The study highlights the statistical procedure employed in developing a shortterm forecasting model that explores the volatility feature of Philippine inflation from 1995 up to August 2007. To build such a model, we identify first the stationary series. Second, we specify the Autoregressive Moving Average (ARMA)model then include the Seasonal ARMA (SARMA) model if seasonality is present, to represent the mean component using the past values of inflation. Next, we incorporate the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model to represent its volatility. Diagnostic tests and examination of forecast accuracy measures indicate that the specifications D(IR) (the first difference of month-on-month inflation) as the stationary series, AR(1) and SMA(12) for the mean,
GARCH(0,1) or ARCH(1) for the variance with Student’s t distribution having fixed degrees of freedom v = 10 for the errors, performs best in forecasting the volatility of the inflation rate for the Philippines. Lastly, out of sample forecasts for the mean and error variance of Philippine inflation from September 2007 to October 2007 are achieved using dynamic forecasting.
