Youfei Xiao

PhD Student, Accounting
PhD Program Office Graduate School of Business Stanford University 655 Knight Way Stanford, CA 94305

Youfei Xiao

Job Market Paper

Uncertainty, Disagreement and Forecast Dispersion: Empirical Measures from a Model of Analysts' Strategic Conduct

There is a sizable theoretical and empirical literature examining analyst forecast dispersion and its association with equity returns. This paper provides empirical estimates of uncertainty and disagreement about future earnings underlying analyst forecast dispersion from a theoretical model where analysts are motivated by strategic incentives. A parsimonious model which assumes that analysts' payoffs are jointly determined by forecast error and deviation from consensus reproduces many of the descriptive facts observed about forecast dispersion in the data. The strategic behavior that arises from the model distorts both the levels of forecast dispersion and the sensitivity of the measure with respect to cross-sectional variation in uncertainty. The estimated parameters perform better at predicting forecast dispersion out-of-sample than approaches based solely on associations with firm characteristics. The model-implied estimates of earnings uncertainty exhibit a substantially less negative association with future returns relative to the association generated by forecast dispersion and thus partially reconciles previous evidence with theories about the asset pricing implications of uncertainty and disagreement.

Working Papers

An Empirical Investigation of Analysts' Objective Function

Assumptions about sell-side analysts' objective function are critical to empirical researchers' understanding of their incentives and the resulting behavior. This paper provides empirical evidence about the objective function underlying analysts' choice of forecasts. In contrast to approaches used in previous papers which rely exclusively on statistical properties of forecasts, I compare theoretical models with alternate objective functions based on their ability to explain observed forecasts. A linear loss objective function which incorporates the effect future analysts' action on analysts' deviation from peer forecasts is best rationalized by the data. I find that assumptions about the objective function has a substantial impact on the conclusions from empirical tests about analysts' incentives and resulting behavior.