9, 47-73 Jegadeesh, N., Subrahmanyam, A., 993. Liquidiy effecs of he inroducion of S&P 5 index fuures conrac on he underlying sock. Journal of Business. 66, 7-87. Kamara, A., Miller, T., Siegel, A., 992. The effecs of fuures rading on he sabiliy of he S&P 5 reurns. Journal of Fuures Markes. 2, 645 658. Lee, S.B., Ohk, K.Y., 992. Sock index fuures lising srucural change in ime varying volailiy. Journal of Fuures Markes. 2, 493-59. Lockwood, L. J., Linn, S.C., 99. An examinaion of sock marke reurn volailiy during overnigh and inraday eriods 964-989. Journal of Finance. 45, 59-6. Pericli, A., Koumos, G., 997. Index fuures and oions marke volailiy. Journal of Fuures Markes. 7, 957-974. Roll, R., 984. A simle imlici measure of he effecive bid-ask sread in an efficien marke. Journal of Finance. 39, 27-39. Ross, S., 989. Informaion and volailiy: The no arbirage maringale aroach o iming and resoluion irrelevancy. Journal of Finance. 44, -7.

Schwer, W.G., 99. Sock marke volailiy. Financial Analyss Journal, 23 34 (May/June) Silber, W.L., 985. The economic role of financial fuures, in Anne Peck (ed.), Fuures Markes: Their Economic Role, 83-4 Soll, R.H., 989. Infering he comonen of he bid-ask sread: Theory and emirical es. Journal of Finance. 44, 5-34. Soll, R.H, Whaley, R., 988. Volailiy and fuures: Message versus messenger. Journal of Porfolio Managemen. 4, 2-22. Soll, R.H, Whaley, R., 99. Sock marke srucure and volailiy. Review of Financial Sudies. 3, 37-7. Subrahmanyan, A., 99, A heory of rading in sock index fuures. Review of Financial Sudies. 4, 7-5. Yang, J., Balyea, R.B., Leaham, D.J., 25. Fuures rading aciviy and commodiy cash rice volailiy. Journal of Business Finance and Accouning, 32, 297 323.

Table Summary saisics for he FTSE Xinhua A5 index Period n Mean Sd. Dev. Skewness Kurosis JB Whole eriod 484.5.76 -.77 7.9 384.89 *** Pre-fuures 242.4.5 -.5 4.4 2.97 *** Pos-fuures 242.25.94 -.3 6.4 42.25 *** Noe: This ables reors he mean, sandard deviaion, skewness, and kurosis of he daily log reurns of he FTSE Xinhua A5 index. Mean and sandard deviaion of reurns are reored in ercenages. JB is he Jarque-Bera es saisic for normaliy, and *** denoes he % significance level for rejecion of normaliy. The whole eriod is from 9//25 o 9/3/27, re-fuures eriod is from 9//25 o 9/4/26, and os fuures from 9/5/26 o 9/3/27. Afer excluding bank holidays and non-rading days, we end u wih 484 days of daa for he whole eriod, 242 for he re-fuures, and 242 days of daa for he os-fuures eriod.

Table 2 GARCH(,) esimaions Period n a a α α β Whole 484.34 (2.28 * ).56 (8.87 *** ).5 (.73 * ).2 (3.27 *** ).8 (3.88 *** ) Pre-fuures 242.4.57.4.4.82 (.94) (23.48 *** ) (.95) (2.53 ** ) (9.8 *** ) Pos-fuures 242.6.55 2.2.29.28 (3.62 *** ) (8.4 *** ) (2.29 ** ) (2.5 ** ) (.24) γ 3.43 (.98 ** ) - - Noe: The GARCH(,) secificaions are given by R h ε ε ψ ~ N (,h ) s = a ar, = for he whole eriod, and R h 2 α α ε β h γdf ε ε ψ ~ N (,h ) s = a ar, = 2 α αε βh s for he sub-eriods. R is he daily change in log rices of he FTSE Xinhua A5, R is he daily change in log rices of he FTSE Xinhua SmallCa index, and DF is a dummy variable ha akes on values re-fuures, and os-fuures. The coefficien a is mulilied by 3, and he α, and γ are mulilied by 6 coefficiens for exosiory uroses. The numbers in he aranheses are he -saisics, and ***, **, * denoe he %, 5%, and % significance levels, resecively. All he esimaions are adjused for Bollerslev-Woolridge robus sandard errors and covariance.

