Performance of Garch Models in Forecasting the Volatility of Exchange Rate for Developing Countries

This paper empirically investigates the performance of GARCH model in forecasting the volatility of exchange rate of some developing countries. We apply linear GARCH model and non-linear GARCH model. We fit these two models to some developing countries exchange rate index from January, 1998 to February, 2005. The return series of the developing countries＇ foreign exchange rate are leptokurtic, significantly skew, deviating from normality and volatile clustering as well. We find within-sample and out-of-sample evidence that conditional estimates of non-linear GARCH model outperform the conditional estimations of linear GARCH models. In our comparisons in most of the developing countries, the non-linear GARCH model produce better results than the linear GARCH model tor forecasting the volatility of exchange rate.

Performance of Garch Models in Forecasting the Volatility of Exchange Rate for Developing Countries