Abstract

This thesis investigates the extent to which investors are rational with respect to two well-known empirical properties of analyst consensus forecasts: optimism and auto-correlated revisions.

Consistent with previous research, I find that investors are aware of the optimistic bias in analyst forecasts, but do not adjust for it completely. Using Mishkin's test, I find that the market's expectation adjusts for approximately 70% of the optimistic bias in analyst forecasts. The market is fully rational about the auto-correlation in two-year-ahead and three-year-ahead forecast revisions, but not so for one-year-ahead revisions: it recognizes 82% of the auto-correlation in one-year-ahead forecast revisions. I find that investing long (short) in firms in the top (bottom) decile of one-year-ahead forecast revisions generates a hedge portfolio return of about 6% in the following year. I investigate the market's rationality across industries. With respect to the optimistic bias, the market seems to be more rational for firms in the consumer service industry and the non-durable goods manufacturing industry. With respect to the auto-correlation in forecast revisions, the market seems to be more rational for firms in the transportation industry and the non-durable goods manufacturing industry. I further examine whether the degree of investor rationality is positively correlated with commonly used proxies for sophistication of marginal investors: size, analyst following and institutional ownership. I fail to find evidence that those proxies are positively correlated with the degree of investor rationality. Finally, I examine time-series variation in investor rationality, and find that investors seem to be more rational in recent years.