Abstract

This study sought to evaluate the prediction ability of ARIMA models on the current account deficit in Nigeria using symmetric loss functions like Mean Absolute Error (MAE), Root Mean Absolute Error (RMAE), Mean Absolute Percentage Error (MAPE) and Theil inequality Coefficient. The results of the unit root tests show that the data is stationary at lag 3 after taking the first difference of the series. Of two possible models, ARIMA (2,1,2) model is found to be the best model because it has lower AIC and BIC. The accuracy of volatility forecast of the two models were evaluated using the above symmetric loss functions and the results show that ARIMA (2,1,2) produces the most accurate forecast as it has least variance and maximum covariance proportions

Key words: ARIMA, Current Account Deficits,Symmetry Lost Functions