Abstract

The main objective of this study is to give theoretical and empirical analysis of the volatility of Afghani (AFN) against United States Dollar (USD) exchange rate. Specifically, the study focuses on the behavior of exchange rate volatility using ARCH/GARCH family of models and shows that errors in the conditional mean equation have time-varying volatility. In addition, the current study aims to select the best model from four types of time series models (ARMA, ARCH, GARCH, EGARCH and TGARCH) to estimate and forecast AFN/USD exchange rate movements with the highest accuracy. This research uses data on the daily exchange rate of AFN/USD from Da Afghanistan Bank. The dependent variable is the AFN/USD exchange rate return, while its conditional volatility is modelled using an ARMA framework combined with ARCH, GARCH, EGARCH, and TGARCH models, in which volatility is explained by past innovations (ARCH effects) and past conditional variances (GARCH effects). The empirical results indicate that the GARCH (1,1) model provides the best overall performance in modeling and forecasting AFN/USD exchange rate volatility. The estimated GARCH parameter is highly significant, indicating strong volatility persistence, whereas the asymmetric effects captured by the EGARCH and TGARCH models are not statistically significant during the sample period. These findings provide useful implications for policymakers, financial institutions, investors, and researchers involved in exchange rate risk management and forecasting.

Keywords

AFN/USD EGARCH Exchange Rate Volatility GARCH Time Series Analysis