A Comparison of Multi-step GDP Forecasts for South Africa

To forecast at several, say h, periods into the future, a modeller faces two techniques: iterating one-step ahead forecasts (the IMS technique) or directly modeling the relation betwen observations separated by an h-period interval and using it for forecasting (DMS forecasting). It is known that structural breaks, unit-root non-stationarity and residual autocorrelation benefit DMS accuracy in finite samples, all of which occuring when modelling the South African GDP over the last thirty years. This paper analyzes the forecasting properties of the model developed by Aron and Muellbauer (2002) and compares with them that of 30 derived or competing models. We find that the GDP of South Africa is best forecast, 4 quarters ahead, using the technique developed by these authors and its variants as derived in the present paper. Rankings of other models vary over time and it is difficult to recommend one of them as a rule in this exercise.