Modeling and forecasting exchange rate dynamics in Pakistan using ARCH family of models

Abstract

The main objective of this paper is to provide an exclusive understanding about the theoretical and empirical working of the GARCH class of models as well as to exploit the potential gains in modeling conditional variance, once it is confirmed that conditional mean model errors present time varying volatility. Another objective is to search the best time series model among autoregressive moving average (ARMA), autoregressive conditional heteroscedasticity (ARCH), generalized autoregressive conditional heteroscedasticity (GARCH), and exponential generalized autoregressive conditional heteroscedasticity (EGARCH) to give best prediction of exchange rates. The data used in present study consists of monthly exchange rates of Pakistan for the period ranging from July 1981 to May 2010 obtained from the State Bank of Pakistan. GARCH (1,2) is found to be best to remove the persistence in volatility while EGARCH(1,2) successfully overcome the leverage effect in the exchange rate returns under study.