Managing portfolio risk during crisis times: A dynamic conditional correlation perspective

Abstract

In this paper, we examine correlations between major European government bonds during the sovereign debt crisis. We apply an intraday Dynamic Conditional Correlation (DCC) model to the high-frequency quote data of the MTS market. We find that the Italian and Spanish government bonds become less correlated with other countries’ debts and the correlation between the two countries’ debts fluctuates heavily over time, ranging from 0.1 to 0.9. The Securities Markets Programme of the ECB is successful in restoring the market confidence for the integrity of the Eurozone, increasing the correlations towards the level before the crisis. In addition, we examine four different methods for computing and forecasting intraday VaR, namely, historical simulation, the Constant Conditional Correlation (CCC) model, the bivariate DCC model, and the multivariate DCC model estimated by composite likelihood. We demonstrate that the bivariate DCC model is most capable of forecasting intraday VaR for the tail of the distribution.