The Rational Modeling Hypothesis to Explain Analyst Underreaction to Earnings News*

Analysts publish earnings forecasts with serially correlated errors. We assess rational versus cognitive limitation explanations for analysts’ underreaction to earnings news. Institutional investor voting for all-star analyst selections reveals whether these investors prefer analysts to issue forecasts with less serially correlated errors. Consistent with it being potentially rational for analysts to underreact to earnings news, institutional investors appear indifferent to serially correlated errors, although votes do indicate a preference for lower total error in earnings forecasts. Further evidence that analysts seem to be behaving rationally when they underreact to earnings news is that, despite consistently underreacting to their most recent forecast error, analyst reaction does a good job of capturing the extent to which earnings news is transitory versus permanent. We further probe the plausibility of our hypothesis by investigating potential explanations for why investors demand analyst forecasts with serially correlated errors.