Price Discovery in Turkish Index Markets: Empirical Evidence from ISE-30 Index

Transcription

1 Inernaional Research Journal of Finance and Economics ISSN Issue 57 (200) EuroJournals Publishing, Inc. 200 hp:// Price Discovery in Turkish Index Markes: Empirical Evidence from ISE-30 Index Musafa Mesu Kayali Associae Professor of Finance, Dumlupinar Universiy Turkey Sibel Celik Research Assisan (Ph.D. Candidae), Dumlupinar Universiy Turkey Absrac This paper invesigaes he price discovery funcion in Turkish index markes. More specifically i examines he price discovery in ISE-30 index markes afer he inroducion and rading of IST30 Exchange Traded Fund (ETF) on April ISE-30 index is an index consising of large capializaion firms and IST30 ETF is a derivaive securiy designed o rack he performance of ha index. Also IST30 is a financial insrumen rading like shares of a sock on he Isanbul Sock Exchange (ISE). Our findings suppor ha he ISE-30 index markes are no weak form efficien and here is arbirage opporuniy for invesors. Moreover resuls confirm he leader role of fuures markes in price discovery process in Turkey. Keywords: Price Discovery ISE-30 index Exchange Traded Funds IST30. JEL Classificaion Codes: G2 G4 G5 I. Inroducion Since he launch of Sandard and Poor s Deposiary Receips (SPDRs) he granddaddy of all Exchange Traded Funds (ETFs) in 993 ETFs have become one of he world s fases growing invesmens. SPDRs have he highes marke value among he more han 939 ETFs ha exis worldwide oday. ETFs can be raded like he componen socks in he index all day long and are subjec o similar risks. Alhough he law of one price implies ha here should be no difference beween he prices of he same asse rading in differen markes here may be discrepancies beween he prices in respecive markes due o marke imperfecions such as price discreeness rounding effecs invenory consideraions and informaion asymmeries. However markes linked by arbirage should be able o eliminae hose discrepancies beween prices hrough buying relaively cheaper asses and selling he expensive ones. The relaively lower ransacion coss and managemen fees for ETFs make i possible o implemen such arbirage sraegies and exploi hose opporuniies as hey arise. Once a new securiy is inroduced and sars rading one of he issues ha is of grea imporance in financial markes is fragmenaion. I occurs when he order flow from he exising marke moves owards he new marke. This migh affec he liquidiy and price discovery process of he exising marke adversely. However recen sudies in marke microsrucure lieraure provide empirical evidence ha alernaive markes no only help improve he liquidiy of exising markes bu also conribue o he price discovery.

2 Inernaional Research Journal of Finance and Economics - Issue 57 (200) 227 The research on ETFs is comparaively limied due o relaively shor hisory of hose innovaive invesmen vehicles wih recen sudies including he following: Swizer e al. (2000) Chu and Hsieh (2002) Kurov and Lasser (2002) and Chu and Kayali (2006) who explore he impac of ETFs on he pricing efficiency and marke qualiy of index fuures; Acker and Tian (2000) Elon e al. (2002) Kayali (2007a) and Kayali (2007b) who invesigae he pricing efficiency of ETFs; and Chu e al. (999) and Hasbrouck (2003) who examine price discovery among index spo index fuures and ETFs. Similarly Sevil e al. (2008) Kasman and Kasman (2008) Basdas (2009) and Ozen e al. (2009) look ino he relaionship beween index spo and index fuures prices in he ISE-30 index markes. The inroducion and rading of IST30 ETF in Turkey provides us unique opporuniy o revisi his relaionship. We exend previous research on price discovery in Turkish index markes by including he prices of IST30 ETF in he analysis in addiion o he prices of index spo and index fuures. The res of he aricle proceeds as follows. Firs we review he relevan lieraure. Second we describe our daa and mehodology. The empirical resuls are discussed nex. Finally we summarize our findings and offer conclusions. II. Lieraure There have been many sudies on he relaionship among spo prices