COMPARISON OF ARIMA AND SINGULAR SPECTRUM ANALYSIS IN FORECASTING THE PHILIPPINE INFLATION RATE

In this study, an appropriate forecasting method for the Philippine inflation rate was determined. Specifically, the forecasting performances of the autoregressive integrated moving average (ARIMA) model and singular spectrum analysis (SSA) were compared. The dataset on monthly inflation rate was divided into two samples, the in-sample data and out-of-sample data. These samples were analyzed and after application of the Box-Jenkins approach, the resulting seasonal model is ARIMA (1, 1, 0) x (0, 0, 1)(12). This model has the least AIC among all tentative models and the behavior of its residuals and forecast errors are satisfactory. SSA was also applied on the same samples and the most appropriate window length for the trajectory matrix is 25. In reconstructing the series, 10 eigentriples under three groups were used. The first group contains the trend component while the second and third groups contain the oscillatory components. Based on the computed in-sample and out-of-sample RMSE, SSA outperforms the ARIMA model. Thus, SSA is better than ARIMA in forecasting the Philippine inflation rate based on the sample data set.