Alternative Indicators of Forecast Revision and Improvement

A number of alternative measures of subjective forecast revision and forecast improvement/degradation are examined, their relative merits and limitations discussed, and some guidelines for selection provided. The analysis is restricted to indicators suitable for cross‐sectional data, and the main focus is on the mathematical properties of the measures. It is shown that, with respect to forecast revision measures, those indicators which are most easily understood (and, therefore, intuitively appealing) are limited in terms of their practical usefulness. As far as the indicators of forecast improvement/degradation are concerned, the analysis shows that universally applicable percentage measures cannot be established with any degree of confidence, although they may be appropriate in a limited number of situations.