fuures prices and exchange raded funds (ETF) in he lieraure. Alhough here are sudies on he relaionship beween spo prices-etfs and fuures prices-etfs empirical sudies mosly focused on he relaionship beween spo prices and fuures prices. Researchers es he lead-lag relaionship beween index spo prices and fuures prices for differen markes [Sock Markes: (Kawaller e al. 987; Hasbrouck 2003); Bond Markes: (Scalia998); Foreign Exchange Markes: (Tse e al. 2006) Commodiy Markes: (Hudson e al. 996)]. While mos of hese sudies find evidence on index fuures prices leading spo prices (Garbade and Silber 983; Kawaller e al.987; Herbs e al. 987; Chan 992; Anoniou and Garre 993; Tse 995; Cheung and Fung 997; Nieo e al. 998; Scalia 998; Jong and Donders 998; Min and Najand 999; Gwilym and Buckle 200 Ryoo and Smih 2004; Kenourgios 2004; Floros and Vougas 2007; Theissen 2009; Cheung and Fung 997) some of hem find evidence on index spo prices leading fuures prices (Chan 992; Anoniou and Garre 993; Jong and Nijman 997; Cheung and Fung997; Ryoo and Smih 2004; Kenourgios 2004; Wang e al. 2009). Addiionally here are sudies finding no evidence on lead-lag relaionship beween index spo and fuures prices (Brooks e al. 200; Hasan 2005; In and Kim 2006). When we evaluae he sudies above alhough sudies on he lead-lag relaionship beween index spo prices and fuures prices suppor he leadership of fuures markes in price discovery process he resuls are conroversial in he lieraure. In addiion researchers analyze he lead lag relaionship beween spo prices-etfs (Wang e al. 2009) and fuures prices-etfs (Chou and Chung 2006; Schlusche 2009) in he lieraure. For example Wang e al. (2009) invesigae he impac of spo index on he exchange raded funds indices in Taiwan by using VAR model and find evidence on spo index leading exchange raded funds indices. Chu e al. (999) invesigae he price discovery roles of he S&P 500 spo index index fuures and he SPDRs. They propose four hypoheses namely he leverage hypohesis rading cos hypohesis upick rule hypohesis and markewide informaion hypohesis o disinguish he relaive price discovery roles of he hree price series. They employ mached synchronous inraday rading daa and find ha he hree price series are a coinegraed sysem wih one common long-run sochasic rend. As a resul of Gonzalo and Granger s (995) common-facor model hey conclude ha alhough S&P 500 index fuures play a dominan role in he price discovery process SPDRs conribue more o he common facor han he spo marke. They aribue his finding o he leverage effec and upick rule. Hasbrouck (2003) empirically invesigaes price discovery in US equiy index markes. He repors ha mos of he price discovery occurs in he E-mini marke where index fuures conracs are raded elecronically for he S&P 500 and Nasdaq-00 indexes. Over he sample period he covers in his

3 228 Inernaional Research Journal of Finance and Economics - Issue 57 (200) sudy Hasbrouck shows ha ETFs play a smaller role in he price discovery process. In he S&P 400 index marke however no E-mini fuures conracs are being raded during he period sudied and boh regular fuures conracs and ETFs conribue o he price discovery. As for he price discovery beween he S&P 500 ETF and he secor ETFs he former dominaes he laer. So and Tse (2004) examine he price discovery in Hong Kong index markes by using he Hang Seng spo index index fuures and he Tracker Fund (ETF) prices. They find ha index fuures lead spo index in he price discovery process. Also he Tracker Fund does no conribue o he price discovery and has no impac on he processing of informaion in he Hang Seng index markes. Chou and Chung (2006) examine he impac of reducion in rading coss due o decimalizaion on informed rading and speed of informaion ransmission beween ETFs and heir corresponding index fuures for he S&P 500 Nasdaq-00 and Dow Jones 30 index markes. They analyze wheher he decrease in he minimum ick size of ETFs affecs he relaive performances of boh securiies in he price discovery process. They conclude ha alhough index fuures have a dominan role in informaion discovery ETF