A predictive Autoregressive Integrated Moving Average (ARIMA) Model for forecasting inflation rates

Abstract

This paper considers the modelling and forecasting of monthly All-items (12 months average change) inflation rates in Nigeria using the Box-Jenkins (ARIMA) model. Time series data used in the study was collected from the Central Bank of Nigeria statistical web database. The data was differenced twice to achieve stationarity in the series as required. Based on the evaluation and diagnostic criteria, the most accurate model is selected. The order of the best ARIMA model was found to be ARIMA (1, 2, 1). The diagnostic analysis of the model residuals showed that they are normally distributed uncorrelated random shocks. The findings in this study showed that the selected ARIMA model captured the dynamics in the series and produced forecasted values which had minimal forecast errors when compared with the actual inflation values in the validation period.