The accuracy of OECD forecasts

This paper examines the accuracy and properties of forecasts by the OECD for 24 countries and 8 variables. First the forecasts made in December of yeart?1 for yeart are examined, with the largest errors being for investment, industrial production and foreign balance. Next the way forecasts are revised between December and July is considered. Systematic revisions occur for Iceland, Turkey and Luxembourg. Finally the accuracy of forecasts made before, during and after the 1979 oil price rises are compared, and no evidence of a worsening of accuracy is found.