prices sar o lead index fuures afer decimalizaion. This resul is consisen wih he findings of Beaulieu e al. (2003) who repor similar resuls for he Torono Sock Exchange 35 index marke. Chen and Sroher (2008) analyze he conribuion of ETFs o price discovery in he presence of price limis wihou shor selling on he Shanghai Sock Exchange (SSE). They empirically show ha price discovery moves from he SSE 50 index o he SSE 50 ETF when a significan porion of he index componen socks reach daily price limis. Their resuls indicae ha ETFs play an imporan role in he price discovery process of he underlying componen socks in he presence of regulaoryimposed pricing consrains. Schlusche (2009) invesigaes he relaive conribuion o price discovery of he ETF marke and he fuures marke in Germany wo derivaive producs of he German blue chip index DAX. He employs a Vecor Error Correcion Model (VECM) of DaxEx (ETF) prices and DAX fuures prices adjusing for he cos-of-carry. He shows ha fuures prices lead ETF prices in Germany. However when volailiy is high he conribuion of he ETF marke o price discovery process increases. In Turkey sudies on he price discovery process appear in he lieraure recenly. For example Sevil e al. (2008) invesigae he lead lag relaionship beween ISE-30 index spo and fuures markes for he period beween February 2005 and Sepember 2007 in Turkey by using Granger Causaliy mehod. They find ha spo prices end o lead fuures prices in Turkey. This resul conradics wih he findings of mos of he previous sudies focusing on differen index markes all over he world. Kasman and Kasman (2008) invesigae he impac of he inroducion of ISE-30 index fuures on he volailiy of ISE-30 index employing EGARCH model for he period July 2002-Ocober They find ha he inroducion of index fuures diminishes he volailiy of he underlying index. They also show ha here is a long run relaionship beween spo and fuures prices and ha he direcion of boh long and shor run causaliy is from spo prices o fuures prices. This finding is consisen wih ha of Sevil e al. (2008). Basdas (2009) revisi he lead-lag relaionship beween ISE-30 index spo and fuures prices using daily observaions from February 2005 hrough May The auhor repors ha spo prices lead fuures prices in he ISE-30 index markes. This resul is consisen wih he findings of Kasman and Kasman (2008). Also he auhor empirically shows ha he forecasing performance of he Error Correcion Model (ECM) is superior in comparison wih he performances of ECM wih Cos of Carry (COC) Auoregressive Inegraed Moving Average (ARIMA) and Vecor Auoregressive (VAR) models. Ozen e al. (2009) analyze he shor erm and long erm causaliy beween spo and fuures markes for he period beween February 2005 and February 2009 by using vecor error correcion mehod (VECM). They conclude ha fuures marke leads spo marke in he long erm alhough spo marke is he cause of fuures marke in he shor erm. This resul conradics wih he findings of

4 Inernaional Research Journal of Finance and Economics - Issue 57 (200) 229 Kasman and Kasman (2008) who repor boh long and shor run causaliy from spo prices o fuures prices in he ISE-30 index markes in Turkey. III. Daa and Mehodology In his paper we used he closing prices of IST-30 ETF, ISE-30 index raded in Isanbul Sock Exchange (henceforh SISE-30) and ISE 30 index raded in Turkish Derivaives Exchange (henceforh FISE-30) for he period beween April 7 h 2009 and July 6 h 200. We obained IST-30 ETF daa from is own web sie SISE-30 index daa from Isanbul Securiies Exchange web sie and FISE-30 index from Turkish Derivaives Exchange web sie. In his sudy we analyzed he bivariae relaionship among he variables of IST-30 ETF SISE 30 index and FISE 30 index. For his purpose firsly we analyzed he saionary of all variables by using Augmened Dickey Fuller uni roo es (Dickey and Fuller 98). Since non saionary series cause spurious regression resuls i is imporan o decide wheher he series are I(0) or I() before any empirical sudy. In he case which boh series are I() i would be expeced o hold long run relaionship wih one anoher and i should be