Journal Article

2015-05-28

Journal for Studies in Economics and Econometrics

University of Cape Town

This paper considers the ex post accuracy of two published sets of macroeconomic forecasts: the Economist Intelligence Unit's (EIU's) Country Report and the "Economist of the Year" consensus forecast, sponsored by Media 24. Root weighted mean square errors (RWMSEs) of these two competing entities' forecasts are, on average, about 60% of the RWMSEs of adaptive naïve predictions for current-year forecasts and about 75% of the RWMSEs of adaptive naïve predictions for one-year-ahead forecasts. There is no substantial difference in the forecast accuracy of the two entities, despite large differences in forecasting methodology. Forecasts are revised on a monthly basis. Monthly revisions in Media 24's consensus forecast are substantially more likely to reduce forecast errors than the EIU's monthly forecast revisions. Revisions in the next-year forecasts for GDP growth and inflation are only marginally better than random, and thus such revisions have little credibility.