Can realized covariance measure improve VaR forecasting of China’s new energy and financial markets? ——Evidence from the A-MRS-Copula model

This paper proposes a new adaptive Markov regime-switching copula (A-MRS-Copula) model based on the realized covariance measures to enhance the prediction accuracy of portfolio risk of China’ s new energy and financial markets. More specifically, the model relaxes the static transition probability condition of the traditional MRS-Copula model, allowing the copula parameters to vary adaptively based on high-frequency dependence information between assets. Furthermore, given the leverage effect prevalent in financial markets, we employ realized semicovariances as proxies for intraday dependencies in our modelling process. This approach improves the ability of the model to capture asset leverage effect and accurately depict dynamic dependence. Empirical results show that the A-MRS-Copula model exhibits statistically significant improvement in forecasting portfolio risk in new energy markets and other markets. Finally, robustness tests confirm that the new model outperforms the MRS-Copula model in terms of different data, copula function settings, prediction evaluation methods and rolling window.