Modelling the South African inflation rate using Box-Jenkins ARIMA models

This study investigates the performance of the South African inflation rate using Box-Jenkins ARIMA models. Several competing ARIMA specifications were identified through ACF, PACF, and EACF analyses, including ARIMA(1,1,0), ARIMA(2,1,0), ARIMA(1,1,1), and ARIMA(2,1,1). All models were estimated using the maximum likelihood method, with results indicating statistical significance and low standard errors across the board, suggesting strong model fit. The optimal model, ARIMA(1,1,1), was selected based on the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC), aligning with findings by Mondal et al. (2014). Diagnostic tests, including the Ljung-Box test and residual analysis, confirmed that the ARIMA(1,1,1) model is robust and reliable for modelling inflation dynamics in South Africa. The study highlights the usefulness of ARIMA models in forecasting inflation, a crucial task for policymakers and the South African Reserve Bank in managing inflation expectations and guiding monetary policy. While the linear ARIMA model performed well, the study also recognises its limitations in capturing complex macroeconomic behaviours, suggesting future exploration of nonlinear models such as GARCH. Though the findings are specific to South Africa, the approach provides a replicable framework for other macroeconomic applications and geographical contexts.