International Research Journal of Finance and Economics ISSN 1450-2887 Issue 58 (2010) EuroJournals Publishing, Inc. 2010 http://www.eurojournals.com/finance.htm

Volatility Linkages among India, Hong Kong and Singapore Stock Markets

Nikolaos Sariannidis Corresponding Author, Lecturer- Technological Education Institute of West Macedonia

Kila, 50100 Kozani, Greece E-mail: sarn@hol.gr Tel: +302461024694

George Konteos Lecturer- Technological Education Institute of West Macedonia

67 Makedonomachon Str., 50200 Greece E-mail: gkonteos@hol.gr

Tel: +302463022859

Evangelos Drimbetas Assistant Professor-Democritus University of Thrace

P.O.Box 418 Kardia, 57500 Thessaloniki, Greece E-mail: e.drimpetas@solidus.gr

Tel: +302310360701

Abstract

This paper analyzes the volatility linkages among three Asian stock exchange markets, namely India, Singapore and Hong Kong, during the period July 1997 to October 2005. We use a multivariate GARCH model to identify the source and magnitude of spillovers. The empirical analysis showed that the markets exhibit a strong GARCH effect and are highly integrated reacting to information which influence not only the mean returns but their volatility as well.

Keywords: Volatility, Multivariate GARCH, Asian stock markets JEL Classifiaction Codes: G14, G15, C22

1. Introduction The transmission of volatility between markets has been studied extensively in recent years. Globalization has brought about market integration, especially in stock markets, a fact which attracted the researchers interest about the transmission of volatility among markets. A common conclusion of contemporary theoretical and empirical approaches in the literature is that the first and second moments are positively related. Most researchers (Agmon, 1972; Eun and Shim, 1989; Koch and Koch, 1991; Knif and Pynnonen, 1999; Tay and Zhu, 2000; Dekker, Sen and Young, 2001; Sariannidis, Koskosas, Garefalakis and Antoniadis; 2009) concluded that, as markets become integrated, the U.S. market or other large markets, which play a major role in their broader area, exert strong influence on returns and/or volatilities of smaller markets.

International Research Journal of Finance and Economics - Issue 58 (2010) 142 However, in recent years, the investigation of the interaction between volatilities of advanced

stock markets using multivariate GARCH models becomes more and more interesting (Karoly, 1995; Theodossiou and Lee, 1995; Koutmos, 1996; Kanas, 1998), since news is produced in all markets, whether large or small. Under this logic, some researchers (Christofi and Pericli, 1999; Kasch-Haroutounian and Price, 2001; Fernandez-Lzquierdo and Lafuente, 2004; Worthigton and Higgs, 2004) proceeded in testing this hypothesis using multivariate GARCH models.

In recent years the markets of South and Southeast Asia attracted interest due to their dynamism and the relatively large rate of growth of the respective countries. Specifically, In, Kim, Yoon and Viney (2001), investigated the interdependence, the transmission mechanism and the integration of three Asian markets, Hong Kong, South Korea and Thailand, using a multivariate VAR-EGARCH model. Their results indicated that South Korea exerts a weak influence in the other markets, playing a lesser role as information producer, whereas Hong Kong is a major information producer, emitting volatility in other Asian markets. In the same direction, De Santis and Imrohoroglu (1997), using a multivariate GARCH model, proved that some emerging markets exhibit reduced volatility as capital mobility increases. This has been attributed to two reasons: firstly, to the increase in the number of stocks included tin the national indexes which means increased diversification and secondly to the increased participation of foreign investors after capital movements liberalization, which means increased depth, information and efficiency of the market. Worthington and Higgs (2004), examined the transmission of equity returns and volatility among three developed markets (Hong Kong, Japan and Singapore) and six emerging markets (Indonesia, Korea, Malaysia, the Philippines, Taiwan and Thailand). The results of the multivariate GARCH model that they have used showed not homogeneous mean spillovers from the developed to the emerging markets and higher own volatility spillover than cross volatility spillovers, especially for the emerging markets. Also, Gunasinghe (2005) examined the integration behavior and volatility spillover transmission across the stock markets of Sri Lanka, India and Pakistan based on a multivariate cointegration test and generalized impulse response functions. He concluded that the Indian stock market may have some marginal spillover effect on Sri Lanka and Pakistan stock price movements. Finally, Li (2007) has examined the linkages among the stock markets of Shanghai Shenzhen, Hong Kong and United States. The approach that he has used is the multivariate GARCH, and more specifically the BEKK model, proposed by Engle and Kroner (1995). The results suggested that the magnitude of the volatility linkages is small, indicating a weak integration of the emerging stock exchanges in mainland China with this of Hong Kong.

