Restatements, Managerial Learning, and Optimal Reporting Bias

Abstract

Financial restatements are costly, but frequent, events and many firms restate several times. This paper asks why rational managers engage in misreporting, in spite of the costly consequences. We present a simple extension to the Fischer and Verrecchia (2000) model, which provides testable predictions regarding the relation between the cost of restatements and reporting bias. In our model, the observed cost of a restatement conveys information to the manager about the true reputation cost of biasing financial statements, which he incorporates into his optimal choice of bias. A restatement hence offers the manager an opportunity to learn about the true cost of reporting bias, and he only updates his biasing strategy if the observed cost differs from his expectation. Consistent with this prediction, we find that firms reporting bias is decreasing in the reputation costs observed after a restatement, such that managers only reduce future bias if the observed cost of restating was larger than expected. This effect seems stronger when we allow firms to update the expected cost of restatement based on more restatements. Our results indicate that rational managers use the insights from prior restatements to improve their future bias strategy.