Modeling Exchange Rate Volatility in the Presence of Serial Correlations
Modeling Exchange Rate Volatility in the Presence of Serial Correlations

Summary/Abstract: The objective of this study was to model the volatility in GHC/US$ exchange rate series taking into consideration the presence of serial correlation. The data used was GHC/US$ exchange rate from January 2000 to August 2019. To select appropriate model for modeling volatility GARCH(p,q), TARCH(p,q) EGARC(p,q), PARCH(p,q) and APARCH(p,q) were estimated and evaluated. The ARMA(3,3)-TARCH(2,1)-GED was selected as the appropriate model. It was found out that the return series had serial correlation problem. It was found that heteroscedasticity was present and was captured by ARMA(3,3)-TARCH(2,1) model under general error distribution but could not account for the account for the serial correlation in the return series. However, the corresponding GARCH-M-TARCH(2,1) model under general error distribution sufficiently captured the presence of serial correlation. From the results when the existence of serial correlations were ignored in the return series the parameters estimated will be bias and inefficient. Hence, the application of GARCH-M types of provided possible feedback between the variance and the mean equation. It was also found out that previous information about volatility and the previous volatility had significant effect on the current day volatility. From the result there was no leverage effect and the impact of news was asymmetric.

• Details
• Contents