Even though there are a large number of studies on the volatility transmission among Asian markets, the relationship of the volatilities of the developed markets of Hong Kong and Singapore and the rapidly developed Indian market, which plays a dominant role as an investment centre in South Asia, has not been described adequately. The contribution of this paper is that the volatility and shock transmission mechanism among the above markets is examined using a multivariate GARCH model. Specifically, we employ a trivariate GARCH model to simultaneously estimate the mean and conditional variance using daily returns from July 1, 1997 to October 31, 2005. We find significant volatility transmission among the markets under investigation. Our results are important for buildings accurate asset pricing models, forecasting volatility, and will further our understanding of the equity markets. Additionally, since different financial assets are traded based on these market indexes, it is important for financial market participants to understand the volatility transmission mechanism over time and across markets in order to make optimal portfolio allocation decisions.

2. Indices used, Data and Preliminary Results Of the 22 stock exchanges in the country, Mumbai's (earlier known as Bombay), Bombay Stock Exchange (BSE) is the largest, with over 6,000 stocks listed. The BSE accounts for over two thirds of the total trading volume in the country. Established in 1875, the exchange is also the oldest in Asia.

143 International Research Journal of Finance and Economics - Issue 58 (2010) The market capitalization of the BSE is Rs. 5 trillion (some 800bn US$). The BSE Sensex (BSE) is a widely used market index for the Bombay Stock Exchange.

The Hong Kong Stock Exchange (HKSE), established at the end of the 19th century, today, with its total securities market capitalization of a record sum of HK$ 8,260.3bn (US$ 1,063.9 trillion), ranks 8th place by market capitalization in the world. The Hang Seng (HSI), started in 1969, is the leading index for shares traded in the HKSE. The index consists of the 33 largest companies and represents approximately 70% of the value of all stocks traded in the HKSE.

With Singapore now a leading financial center in the Asia-Pacific, the Singapore Exchange (SGX) has become one of the premier exchanges in the region. It is a highly international exchange, with 40% of its market capitalization coming from foreign companies. The Strait Times Index (STI) is the main index of SGX, consisting of 50 constituent stocks listed on the Singapore Exchange Securities Trading Limited (SGX-ST), formerly known as the Stock Exchange of Singapore (SES).

This study uses daily closing stock index data for three major markets of Southeast Asia, for the period July 1, 1997 to October 31, 2005. The data have been obtained from the Comstock database. The three indices analyzed are the BSE (India), the HSI (Hong Kong), and the STI (Singapore). In addition, the daily values of the DJ index have been used to isolate the systematic factors concerning general international factors.

As shown in Table 1, the HSI and the STI indices are highly correlated (0.62). This fact was expected since increased interactions exist between the two countries which are geographically close to each other and they have extensive commercial, economic and political links. Moreover, the correlations of BSE with HSI and STI are 0.207 and 0.243 respectively, a fact attributed to the extroversive features of the Indian economy in recent years. The correlations of DJ, with BSE, HSI and STI are 0.126, 0.303 and 0.334 respectively.

Table 1: Correlations of Indices

DJ (with one lag) BSE STI HSI DJ (with one lag) 1.000 0.126 0.303 0.334 BSE 0.126 1.000 0.207 0.243 SGX-ST 0.303 0.207 1.000 0.620 HIS 0.334 0.243 0.620 1.000

The mean returns of all series are not statistically different from zero. The coefficients of skewness and kurtosis indicate that the series have asymmetric and leptokurtic distribution. The Jarque Bera test rejects the normality of the return series. The augmented Dickey - Fuller (ADF) test, allowing for both an intercept and a time trend, showed that the BSE returns series had been produced by a stationary series. The Ljung Statistics for 1, 3 and 12 lags applied on returns (denoted by LB(n)) and squared returns (denoted by LB2(n)) indicate significant time dependence of the first and second moments. Therefore, a multivariate GARCH model that takes into account the ARCH effects that all series exhibited in the mean is an appropriate model.

