Improving forecasting in an emerging economy, South Africa: Changing trends, long run restrictions and disaggregation

Forecasting inflation is particularly challenging in emerging markets, where trade and monetary policy regimes have shifted and the exchange rate and food prices are highly volatile. This study shows that the information in long-run co-integrated relationships for relative prices in South Africa is helpful in outperforming univariate benchmark models for forecasting inflation. It also investigates gains to the inflation forecast accuracy as a result of aggregating weighted forecasts of the sub-component price indices, versus forecasting the aggregate consumer price index itself. Rich multivariate equilibrium correction models employ general and sectoral information for ten sub-components, including structural breaks and institutional changes. Model selection over the period 1979–2003 generates pseudo out-of-sample forecasts, four quarters ahead, until 2007. The largest gain in forecast accuracy against naïve benchmark models comes from incorporating equilibrium correction into the long-run. For more sophisticated models, aggregating the weighted forecasts of the sub-components outperforms the aggregate forecasts. The analysis also contributes to an improved understanding of sectoral inflationary pressures.