The estimation and determinants of emerging market country risk and the dynamic conditional correlation GARCH model

Country risk assessment is central to the international investment, which recently has increasingly focused on emerging markets (EM). In this paper we proxy for country risk in EM by using time-varying beta. We extend existing literature by applying a dynamic conditional correlation GARCH model. After confirming beta is time varying in twenty EM over the period January 1995 to December 2008 we investigate the GARCH (1,1) model and find the t-distribution generates the lowest forecast errors compared to the normal error distribution and a generalised error distribution. In a comparison of previous modelling techniques the results of our modified Diebold-Mariano test statistics suggest that the Kalman Filter model outperforms the GARCH model and the Schwert and Seguin (1990) model. Using a DCC-GARCH model our evidence suggests that considering dynamic betas can improve beta out-of-sample predicting ability and therefore offers potential gains for investors. Finally, we find dynamic betas across EM are strongly associated with each nation's interest rates, US interest rates and the Consumer Price Index (CPI) and to a lesser extent the exchange rates. Our results have some similarities to those in previous studies of developed markets in the economic determinants of time-varying beta but differences exist in the results on best model to forecast time-varying beta. These findings have implications for estimating country risk for investment and risk management purposes in EM.