searched he coinegraion relaionship beween he variables. In he conrary case we should esimae he model aking firs differences of series. Afer analyzing he saionary of variables we es he long run relaionship beween variables. In he lieraure hree differen mehods are used o es coinegraion. These mehods are he Engle Granger (987) 2 sep mehod Engle Yoo (987) 3 sep mehod and lasly Johansen procedure (990) based on VAR. Boh Engle Granger 2 sep mehod and Engle Yoo 3 sep mehod have some shorcomings like finie sample problems simulaneous equaion bias if he causaliy occurs in boh direcions and no o be able o hypohesis esing. However Johansen procedure overcomes hese problems so in his paper we used he Johansen procedure o es he coinegraion beween variables. Johansen procedure is based on VAR approach; y = β y + β 2 y β k y k + ζ () In Equaion () y is vecor of variables I() and ζ is vecor of residuals. Equaion () can be wrien as in Equaion (2); y = y k + Γ y + Γ2. y Γk y ( k ) + ζ (2) where =( ( k j= β ) I and Γ = ( β ) I i i i j= j The Johansen es focuses on examinaion of he marix. can be inerpreed as a long run coefficien marix since in equilibrium all he y i will be zero and seing he error erms ζ o heir expeced value of zero will leave y k = 0. The coinegraion beween variables is calculaed by looking a he rank of he marix via is eigenvalue. There are wo es saisics race saisic and max eigenvalue saisic for coinegraion under he Johansen approach and described as; g λ ( r) = T ln( ˆ λ ) and λ r, r ) = T ln( ˆ λ ) (3) race i= r+ i max ( + r+ where r is he number of coinegraing vecors under he null hypohesis and λˆ i is he esimaed value for he ih ordered eigenvalue from he marix (Brooks 2002). Thirdly we apply he Vecor Error Correcion Model (VECM) o undersand wheher he variables have long run equilibrium relaionship or no. A simple VECM model can be described as in Equaion 4 z = α y + α 2 ( z θy ) + ζ (4)

5 230 Inernaional Research Journal of Finance and Economics - Issue 57 (200) where z θ y is error correcion erm. In equaion 3 θ defines he long run relaionship beween z and y and α i describes he shor run relaionship beween changes in z and changes in y. α 2 defines he speed of adjusmen back o equilibrium (Brooks 2002). Fourhly we presen resuls of variance decomposiion analysis o see he proporion of he movemens in he variables ha are due o heir own shocks versus shocks o he oher variables. Lasly we apply VECM Granger Causaliy es o deermine he direcion of he causaliy beween variables. VECM Granger Causaliy es is defined as in Equaion 5 z y p p α i z i + φi y i + ψ ε i= i= = α + + ν 0 p p 2 + σ i y i + γ i z i + δε ξ (5) i= i= = α + where δ and ψ are coefficien of error correcion erms. The null hypohesis is z does no Granger cause y and y does no Granger cause z. IV. Empirical Findings Firs we presen he descripive saisics of variables in Table. Table : Descripive Saisics Mean Median Maximum Minimum Sd.Dev. Kurosis Skewness IST-30 ETF SISE FISE Then we es he saionariy of variables by using Augmened Dickey-Fuller es and presen he resuls in Table 2. I is seen from Table 2 ha alhough all series are no saionary in price level hey are saionary in reurn level. We can say ha series are I(). Thus here may be a coinegraion relaionship beween hese pair of variables. Table 2: The Resuls of Augmened Dickey -Fuller Uni Roo Tess In price Tes Saisic Model Lag IST 30 ETF # 0 SISE # 0 FISE # 0 In Reurn Tes Saisic Model Lag IST 30 ETF *** # 0 SISE *** # 0 FISE *** # 0 Noes: For % and %5 significance level Mac Kinnon criical values are -345 and -287 (wih inercep) respecively. *** represens he rejecion of null hypohesis of series have uni roo a % significance level. # represens wih inercep model. We used Schwarz Informaion Crierion for lag order selecion. L refers he log operaor. Since all variables are I() we invesigae coinegraion beween hem. Firs we should deermine he opimum lag order of variables. Table 3 shows he seleced lags by Akaike Informaion Crierion and Schwarz Informaion Crierion. We use Schwarz Crierion as a base and used lag for all series.

