The objective of this paper is to modelize the volatility of the Algerian exchange rate against the American dollar (DZA / US Dollar) and to predict the rate for the first three months of 2014. Our study showed that our series is characterized by the volatility of the phenomenon, by asymmetric specification and by the presence of excessive kurtosis. ARCH test was performed. This test has rejected the null hypothesis of homoscedasticity. Therefore we have concluded that a nonlinear ARMA model type ARCH is adequate. Then, we have estimated four ARCH type models: AR (2) -ARCH (2), AR (2) -GARCH (1,1), AR (2) -EGARCH (1,1) and AR (2) - TGARCH (1,1). The AIC leads us to choose the TGARCH (1,1) model as a suitable one for the prediction. The Forecasting quality criteria (RMSE and U Theil) indicate that we have reached a good prediction model.
Show abstract
Full Text
