A nonlinear time-varying copula using kink approach

Abstract

The structural change in copula parameter is motivates us to propose a flexible non-linear time varying copula that allows for capturing the structural change in the time varying dependence between variables. In this study, two families of Elliptical copula, namely Gaussian and Student-t copulas, are considered. We conduct a simulation study to examine the performance and accuracy of the proposed model and we obtain the reliable and acceptable results. In addition, the new model is applied to explain the dependence between S&P 500 and FTSE 100 stock markets. The proposed model fits well with these datasets and shows evidence of structural change in the dependence structure overtime. Moreover, the nonlinear time varying copula outperforms the conventional linear time varying copula suggesting that our model can detect better the dependence structure of these two stock markets.