6 Inernaional Research Journal of Finance and Economics - Issue 57 (200) 23 Table 3: Lag Order Selecion of Variables Lag AIC SC IST 30 ETF- SISE * * IST 30 ETF- FISE * * SISE-30 - FISE * * Noes: AIC represens Akaike Informaion Crierion SC represens Schwarz Informaion Crierion * represens lag order selecion Then we employ Johansen approach o undersand he coinegraion relaionship beween he variables. Table 4 shows he resuls of Johansen approach. Table 4 represens wo differen es saisics race saisic and max eigenvalue saisic. Table 4: The Resuls of Coinegraion Tess Trace Max Eigenvalue H0 Eigenvalue Trace 0.05 criical Prob Max-eigen 0.05 criical Prob Saisic value saisic value IST 30 ETF- SISE-30 r= * ** r<= * * IST30 ETF- FISE-30 r= * * r<= * * SISE-30 - FISE-30 r= * * r<= * * * and **represens he rejecion of null hypohesis a %5 %0 significance level respecively. p Mac Kinnon- haug- Michelis (999) p value Trend assumpion is linear deerminisic rend. We used Schwarz Informaion Crierion in Table 2 o deermine opimal lag In Table 4 firs second and hird panel show he coinegraion relaionship beween IST30- ETF and SISE-30 IST30- ETF and FISE-30 SISE-30 and FISE-30 respecively. Since race saisic and max eigenvalue saisic are greaer han %5 criical values we rejeced he null hypoheses of here is no coinegraing equaion and here is a mos coinegraing equaion beween he variables. There are wo coinegraing equaion beween all pair of variables. Thus i is possible o say ha here are long run relaionships beween IST30- ETF and SISE-30 IST30- ETF and FISE-30 SISE-30 and FISE- 30. Table 5: The Resuls of Vecor Error Correcion Model IST30 ETF- SISE-30 LIST30 L SISE-30 EC [-.7986] [0.466] LIST [ ] [ ] L SISE [ ] [ ] IST 30 ETF- FISE-30 LIST30 L FISE-30

7 232 Inernaional Research Journal of Finance and Economics - Issue 57 (200) EC *** [ ] [0.254] LIST [0.779] [-0.542] L FISE *** [ ] [-.6623] SISE-30 - FISE-30 L SISE-30 L FISE-30 EC *** [ ] [0.879] L SISE [0.7487] [-0.327] L FISE *** [ ] [-624] Noes: [.] shows values *** is % significance level is difference operaor L is log operaor. EC is error correcion erm. Since he variables are coinegraed we use VECM for modelling he relaionship beween hem. Table 5 exhibis he resuls of VECM. In he firs panel we can explain he long run and shor run relaionships beween IST-30 ETF and SISE-30 index. The coefficien of Error Correcion erm (EC) is negaive as expeced however i is no significan. This means ha deviaions in shor run are no adjused by variables and here is no long run equilibrium beween IST-30 ETF and SISE-30 index. Indisinguishably coefficiens of shor-run relaionship are no significan. This means we can no expec he shor run and causaliy relaions beween wo variables. The second panel shows he relaionship beween IST-30 ETF and FISE-30 Index. The error correcion erm is negaive and significan as expeced in he lieraure indicaing ha if he difference beween IST-30 ETF and FISE-30 index is posiive in one period he spo price fall during he nex period o resore equilibrium and vice versa. When we examine he shor run relaionship he coefficien of FISE-30 index is significan indicaing ha he fuures marke leads he is-30 ETF. The hird panel shows he relaionship beween SISE-30 index and FISE-30 index. For he long-run relaionship coefficien of error correcion erm is negaive and significan indicaing ha if he difference beween SISE-30 and FISE-30 index is posiive in one period he spo price fall during he nex period o resore equilibrium and vice versa. In addiion he coefficien of shor erm relaionship parameer is significan