Dependence measure of five-minutes returns compared to daily returns using static and dynamic copulas

Studies on dependence between stock markets are crucial because of their indications on the process of decision-making in investment strategies. Many previous studies measure the dependence between stock markets using static copula. However, in recent years, dynamic copula has been used as an alternative for measuring dependence due to its capability of capturing time-varying dependence between stock markets. Many copula studies have been focusing on examining the correlation of the bivariate data of daily, or weekly, or monthly returns to explain the co-movement between financial markets and for possible financial directions on portfolio management. However, information of low-frequency data is unable to accommodate large-scale trading activities. On the other hand, high frequency data contains more information about the stock market and has the ability to reflect stock market volatility more accurately. Therefore, this study aims to compare the dependence of the five-minutes returns (or high-frequency data) with the daily returns (or low-frequency data) to determine whether these data have similar or different dependence structures. Both static and dynamic copula models are utilized to capture the existence of time-varying dependence of the bivariate data. For numerical examples, the bivariate returns series of the Islamic (FBMHS) and conventional (KLCI) stock markets in Malaysia are utilized to model the dependence of the daily data and the dependence of the five-minute data. Findings of this paper shows that the structure of dependency between daily returns and 5-minute logarithmic realized variance are different, and portfolio diversification between KLCI-FBMHS pair is not advisable. Finally, the 5-minute series and dynamic SJC copula model are chosen as the best data set and the best dependency model, respectively