Three essays on analyst earnings forecast: Dispersion, accuracy and applications

Financial analyst earnings forecast dispersion has been used as a proxy for differences of opinion or earnings uncertainty, yet this dispersion also results from the capability, bias and incentives of analysts. I design a new measure of forecast dispersion which has analyst-specific effects purged and thus better reflects the drivers of firm uncertainty. This measure is statistically different from the commonly used dispersion measures, and it yields some different results when measuring the association between firm characteristics and firm uncertainty.;I find that analyst-specific effects are important determinants of forecast accuracy, too, and that it is necessary to control for them when examining the relation between accuracy and other analyst characteristics, such as experience and coverage complexity. I also show that investment banking affiliation generally improves accuracy. In a sample of seasoned equity offerings, I show that new affiliated analysts exhibit better accuracy overall, which suggests that the superior information and incentives of maintaining underwriting business by new affiliated analysts dominate the effect of previously accumulated firm-specific expertise of old affiliated analysts. I also find that after controlling for analyst affiliation and timing of forecasts made, issuer firm's forecast dispersion decreases before the SEO filing date, indicating that prior voluntary disclosure reduces differences of opinions more than the official announcement discloses clearer information after filing, or that a seasoned equity offering adds more uncertainty to the issuer firm.