International portfolio of stock indices with spatiotemporal correlations: Can investors still benefit from portfolio, when and where?

Abstract

This paper revisits the topic of international portfolio of stock indices under spatiotemporal correlations, to help people to get better portfolio performance in international stock markets. We firstly develop a mean-VaR framework as well as a mean-CVaR one, where both of which are with spatiotemporal correlation and other constraints. Then we apply these two frameworks to investigate whether investors can gain in international stock markets or not under various constraints. Our empirical results find that 1) Investors can still benefit from the international portfolio with spatiotemporal correlations, either in the view of avoiding risk or pursuing the profit; 2) the spatiotemporal correlation and exchange rate contribute to the performance of portfolio significantly, and transaction cost and fixed income rarely have effect on the portfolio, both in crisis and calm periods. Additionally, in the period of calm, the skewness of each single return series has some significant impact on the portfolio performance; 3) the portfolio with lowest spatiotemporal correlation with other markets is the optimal choice. In addition, in the calm period, another suitable area can be the one with positive mean and negative skewness of returns, such as in the U.K. market; 4) the mean-CVaR framework outperforms the mean-VaR one in financial calm period, but equals to the latter in crisis time. Our results demonstrate that the proposed mean-CVaR programming framework with spatiotemporal correlation provides a more flexible and effective decision support tool for international portfolio.