Modelling and Forecasting of Exchange Rate Dynamics: Empirical Evidence from AFN/USD

This empirical study explores the changing aspects of the exchange rate, AFN versus USD, in Afghanistan, over an extended period ranging from March 1, 2003, until December 31, 2020, starting with the beginning of demonetization up to the eve of a political change in Afghanistan. The study focuses on illuminating the characteristics pertinent to modelling and predicting the instability of the exchange rate. The current study used univariate time series approaches to outline the pattern of the exchange rate and the ARCH and GARCH models. Both in-sample and out-of-sample datasets were developed to test the fitness and ability of models to predict the data. The results confirm that, over the selected time period, the exchange rate exhibits persistence in volatility with a stylized fact, namely, volatility clustering. The results highlight that the standard GARCH (1,1) model is effective in capturing and predicting the volatility dynamics of AFN/USD exchange rate returns, exhibiting strong volatility clustering and persistence. The models employed in this paper are highly useful for people who are engaged in the foreign exchange market and for policymakers while formulating economic policies in Afghanistan.