In this study we examine whether there are differences in the accuracy of forecasts produced by brokers for a sample of almost 300,000 forecasts of the earnings per share of UK firms, over a period which spans accounting year ends from 1987-95. We report evidence of differential forecast accuracy which is contrary to much of the existing US research, in particular O'Brien (1985 and 1990). We use a forecast error that is controlled for the size of the firm being forecast, the industry to which the firm belongs, the accounting year being forecast, and the forecast horizon. There is evidence of short term persistence in the relative performance of the brokers across the full sample, and for some brokers there is evidence of forecast superiority over an extended period. We also find evidence that the relative accuracy of brokers differs between industries, which suggests that brokers tend to develop expertise in particular sectors. There are significant differences between brokers' accuracy in seven of the 11 industries in the sample, and in some industries there is evidence of persistently superior performance. Overall, these findings have implications for the construction of the earnings expectations proxies that are used in capital market research.
- forecast accuracy