Model GARCH(,).6 GARCH-M(,) 2. TARCH(,).4 EGARCH(,).34 Table 3 Comarison of condiional volailiy models in he os-fuures eriod a a a 2 α α.55 2.2.29.28 (3.62 *** ) (8.4 *** ) (2.29 ** ) (2.5 ** ) (.24).56 -.8 2.27.29.25 (.92) (8.55 *** ) (-.2) (2.4 ** ) (2.6 ** ) (.94).54 2.48.6.8 (3.35 *** ) (.7 *** ) (3.22 *** ) (.92) (.5) (3.6 *** ) (.6 *** ) (-3.38 *** ) (3. *** ) (3. *** ) (-2.8 ** ) β γ 2LL AIC SC 87.55-7.5-7.8 87.9-7.4-7.5.49 874.65-7.8-7.9 (2.5 ** ).54-5.52.46.49 -.22 874.78-7.8-7. Noe: The GARCH(,) secificaion is given by s R a a R h ε ε ψ ~ N (,h ) =, = 2 αε βh α, he GARCH-M(,) secificaion is given by s R a a R a h h ε ε ψ ~ N (,h ) = 2, = 2 αε βh α, he TARCH(,) secificaion is given by s R a a R h ε ε ψ ~ N (,h ) =, 2 2 = αε βh γε d α, and he EGARCH(,) secificaion is given by s R a a R log ε ε ψ ~ N (,h ) =, ( h ) = α z β log( h ) γz α. R is he daily change in log rices of he FTSE Xinhua A5, and R is he daily change in log s rices of he FTSE Xinhua SmallCa index. h is he condiional volailiy, d is a dummy if ε < ε variable where d =, and z = is he sandardized residual. The coefficien o. w. h a is mulilied by 3 for exosiory uroses. Log-likelihood (LL) denoes he logarihm of he likelihood (robabiliy) ha he observed values of he deenden may be rediced from he observed values of he indeendens, and calculaed by using maximum likelihood esimaion (MLE), and AIC, and SC denoe Akaike informaion crierion and Schwarz crierion, resecively. The numbers in aranheses are he -saisics, and ***, **, * denoe he %, 5%, and % significance levels, resecively. All he esimaions are adjused for Bollerslev-Woolridge robus sandard errors and covariance.

Table 4 Average ercenage sreads, average monh-end rice, average monhly variance, and average monhly rading volume (millions of shares) in re-fuures and he os-fuures eriods. FTSE Xinhua A5 (N=28) Mean Percenage Sreads - Before fuures.8 - Afer fuures.62 - -Saisic -2,3 ** Average monh end closing rice - Before fuures 29.77 - Afer fuures 96.49 - -saisic 8,22 *** Average monhly variance - Before fuures.6 - Afer fuures.2 - -saisic 2,64 ** Average monhly volume - Before fuures 5.93 - Afer fuures 38.77 - -saisic 2,34 ** Noe: This able reors he average ercenage sreads, average monh-end rice, average monhly variance, and average monhly rading volume (millions of shares raded) in he re-fuures and he osfuures eriods. The samle consiss of N=28 socks ha have been included in he index hroughou he samle eriod sudied. The -saisic ess for he equaliy of means re- and os-fuures, and ***, **, * denoe he %, 5%, and % significance levels, resecively.

Table 5. Esimaes of he ooled cross-secional ime-series regression Variable Coefficien -saisic Inerce -3.96-2.63 ** LNPRC -.89 -.99 *** LNVOL -.32-4.2 *** LNVAR -.4-2. ** DUMMY.82 3.56 *** Noe: The model esimaed is reresened as: LNSPRD a a LNPRC a LNVOL i = i =, K, N and =, 2. i 2 i a LNVAR 3 i bdf ε, where LNSPRDi is he naural logarihm of he average quoed ercenage sread (he quoed sread as he ercenage of he rice level), and LNPRC, LNVOL, and LNVAR are he naural logarihms of he average monh-end rices, average monhly rading volume, and he monhly reurn variance, resecively. The number of socks included in he regression is denoed by N, and = or 2 denoes he re- and os-fuures eriods. The dummy variable DF akes on he value re-fuures, and os-fuures. ***, **, * denoe he %, 5%, and % significance levels, resecively. All -values are correced for auocorrelaion (wih lag = 3) and heeroskedasiciy as suggesed by Newey and Wes (987). i i i i i

Figure The CUSUM lo for he log reurns of he FTSE Xinhua A5 index Noe: This char los he cumulaive sums for he log reurns of he FTSE Xinhua A5 index for he i i i, =,2, K, = where, eriod 8//25 o /9/27, given by he equaion S S ( X X ) i n X, X 2,, reresen consecuive observaions of he reurns, X is he mean reurn, S =, and n = 538. The red line indicaes he inroducion dae of he FTSE Xinhua A5 index fuures.