Table 2: Sample statistics of BSE, HSI and STI returns

Statistics BSE HSI STI Mean 0.000294 -0.000078 0.000139 Std. Dev. 0.016153 0.017318 0.014684 Skewness -0.32349 -0.204156 0.405903 Kurtosis 6.640 10.513 12.806 Jarque-Bera 1175 4866 8323 Augmented Dickey-Fuller (ADF) -42.3 -24.4 -39.2 Observations 2063 2063 2063 LB(1) 10.553 8.4983 43.449 LB(3) 15.432 22.858 43.792 LB(12) 43.285 34.067 55.785 LB2(1) 183.36 76.686 58.301 LB2(3) 256.73 344.01 146.83 LB2(12) 370.44 581.72 474.8

International Research Journal of Finance and Economics - Issue 58 (2010) 144

3. Empirical Framework and Estimation Results Considering the international literature and the preliminary results cited above, the BEKK model of Baba, Engle, Kraft, and Kroner (1989) renders a very good choice for modeling the volatility transmission among BSE, HSI and STI indexes. The following mean equations were estimated for each markets own returns and the returns of other markets lagged one period:

ttt GRdR ++= 1 (1)

where Rt is an nx1 vector of daily returns at time t for each market, t/It-1~N(0, Ht) is an nx1 vector of random errors for each market at time t, It-1 represents the market information that is available at time t-1 with its corresponding nxn conditional variance-covariance matrix, Ht. The diagonal elements gii of matrix G are the respective markets own returns lagged one period, while the off-diagonal elements gij represent the mean spillover effect across markets. The 3x1 vector di contains constants.

The variance equation that we used in this paper of the following form: Ht = CC + At-1t-1 +t-1 (2)

where C is a 3x3 lower triangular matrix of constants, and A is a 3x3 square matrix which shows how the conditional variances are correlated with past squared errors. The elements of matrix A measure the effects of shocks or news on the conditional variances. The 3x3 square matrix B shows how past conditional variances affect the current levels of conditional variances, in other words, the degree of volatility persistence in conditional volatility among the markets. For the empirical implementation it is desirable to restrict further the above parameterization. Of the two most popular parameterizations for the multivariate GARCH models, VECH and BEKK, we adopted the BEKK parameterization, because this model is designed in such a way that has less parameters and the estimated covariance matrix will be positive semi-definite, which is a requirement needed to guarantee non-negative estimated variances. Thus, we have restricted the trivariate BEKK model with 24 parameters to the reduced form of the diagonal BEKK with 12 parameters. The elements of the covariance matrix Ht, depends only on past values of itself and past values of tt, which means that the variances depend only on past own squared residuals, and the covariances depend only on past own gross products of residuals. The restriction of the diagonal is justified, since information about variances is usually revealed in squared residuals, and, moreover, if the variances develop slowly, then past squared residuals should be able to forecast future variances (the same argument holds for the covariances).

In order to estimate the above BEKK system we maximized the following likelihood function using the Student t-distribution, which accounts for the possible excess kurtosis usually found in the residuals of stock series in ARCH models:

=

++=1

1' )(ln21)2ln(

2)(

ttttt HH

TnL pi (3)

where () is a vector of parameters to be estimated, T is the number of observations and n is the number of markets. The Marquardt optimization algorithm is used to produce the maximum likelihood parameter estimates and their corresponding asymptotic standard errors.

The estimation results of the multivariate GARCH model with diagonal BEKK parameterization for each mean equation are reported in Table 3. The results of the mean equation manifest the huge influence (gi4 = 0.267) that the USA market exerts in the formation of the returns in all other markets. Also, the coefficients gi1- (0.032) and gi3 (0.051) show that the mean return of the Indian and the Singapore markets are influenced by they own lagged returns and exhibit mean return spillovers from lagged returns in the other markets.

145 International Research Journal of Finance and Economics - Issue 58 (2010) Table 3: Estimated coefficients for conditional mean returns equations

Estimated coefficient

Standard error

d1 0.001021* 0.000273 d2 0.000073 0.000244 d3 0.000211 0.000212 gi1 0.032354* 0.012278 gi2 0.012633 0.015607 gi3 0.050972* 0.017619 gi4 0.266829* 0.015897

i takes values from 1 to 4 and refers to the indices BSE, HSI, STI and DJ respectively. * indicates statistical significance at the 1% level. ** indicates statistical significance at the 5% level.

The coefficient gi2 is not statistically significant, possibly because the information for the Hong Kong investors derives mainly from the USA market as well as from other large Asian markets which close after the HKSE (Table 4). This manifests the efficiency of the HKSE, as the prices in this market incorporate all available information and any information available after market close affects next day prices. This is probably the reason that we have autocorrelation in the return series of the HSI.

