Forecasting Core Inflation: The Case of South Africa

Forecasting and estimating core inflation has recently gained attention, especially for inflation targeting countries, following research showing that targeting headline inflation may not be optimal; a Central Bank can miss the signal due to the noise. Despite its importance there is sparse literature on estimating and forecasting core inflation in South Africa, with the focus still on measuring core inflation. This paper emphasises predicting core inflation using large time-varying parameter vector autoregressive models (TVP-VARs), factor augmented VAR, and structural break models using quarterly data from 1981Q1 to 2013Q4. We use mean squared forecast errors (MSFE) and predictive likelihoods to evaluate the forecasts. In general, we find that (i) small TVP-VARs consistently outperform all other models; (ii) models where the errors are heteroscedastic do better than models with homoscedastic errors; (iii) models assuming that the forgetting factor remains 0.99 throughout the forecast period outperforms models that allow for the forgetting factors to change with time; and (iv) allowing for structural break does not improve the predictability of core inflation. Overall, our results imply that additional information on the growth rate of the economy and interest rate is sufficient to forecast core inflation accurately, but the relationship between these three variables needs to be modelled in a time-varying (nonlinear) fashion.