Has The Dependence Between The Sino And Foreign Stock Market Increased Through Time?

In the study of financial theory, the dependence between different asset plays a vital role. Understanding, collaboration and measurement of the dependence of financial assets is a subject of concern in asset pricing, portfolio allocation and financial risk management field. In the financial business community, many investors will investment their capital in different stock markets of several countries in order to benefit from diversification. Thus, the stock market linkage between different countries will directly affect the performance of the portfolio:if the dependence between the different stock markets which the investors have collocated their capital is high, the diversification benefits is relatively low; if the dependence between the different stock markets which the investors have collocated their capital is low, the diversification benefits is relatively high.Between the research methods for dependence of the stock markets, the most common model assumes that the distribution of multiple assets return series is multivariate normal. However, the correlation estimated by multivariate normal distribution is completely symmetrical, the correlation between left tails and right tails can not been distinguished, which does not match with the real situation:the dependence rise between stock markets is higher when prices fall than when prices rise (Login and Solnik,2001; Ang and Bekaert,2002). Secondly, since the return series of financial markets tend to exhibit characteristics of fat tail, normal distribution is not a good characterization of this feature, the correlation estimation results are also questionable (Patton,2006). Finally, in reality, a plurality of dependency between the distribution is not always normal, multivariate normal distribution may not be an appropriate distribution model. Compared to the methods that model the return series directly and set up a number of restrictive condition, copula is more flexible, making the model more realistic.Earlier research on copula is focused on binary models, that is only two-dimensional data modeling, and relevant theoretical approaches are relatively mature. However, in reality, a portfolio is often far more than two securities. Directly use the binary copula may cause the "curse of dimensionality" problem. In addition, directly extend the binary Copula to high-dimensional, will also cause limitations of parameters (Eike et al.,2013; GAO Jiang,2013), such as high-dimensional Archimedean Copula model family.This paper adopts two-dimensional time-varying SJC-copula model and vine copula to model the multivariate distribution of return serious, so that the multivariate distribution model of the return series can meet the time-varying and asymmetric feature of the dependence of financial return series, while the two-dimensional Copula extended to higher dimensional.In conclusion, this paper has studied the changes of dependence between the performance of Chinese stock market and the performance of the stock markets of many other foreign countries over time. As nonlinearity and asymmetry are the most common two dependence features of finance return series, this paper paid special attention to the time varying features of tail dependence. We adopted two different copulas, i.e., vine copula and time-varying SJC copula to estimate the dependence between Chinese stock market and the stock market of developed countries and Chinese stock market and the developing countries respectively.By comparing the back-testing results of the two asset portfolios, we can find out which copula is better:the more accurate the VaR is, the better is the estimation of the dependence. Finally, this paper got the three following results:firstly, Chinese stock markets is not closely related any other foreign stock market. But the lower dependence is more distinct than the upper dependence. Secondly, the dependence of Sino and developing stock markets is more obvious than the dependence of Sino and developed countries. Thirdly, the dependence between the Sino and foreign stock markets has not increased through the years. Lastly, by comparing the back-testing results of VaR, this study found that the estimation of time-varying SJC copula is more accurate than that of vine copula.