The Risk-return Analysis Of International Indices Portfolio:Based On GAS Dynamic Vine Copula

This paper considers the problem of modeling the complex non-linear dependence among high dimensional financial time-series.We extend the Generalized Autoregressive Score model in different kinds of pair-copula models and show the way of parameters estimation.The results of simulation confirm that GAS model outperforms the ARMA model and static model in estimating the dynamic parameters of pair-copulas.Meanwhile,simulation shows the AIC is more preferable than BIC on the issue of copula selection.In the empirical study,a portfolio contains both developing and developed market indices has been built.We firstly estimate all the parameters of two kinds of dynamic vine-copula model,then evaluate Va R of the portfolio based on Monte Carlo method and then use Va R backtesting to test models’ performance.Lastly,optimizing the portfolio based on three copula models and forecasting the outof-sample return.Empirical studies show the international indices portfolio has significant non-linear dynamic correlation and the time-varying pair-copula in the first tree propagates to the whole distribution.Besides,the selection of vine structure also has a great impact on the goodness-of-fit.The GAS dynamic vine-copula passes the Va R test in three significant level and verifies its outperformance of earning capacity.