Table 4: Marketing Opening Times in UTC Time Zone

Exchange Name Opening Time Closing Time Bombay Stock Exchange 4:25 am 10:00 am Hong Kong Stock Exchange 2:00 am 8:00 am Singapore Exchange 1:00 am 9:00 am New York Stock Exchange 2:30 pm 9:00 pm

Table 5 presents the estimated coefficients for the variance covariance matrix of equations. The conditional variance covariance equations effectively capture the volatility and cross volatility spillovers among the three Asian markets. From the empirical results we conclude a time variation in market risk, a strong evidence of GARCH effect and the presence of a weak ARCH effect. Specifically, equations (4)-(9) present the effects of variance equations implied by the diagonal BEKK specification. The volatilities h11, h22 and h33 in the Indian, Hong Kong and Singapore markets

respectively, show that the coefficient of GARCH effect, which reflects the influence of2

1th , i.e. the older information (residuals ut-2, ut-3, ), is much higher than the value of the ARCH coefficient, which correlates the price variation of the present day to the price variation of the previous day. The results for the covariance equations are similar, indicating that there is a statistically significant covariation in shocks, which depends more on its lag than on past innovations. Consequently, market shocks are influenced by information which is common to the respective markets, and as a result of this we have statistically significant covariance in the variance covariance equations. The conditional variances covariances estimated by the diagonal BEKK model are presented in Figure 1.

1,112

1-t1,11 79.00.140.0000169 ++= tt hh (4)

1,121,122

1-t2,2

1-t1,12 884.00.0480.00000288 ++= ttt hhh (5)

1,222

1-t2,22 98.00.0160.00000032 ++= tt hh (6)

1,131-t3,1-t1,13 88.00.0570.00000184 ++= tt hh (7)

1,231,231-t3,1-t2,1-t3,1-t2,23 978.00.0190.00000015 ++= ttt hhh (8)

1,332

1-t3,33 974.00.0220.00000043 ++= tt hh (9)

International Research Journal of Finance and Economics - Issue 58 (2010) 146 Table 5: Estimated coefficients for variance covariance equations

Estimated coefficient

Standard error

C11 0.0000169* 0.0000031 C12 0.00000288* 0.0000007 C13 0.00000184* 0.0000006 C22 0.00000032* 0.0000002 C23 0.00000015** 0.0000001 C33 0.00000043** 0.0000002 A11 0.37707* 0.0250440 A22 0.128127* 0.0099990 A33 0.151215* 0.0102380 B11 0.891956* 0.0126080 B22 0.99153* 0.0014220 B33 0.987217* 0.0018490 * indicates statistical significance at the 1% level. ** indicates statistical significance at the 5% level.

Figure 1: Conditional variances covariances estimated by the diagonal BEKK model

.000

.001

.002

.003

.004

500 1000 1500 2000

Var(BSE)

-.0002

-.0001

.0000

.0001

.0002

.0003

.0004

500 1000 1500 2000

Cov(BSE,HSI)

.0000

.0004

.0008

.0012

.0016

500 1000 1500 2000

Var(HSI)

-.0002

-.0001

.0000

.0001

.0002

.0003

.0004

500 1000 1500 2000

Cov(BSE,STI)

.0000

.0002

.0004

.0006

.0008

.0010

500 1000 1500 2000

Cov(HSI,STI)

.0000

.0002

.0004

.0006

.0008

.0010

.0012

.0014

500 1000 1500 2000

Var(STI)

Conditional Variance Covariance

Observations Observations

Observations Observations

ObservationsObservations

147 International Research Journal of Finance and Economics - Issue 58 (2010) The overall persistence of the variance covariance equations measured by the sum of the

ARCH and GARCH coefficients provides evidence of strong volatility, which possibly establishes unit root in variance. Finally, the Ljung Box Q statistics for the 1st (0.1035), 12th (0.1170) and 24th (0.1687) orders in squared standardized residuals show that there is no series dependence in the squared standardized residuals, indicating the appropriateness of the diagonal BEKK GARCH model.

4. Conclusions This study employs the multivariate diagonal BEKK GARCH model to examine the transmission of returns and volatilities among the markets of India, Hong Kong and Singapore. The estimated coefficients from the conditional mean return equations indicate that all examined markets are highly integrated, reacting in common information which mostly derives from the USA market, the largest information producer in the world. Furthermore, the results indicate that the Hong Kong market is the most efficient, followed by the markets of India and Singapore.

