Many empirical studies find that the distribution of stock returns departs from normality. In such cases, it is desirable to employ a statistical estimation procedure that may be more efficient than ordinary least squares. This paper describes various robust methods, which have attracted increasing attention in the statistical literature, in the context of estimating beta risk. The empirical analysis documents the potential efficiency gains from using robust methods as an alternative to ordinary least squares, based on both simulated and actual returns data.

Analysts' earnings forecasts are influenced by their desire to win investment banking clients. We hypothesize that the equity bull market of the 1990s, along with the boom in investment banking business, exacerbated analysts' conflicts of interest and their incentives to strategically adjust forecasts to avoid earnings disappointments. We document shifts in the distribution of earnings surprises and related changes in the market's response to surprises and forecast revisions. The evidence for shifts is stronger for growth stocks, where conflicts of interest are more pronounced. However, shifts are less notable for analysts without ties to investment banking and in international markets.

ol ion ix3i b noiX3 .31UO2 r1I 01 n3'ig i 3iiOfl nibu!3ni 3ib13 11th r&D ELOLJJ 4JJG 2c9'UqbOiJJc o pocp cp6 9'C9'q6WIC LG2G9'LCGL 9'Uq cp iuiwuc bL9'ccJvJoucL GLGOL& (J2) OL 9' qfl,GL6IJ 9'bbLO9'Cp)