Forecasting Maltese inflation using ARIMA models

Abstract

In view of the fact that the main aim of the European Central Bank is to maintain price stability, it is of great importance that members of the Eurosystem can forecast inflation accurately. This dissertation evaluates the performance of ARIMA models in forecasting overall HICP inflation in Malta, and, by doing so, highlights the main procedure that one needs to undertake to build an ARIMA model suitable for forecasting. The main objective of this dissertation is to try and find an appropriate ARIMA model which would be able to forecast Maltese inflation with sufficient accuracy. Towards this aim, this study considers a number of candidate models, some of which are then used to forecast monthly inflation 12-months ahead using out-of-sample forecasting. The model with the lowest forecasting error is then selected. The results reveal that ARIMA models produce reasonable results, particularly in the short-term, and hence it might be useful for the Central Bank to include them in its suite of forecasting models.