indicaing fuures marke (FISE-30 index) leads he spo marke (SISE-30 index). Afer he VECM we presen he variance decomposiion analysis in able.6. In he firs panel he variance decomposiion of IST-30 ETF is explained approximaely %99 by is own pas shocks however he variance decomposiion of SISE-30 index is explained approximaely %99 by Is-30 ETF. Since he IST-30 ETF is radable securiy his resul is expeced. Table 6: The Resuls of Variance Decomposiion Analysis IST30 ETF SISE-30 Period Sd. error IST 30 ETF SISE-30 Sd. error IST 30 ETF SISE IST 30 ETF FISE-30 Period Sd. error IST 30 ETF FISE-30 Sd. error IST 30 ETF FISE-30

8 Inernaional Research Journal of Finance and Economics - Issue 57 (200) SISE-30 FISE-30 Period Sd. error SISE-30 FISE-30 Sd. error SISE-30 FISE In he second panel while he variance decomposiion of IST-30 ETF is explained approximaely %95 by is own pas shocks he variance decomposiion of FISE-30 index is explained approximaely %94 by IST-30 ETF. In he hird panel while he variance decomposiion of SISE-30 index is explained approximaely %96 by is own pas shocks he variance decomposiion of FISE-30 index is explained approximaely %96 by FISE-30 index. Afer he deerminaion of relaionships beween variables we have applied VECM Granger Causaliy es o see he direcion of relaionship. Table 7 shows he resuls of VECM Granger Causaliy es. Table 7: Resuls of VECM Granger Causaliy Tes Null Hypohesis Wald (χ2) Prob IST 30 ETF does no Granger cause SISE SISE-30 does no Granger cause IST 30 ETF IST 30 ETF does no Granger cause FISE FISE-30 does no Granger cause IST 30 ETF *** SISE-30 does no Granger cause FISE FISE-30 does no Granger cause SISE *** Noes: *** indicaes he % significance level In Table 7 we can no rejec he null hypoheses ha FISE-30 does no Granger cause IST 30 ETF and FISE-30 does no Granger cause SISE-30 a % significance level. The resuls of VECM Granger Causaliy ess are compaible wih he resuls of VECM analysis. V. Conclusion In his sudy we invesigae he price discovery funcion in Turkish index markes by employing coinegraion and VECM analysis. Firs we es he saionariy of variables second we analyze long run relaionship beween variables. Third we use VECM analysis for modelling he relaionship beween hem. Las we es he direcion of causaliy beween variables wih VECM Granger Causaliy es. As a resul we observe long run relaionship beween variables. VECM es shows ha deviaions

9 234 Inernaional Research Journal of Finance and Economics - Issue 57 (200) in shor run are adjused by variables for relaionship beween IST30- FISE30 and SISE30- FISE30. However his is no he case for relaionship beween IST-30-SISE-30 index. In addiion VECM Granger Causaliy es shows ha here is causaliy relaionship running from FISE-30 o IST-30 and FISE-30 o SISE-30. The presence of coinegraion beween variables suggess he violaion of weak form efficiency and arbirage opporuniy in Turkish index markes. In addiion since IST-30 SISE-30 and FISE-30 index are no weak form efficien invesors can predic he fuure prices by using he pas informaion. Lasly our findings suppor he leader role of fuures markes in price discovery process in Turkey IST-30 ETF - SISE-30 (price) IST-30 ETF - SISE-30 (reurn) M M0 200M0 200M04 200M M M0 200M0 200M04 200M07 IST SISE DIST DSISE IST-30 ETF - FISE-30 (price) IST-30 ETF - FISE-30 (reurn) M M0 200M0 200M04 200M M M0 200M0 200M04 200M07 IST FISE DIST DFISE SISE-30 - FISE-30 (price) SISE-30 - FISE-30 (reurn) M M0 200M0 200M04 200M M M0 200M0 200M04 200M07 SISE FISE DSISE DFISE

