Time Series Properties and Predictability of Pak Rupee/US Dollar Exchange Rate

This paper investigate the time series properties and predictability of daily percentage changes in the Pakistani rupee exchange rate with respect to the currencies of major trading partner country USA. The daily data is used for the time period of October 1988 to April 2012. In this study, we employed the EGARCH-M model along with the power exponential distribution. We find that the volatility of the Pakistani rupee is best elucidated by the EGARCH process as using the past volatility. The coefficient of moving average of the conditional mean equation suggests that percentage change in the exchange is predictable using the past innovation. The results for the conditional variance equation indicate the presence of conditional heteroskedasticity in the series. The coefficient for past innovation and past conditional variance are statistically significant. The coefficient of conditional variance is less than one indicating that the conditional variance is stationary and mean reverting. The asymmetry coefficient is also statistically significant, which allow us to accept the hypothesis of asymmetry volatility. The half-life of a shock on volatility is almost 3 days for rupee and us dollar exchange rate. Asymmetry effect is 1.06, which explains that the negative impact is 1.06 times more than the positive impact