ARIMA ve VAR Modellerinin Tahmin Başarılarının Karşılaştırılması

This paper considers the forecast accuracies of VAR and ARIMA models. The paper, hence, employs monthly Turkish CPI, Exchange Rate and Interest rate variables for the period 1994:1-200:07, and, observes the ex-post forecast values of the relevant variables. To this end, paper first determines the final AR and MA parameters through ACF and PACF estimations, later, identifies the best VAR model among others through Sims, LR and SC, and, AC criteria. Eventually, statistical analyses throughout MAE, MAPE, MSE, RMSE, Theil U1 and Theil U2 criteria evaluations, this paper reveals that VAR forecast is superior to ARIMA forecast for the relevant variables