Two essays on self-selection bias in consensus analysts' forecasts

In this dissertation, I investigate a link between self-selection bias in analysts' forecasts and optimism in consensus analysts' forecasts. Self-selection arises when analysts drop coverage of firms with unfavorable future prospects, which adds to optimism in analysts' forecasts.; In my first essay, I examine four primary issues. First, I investigate whether self-selection explains the cross-sectional variation of analyst optimism after controlling for other explanations associated with analyst optimism. Second, I empirically examine factors that may determine analysts' behavior, specifically the dropping of coverage. Third, I revisit analyst underreaction to past information about the firm's performance (Abarbanell and Bernard, 1992; Easterwood and Nutt, 1999). I propose that analyst underreaction to past news is a manifestation of self-selection. Finally, I examine whether the market fully recognizes the implications of self-selection in assessing analyst optimism.; By using two firm specific measures of the degree of self-selection, I provide evidence suggesting that self-selection is significant in explaining analyst optimism even after controlling for incentives and analyst underreaction. Also, I find some firm characteristics that affect the degree of selection. Firms experiencing financial distress, low long-term growth, and stock price declines appear more likely to be subject to self-selection by analysts. My findings also suggest that self-selection is responsible, at least in part, for analyst underreaction to past information about the firm's performance. Moreover, I find empirical evidence suggesting that investors do not fully recognize self-selection, and overprice firms which are dropped by analysts.; In my second essay, I reexamine the negative relation between dispersion in analysts' earnings forecasts and future returns. Recently, Diether, Malloy, and Scherbina (2002) attribute this relation to heterogeneous expectations and short-sales constraints advanced by Miller (1977). I argue that the negative relation between dispersion and future returns is attributable to analyst self-selection.; First, I examine the validity of the Diether et al (2002) conjecture. Unlike the predictions from heterogeneous expectations and short-sales constraints, changes in dispersion do not seem to explain future returns. The level of dispersion appears to predict future returns regardless of the degree of short-sales constraints (proxied by firm size and short interest). More telling, the negative relation between dispersion in forecasts and stock returns disappears once I control for future performance. I believe that prior results supporting Miller (1977) are likely driven by omitting an important correlated variable, dispersion and future performance.; Instead, my evidence suggests that the negative association between dispersion in analysts' forecasts and subsequent returns can be attributed to analyst self-selection. Given the same threshold (below which analysts do not report their expectations), high dispersion stocks tend to suffer a larger truncation from below compared to low dispersion stocks. Consistent with this notion, I find that high dispersion stocks are more likely to have optimistic and right-skewed forecasts, a significantly lower analyst coverage, and a higher incidence of delistings than low dispersion stocks.