The magnitude and persistence of the coefficients of the variance covariance equations indicate that all markets exhibit strong GARCH effects, as old news have a higher impact on conditional volatility than current news, a fact which implies that markets are probably inefficient and shocks (news) are slowly absorbed by the market. Furthermore, the magnitude and the statistical significance of the covariances indicate that volatility formation underlies in commonly available information, which confirms the argument that stock markets are becoming more integrated.

References [1] Aggarwal, R. 1988. Stock index futures and cash market volatility, Review of Futures

Market, 7, pp. 290-99. [2] Akgiray, V. 1989. Conditional heteroskedasticity in time series of stock returns: evidence and

forecasts, Journal of Business, 62, pp. 55-80. [3] Andersen, T.G. and T. Bollerslev 1998. Answering the skeptics: yes, standard volatility

models do provide accurate forecasts, International Economic Review, 39, pp. 885-905. [4] Andersen, T.G., T. Bollerslev and S. Lange 1999. Forecasting financial market volatility:

sample frequency vis--vis forecast horizon, Journal of Empirical Finance, 6, pp. 457-477. [5] Antoniou, A., P. Holmes and R. Priestley 1998. The effects of stock index futures trading on

stock index volatility: an analysis of the asymmetric response of volatility to news, Journal of Futures Market, 8, pp.151-66.

[6] Bae, S.C., T.H. Kwon and J.W. Park, 2004. Futures trading, spot market volatility, and market efficiency: The case pf the Korean index futures markets, Journal of Futures Markets, 24, pp. 1195-1228.

[7] Bessembinder, H. and P.J. Seguin 1992. Futures trading activity and stock price volatility, Journal of Finance, 47, pp. 2015-34.

[8] Black, F. (1986) Noise, Journal of Finance, 41, 529-43. [9] Bollerslev, . 1986. Generalized autoregressive conditional heteroskedasticity, Journal of

Econometrics, 31, pp. 307-27. [10] Bologna, P. and L. Cavallo 2002. Does the introduction of stock index futures effectively

reduce stock market volatility? Is the future effect immediate? Evidence from the Italian stock exchange using GARCH, Applied Financial Economics, 12, pp. 183-92.

[11] Brailsford, T.J. and R.W. Faff 1993. Modelling Australian stock market volatility, Australian Journal of Management, 18, pp. 109-32.

[12] Cha, B. and Y.L. Cheung 1998. The impact of the U.S. and the Japanese equity markets on the emerging Asia Pacific equity markets, Asia-Pacific Financial Markets, 5, pp. 191-209.

[13] Chatrath, A., F. Song and B. Adrangi 2003. Futures trading activity and stock price volatility: some extension, Applied Financial Economics, 13, pp. 655-64.

International Research Journal of Finance and Economics - Issue 58 (2010) 148 [14] Darrat, A.F. and S. Rahman, 1995. Has futures trading activity caused stock price volatility,

Journal of futures markets, 15, pp. 537-57. [15] Dekker, A., K. Sen and M. Young 2001. Equity market linkages in the Asia Pacific region. A

comparison of the orthogonalised and generalized VAR approaches, Global Finance Journal, 12, pp. 1-33.

[16] Edwards, F.R. 1988. Does the future trading increase stock market volatility?, Financial Analysts Journal, 44, pp. 63-9.

[17] Engle, R.F. 1982. Autoregressive conditional Heteroscedasticity with estimates of the variance of United Kingdom inflation, Econometrica, 50, pp. 987-1007

[18] Eun, C.S. and S. Shim 1989. International transmission of stock market movements, Journal of financial and Quantitative Analysis, 24, pp. 241-56.

[19] Fama, E.F. 1963. Mandelbrot and the stable paretian distribution, Journal of Business, 36, pp. 420-429

[20] Fama, E.F. 1965. The behavior of stock market prices, Journal of Business, 38, pp. 34-105. [21] Gunasinghe, W.I.C.S. 2005. Behaviour of Stock Markets in South Asia: An Econometric

Investigation, South Asia Economic Journal, 6, pp. 165-191. [22] Hamao, Y., R.W. Masulis and V.K. Ng 1990. Correlations in price changes and volatility

across international stock markets, Review of Financial Studies, 3, pp. 281-307. [23] Harris, L. (1989) S&P 500 Cash stock price volatilities, Journal of Finance, 44, pp. 1155-75. [24] In, F., S. Kim, J.H. Yoon and C. Viney (2001) Dynamic Interdependence and Volatility

Transmission of Asian Stock Markets. Evidence from the Asian crisis, International Review of Financial Analysis, 10, pp. 87-96.

