Acta Universitatis Danubius. Œconomica, Vol 9, No 1 (2013)

Abstract

In this study, transformations of SPF inflation forecasts were made in order to get more accurate predictions. The filters application and Holt Winters technique were chosen as possible strategies of improving the predictions accuracy. The quarterly inflation rate forecasts (1975 Q1-2012 Q3) of USA made by SPF were transformed using an exponential smoothing technique- Holt Winters- and these new predictions are better than the initial ones for all forecasting horizons of 4 quarters. Some filters were applied to SPF forecasts (Hodrick- Prescott, Band-Pass and Christiano-Fitzegerald filters), but Holt Winters method was superior. Full sample asymmetric (Christiano-Fitzegerald) and Band-Pass filter smoothed values are more accurate than the SPF expectations only for some forecast horizons.