10 Inernaional Research Journal of Finance and Economics - Issue 57 (200) 235 References [] Acker L. Tian Y. (2000). Arbirage and valuaion in he marke for Sandard and Poor s Deposiary Receips. Financial Managemen [2] Anoniou A. and Garre I. (993) To wha exen did sock index fuures conribue o he Ocober 987 sock marke crash? Economic Journal [3] Basdas U. Lead-Lag Relaionship beween he Spo Index and Fuures Price for he Turkish Derivaives Exchange (Ocober ). Available a SSRN: hp://ssrn.com/absrac =49347 [4] Beaulieu M.C. Ebrahim S.K. & Morgan I.G. (2003). Does ick size influence price discovery? Evidence from he Torono Sock Exchange. Journal of Fuures Markes [5] Brooks C. Rew A.G. and Rison S. (200). A Trading Sraegy based on he lead-lag relaionship beween he Spo Index and Fuures Conrac for he FTSE 00 Inernaional Journal of Forecasing [6] Brooks C. (2002). Inroducory Economerics for Finance Cambridge Universiy Press. [7] Chan K (992) A Furher Analysis of he Lead-Lag Relaionship Beween he Cash Marke and Sock Index Fuures Marke Review of Financial Sudies Vol. 5 No. pp [8] Chen G. and Sroher T.S On he Conribuion of Index Exchange Traded Funds o Price Discovery in he Presence of Price Limis Wihou Shor Selling (February 2008). Available a SSRN: hp://ssrn.com/absrac= [9] Cheung Y.W. and Fung H.G.(997). Informaion flows beween eurodollar spo and fuures markes Mulinaional Finance Journal(4) [0] Chou R. K. and Chung H. (2006) Decimalizaion rading coss and informaion ransmission beween ETFs and index fuures Journal of Fuures Markes [] Chu Q.C. Hsieh W.G. Tse Y Price discovery on he S&P 500 index markes: an analysis of spo index index fuures and SPDRS. Inernaional Review of Financial Analysis [2] Chu Q.C. Hsieh W.G Pricing efficiency of he S&P 500 Index Marke: evidence from he Sandards and Poor s Deposiary Receips. Journal of Fuures Markes [3] Chu Q.C. and M.M. Kayali 2006 Sandard and Poor s Deposiary Receips and he marke qualiy of S&P 500 index fuures Applied Economerics and Inernaional Developmen 6: [4] Dickey D.A. and Fuller W.A. (98). Likelihood Raio Saisics for Auoregressive Time Series wih a Uni Roo Economerica 49(4) [5] Elon E.J. M.J. Gruber G. Comer and K. Li 2002 Spiders: Where are he Bugs Journal of Business 75 (3) [6] Engle R.F. and Granger C.W.E. (987). Coinegraion and Error Correcion: Represenaion Esimaion and Tesing Economerica [7] Engle R.F. and Yoo B.S.(987). Forecasing and Tesing in Co-inegraed Sysems Journal of Economerics 35() [8] Floros C. and Vougas D.V. (2007). Lead-Lag Relaionship beween Fuures and Spo Markes in Greece: Inernaional Research Journal of Finance and Economics Issue: 7 pp [9] Garbade K.D. and Silber W.L.(983). Price Movemens and Price Discovery in Fuures and Cash Markes The Review of Economics and Saisics Vol.65 No: 2 pp [20] Gwilym O.A. and Buckle M The lead-lag relaionship beween he FTSE00 sock index and is derivaive conracs Applied Financial Economics (4) [2] Hasan M. (2005) An alernaive approach in invesigaing lead-lag relaionships beween sock and sock index fuures markes - a commen Applied Financial Economics Leers (2) [22] Hasbrouck J Inraday Price Formaion in U.S. Equiy Index Markes. The Journal of Finance 58 (6)

