Whose inflation expectations forecast best? Lessons from South African survey data

As more survey-based measures of inflation expectations have become available, we ask whether some are better than others at forecasting inflation and whether forecast averaging outperforms individual forecasts. We consider three unique quarterly surveys of firms, financial analysts and trade unions from South Africa. These surveys possess unique features including the time span over which they are available as well as having been conducted under a single monetary policy regime of inflation targeting since 2000. We find that a linear combination of forecasts obtained from a factor model can improve the accuracy of forecasting over alternative forms of aggregation. By going beyond the typical reliance on professional forecasters, exploiting the underlying heterogeneity of expectations of firms and trade unions can improve forecasts. We also demonstrate how the state of the economy and the forecast horizon can dictate which forecasts perform best. We believe that our results have broader implications for the study of inflation expectations especially in emerging market economies.