[25] Kan, .C.V. 1999. The effect of index futures trading on volatility of HIS constituent stocks, Pacific-Basin Finance Journal, 5, pp. 105-14.

[26] Knif, J. and S. Pynnonen 1999. Local and global price memory of international stock markets, Journal of International Financial Markets, Institutions and Money, 9, pp. 129-147.

[27] Kyle, S. 1985. Continuous Auctions and Insider Trading, Econometrica, 53, pp. 1315-1335. [28] Lee, S. B. and K. Y. Ohk 1992. Stock index futures listing and structural change in time-

varying volatility, Journal of Futures Markets, 12, pp. 493-509. [29] Lien, D. and L. Yang 2003. Options expiration effects and the role of individual share futures

contracts, Journal of Future Markets, 11, pp. 1107-1118. [30] Martens, M. 2001. Forecasting daily exchange rate volatility using intraday returns, Journal

of International Money and Finance, 20, pp. 1-23. [31] Mazouz, K. and M. Bowe 2006. The volatility effect of futures trading: Evidence from LSE

traded stock listed as individual equity futures contracts on LIFFE, International Review of Financial Analysis, 15, pp. 1-20.

[32] Mckenzie, M.D., T.J. Brailsford and R.W. Faff 2001. New insights into the impact on the introduction of futures trading o stock price volatility, Journal of Future Markets, 21, 237-255.

[33] McMillan, D.G. and A.E.H. Speight 2004. Daily volatility forecast: Reassessing the performance of GARCH models, Journal of Forecasting, 23, pp. 449-460.

[34] Michelfelder, R. and S. Pandya 2005. Volatility of stock returns: emerging and mature markets, Managerial Finance, 31, pp. 66-86.

[35] Najand, M. 2002. Forecasting stock index futures price volatility: Linear vs nonlinear models, The Financial Review, 37, pp. 93-104.

[36] Pan, M.S. and P.L. Hsueh 1998. Transmission of stock returns and volatility between U.S. and Japan: evidence from stock index future markets, Asia-Pacific Financial Markets, 5, pp. 211-225.

[37] Pilar, C. and S. Rafael 2002. Does derivatives trading destabilize the underlying assets? Evidence from the Spanish stock market, Applied Economics Letters, 9, pp. 107-10.

149 International Research Journal of Finance and Economics - Issue 58 (2010) [38] Poshakwale, S. and W.C. Pok 2004. The impact of the introduction of futures contracts on the

spot market volatility: the case of Kuala Lumpur Stock Exchange, Applied Financial Economics, 14, pp. 143-154.

[39] Rahman, S. 2001. The introduction of derivatives on the Dow Jones Industrial Average and their impact on the volatility of component stocks, The Journal of Future Markets, 21, pp. 633-653.

[40] Santoni, G.J. 1987. Has programmed trading made stock price more volatile?, Federal Reserve Bank of St. Louis Review, 69, pp. 8-29.

[41] Sariannidis N., I. Koskosas, A. Garefalakis and I. Antoniadis 2009. Volatility of Stock Returns: The Case of the Belgian Stock Exchange, International Journal of Business Forecasting and Marketing Intelligence, 1, pp. 111-121.

[42] Stein, J.C. 1987. Informational externalities and welfare-reducing Speculation, Journal of Political Economy, 95, pp. 1123 - 45.

[43] Tay, N. and Z. Zhu 2000. Correlations in returns and volatilities in Pacific Rim stock markets, Open Economic Review, 11, pp. 27-47.

[44] Theodossiou, P. and U. Lee 1993. Mean and volatility spillovers across major national stock market: Further empirical evidence, The Journal of Financial Research, 4, pp. 337-350.

[45] Theodossiou, P. and U. Lee 1995. Relationship between volatility and expected returns across international stock markets, Journal of Business Finance and Accounting, 22, pp. 289-300

[46] Von Furstenberg, G.M. and B.N. Jeon 1989. International stock price movements: Links and messages, Brooking Papers on Economic Activity, 1, pp. 125-79.

[47] Worthington, A. and H. Higgs 2004. Transmission of equity returns and volatility in Asian developed and emerging markets: A multivariate GARCH analysis, International Journal of Finance and Economics, 9, pp. 71-80.