11 236 Inernaional Research Journal of Finance and Economics - Issue 57 (200) [23] Herbs A. McCormack J.P. and Wes E.N. (987) Invesigaion of a Lead-Lag Relaionship beween Spo Sock Indices and Their Fuures Conracs Journal of Fuures Markes [24] Hudson D. Ehridge D. and Brown J. (996). Producer prices in coon markes: Evaluaion of repored price informaion accuracy AgriBusiness 2(4) [25] In F. and Kim S. (2006) The Hedge Raio and he Empirical Relaionship beween he Sock and Fuures Markes: A New Approach Using Wavele Analysis The Journal of Business Vol. 79 No [26] Johansen S. and K. Juselius (990) Maximum Likelihood Esimaion and Inference on Coinegraion Wih Applicaions o he Demand for Money Oxford Bullein of Economics and Saisics [27] Jong F.D. and Donders M.W.M. (998). Inraday Lead-Lag Relaionships Beween he Fuures-Opions and Sock Marke European Finance Review : [28] Jong F. and Nijman T. (997). High Frequency Analysis of lead lag Relaionships Beween Financial Markes Journal of Empirical Finance 4(2-3) [29] Kasman A. and Kasman S. (2008) The Impac of Fuures Trading on Volailiy of he Underlying Asse in he Turkish Sock Marke Physica A [30] Kayali M.M. 2007a Pricing efficiency of exchange raded funds in Turkey: Early evidence from Dow Jones Isanbul 20 Inernaional Research Journal of Finance and Economics [3] Kayali M.M. 2007b Do Turkish Spiders Confuse Bulls and Bears?: The Case of Dow Jones Isanbul 20 Invesmen Managemen and Financial Innovaions 4: [32] Kawaller I.G. P.D. Koch and T.W. Koch (987) The Temporal Price Relaionship Beween S&P500 Fuures and he S&P 500 Index Journal of Finance Vol. 42 pp [33] Kenourgios D. Price discovery in he Ahens derivaives exchange: evidence for he FTSE/ASE-20 fuures marke Economic and Business Review Vol pp [34] Kurov A. and D. Lasser 2002 The effec of he inroducion of Cubes on he NASDAQ-00 index spo-fuures pricing relaionship Journal of Fuures Markes 22(3) [35] Min J.H. and Najand M. (999). A Furher Invesigaion of he Lead-Lag Relaionship beween he Spo Marke and Sock Index Fuures: Early Evidence from Korea The Journal of Fuures Markes Vol: 9 No:2 pp [36] Nieo M.L. Fernandez A. and Muñoz M.J. (998). Marke efficiency in he Spanish derivaives markes: An empirical analysis Inernaional Advances in Economic Research 4(4) [37] Ozen E. Bozdogan T. and Zugul M.(2009). The Relaionship of Causaliy beween he Price of Fuures Transacions Underlying Sock Exchange and Price of Cash Marke: The Case of Turkey Middle Easern Finance and Economics (4) [38] Ryoo H and Smih G. (2004) The Impac of Sock Index Fuures on he Korean Sock Marke Applied Financial Economics Vol. 4 No. 4 pp [39] Scalia A. (998). Informaion ransmission and causaliy in he Ialian Treasury bond marke Journal of Empirical Finance 5(4) [40] Schlusche B. (2009) Price Formaion in Spo and Fuures Markes: Exchange Traded Funds vs. Index Fuures Journal of Derivaives 7: [4] Sevil G. Sayılır O. and Yalama A. Lead-Lag Relaionships beween Spo and Fuure Sock Markes in Turkey Academy of World Business Markeing & Managemen Developmen (July ) Brazil 3 () [42] So R.W. and Y. Tse (2004). Price discovery in he Hang Seng index markes: Index fuures and he racker fund. The Journal of Fuures Markes [43] Swizer L. P. Varson and S. Zghidi 2000 Sandard and Poor s Deposiory Receips and he performance of he S&P 500 Index fuures marke Journal of Fuures Markes 20:

12 Inernaional Research Journal of Finance and Economics - Issue 57 (200) 237 [44] Theissen E. (2009). Price discovery in spo and fuures markes: A reconsideraion Unpublished Manuscrip. Universiy of Bonn. [45] Tse Y.K. (995) Lead-Lag Relaionship Beween Spo Index and Fuures Price of Nikkei Sock Average Journal of Forecasing Vol. 4 No. 7 pp [46] Tse Y. Xiang J. and Fung J.K.W. (2006). Price Discovery in he Foreign Exchange Fuures Marke Journal of Fuures Markes 26() 3-43 [47] Wang C.C. Liau Y.S. and Yang J.J.W. (2009). Informaion Spillovers in he Spo and ETF Indices in Taiwan Global Journal of Business Research 3() 7-3