VAR, ARIMA, Üstsel Düzleme, Karma ve İlave-Faktör Yöntemlerinin Özel Tüketim Harcamalarına ait Ex Post Öngörü Başarılarının Karşılaştırılması

The aim of this study is to compare the ex post forecast accuracies of VAR, ARIMA, ES, Combining and Add-factor methods. In this comparison, the ex post forecasts of 2000:1-2000:4 are obtained by using the data of the Turkish private consumption for the period of 1987:1-1999:4. Beside private consumption, for VAR method, the Turkish GDP data is employed for the same periods. Later, the seasonality and stationarity analyses are run for these two series. The series are seasonally adjusted by the additive decomposition method and found as I(1). In the following steps, the ex post forecast models of these methods are established. Forecast outputs are evaluated by the criteria of MAE, MAPE, MSE, RMSE and Theil U. In conclusion of this analysis, the combining model of VAR-ES is found the best among others.