MODELING EXCHANGE RATE VOLATILITY USING UNIVARIATE GARCH MODELS - ACASE STUDY OF THE CEDI/DOLLAR EXCHANGE RATE.

Abstract

This study examines exchange rate volatility with Generalised Autoregressive Conditional Heteroscedastic (GARCH) models using monthly exchange rate data from the bank of Ghana from January 1990 to November 2013. The data was converted to returns to enhance their statistical properties and the returns used to fit a mean equation. The ARCH LM test of the mean equation revealed the presence of conditional heteroscedasticity. The returns were therefore modeled using ARCH (3), GARCH (2,3), Exponential Generalised Autoregressive Conditional Heteroscedastic (EGARCH) (2,2) and Threshold Generalized Autoregressive Conditional Heteroscedastic (2,3). The result revealed that EGARCH (2, 2) was the best since it has the least value of AIC of (-6.2816) and SIC of (-6.1629). Diagnostic test of the EGARCH (2, 2) model residuals with the Ljung-Box and the ARCH LM tests revealed that the models were free from higher order autocorrelation and conditional heteroscedasticity respectively. The Chi-square goodness of fit test showed that the forecasted values obtained from the EGARCH (2, 2) model were not significantly different from the observed values at the 5% significance level. The leverage parameter of the EGARCH (2, 2) model was significant and positive indicating the absence of leverage effect in the returns of the cedi dollar exchange rate. The absence of the leverage effect in the exchange rate indicates that positive shocks increases volatility than negative shocks of equal magnitude. Thus, the implication is that a strengthening dollar (weakening cedi) leads to higher period volatility than when the cedi strengthens by the same amount. It is recommended that the central bank should put in place longterm measures to stabilise the cedi since a weakening cedi increases the uncertainty in the exchange market than a strengthening cedi.