Application of Arima Models to Nigerian Inflation Dynamics

Abstract

The objectives of this study were to empirically develop a univariate autoregressive integrated moving average (ARIMA) model suggested by Box & Jenkins (1976) for Nigerian inflation and analyze the forecasting performance of the estimated model between 1981 and 2010. In this study, the analyses were carried out with the aid of EViews and Excel softwares. The study used the Ordinary Least Squares (OLS) technique for estimation purposes. On the basis of various diagnostic and selection evaluation criteria the best model was selected for the short term forecasting of Nigerian inflation. The study found ARIMA (2,2,3) as the most appropriate model under model identification, parameter estimation, diagnostic checking and forecasting inflation. In-sample forecasting was attempted and the estimated ARIMA model remarkably tracked the actual inflation during the sample period. The study concluded that Nigerian inflation is largely expectations-driven. The major inference that can be drawn in this study is that expectations that are formed about future levels of prices affect the current purchase decisions. It was recommended that, to put inflation under control, there is need for high transparency in monetary policy making and implementation.

Keywords: Inflation dynamics, ARIMA, expectations