A Note on Analysts' Earnings Forecast Errors Distribution, Journal of Accounting and Economics
Abarbanell and Lehavy provide evidence that analysts' forecast errors are not normally distributed exhibiting a high occurrence of extreme negative forecast errors (left-tail asymmetry) and a high occurrence of small positive forecast errors (middle asymmetry). This is important for researchers who rely on techniques that are sensitive to the distributional assumptions of analysts' forecast errors. Many of the conclusions drawn by Abarbanell and Lehavy, however, are based on visual impressions (as opposed to formal empirical tests) or based on methods that are very sensitive to the empirical methods used (e.g., whether the serial correlation of forecast errors is caused by the left-tail asymmetry).
Daniel A Cohen, Thomas Lys
Cohen, Daniel A, and Thomas Lys. 2004. A Note on Analysts' Earnings Forecast Errors Distribution. Journal of Accounting and Economics. 36(1-3): 147-164.LINK