This dissertation examines investors' expectations of loss persistence. Loss persistence varies across firms. I develop a model to forecast loss firms' future earnings based on Joos and Plesko (2005). This model produces smaller forecast errors than random walk models or a model that assumes losses are transitory. Using the forecast earnings generated by the model, I classify losses observations into predicted persistent losses and transitory losses. Predicted persistent losses are loss observations with forecast earnings in the lowest quintile of the quarterly distribution, while predicted transitory losses are loss observations in the highest quintile of the distribution.

Using the Mishkin (1983) framework, I examine investors’ expectations of loss persistence embedded in the stock prices. The results suggest that investors do not fully distinguish the differences in loss persistence captured by the model, and instead appear to assume that all losses are transitory. Consequently, investors are surprised by future announcements of negative earnings for firms with predicted persistent losses, and these firms experience significantly negative abnormal returns over the following four quarters. Consistent with market underreacting to loss persistence of firms with predicted persistent losses, the future negative abnormal returns of these firms are clustered around future earnings announcement dates.

Security analysts and institutional investors are sophisticated market participants. The expectations embedded in analyst forecasts, albeit optimistic, largely capture the difference in loss persistence. As a result, the future negative returns of firms with predicted persistent losses are smaller in magnitude when these firms are followed by more analysts. The future negative returns of firms with predicted persistent losses are also smaller in magnitude when these firms have higher institutional holdings, suggesting that the optimistic bias about loss persistence is concentrated in naive retail investors.

The results are robust to controls for various risk factors and price anomalies. The results are not driven by short sales constraints or influential observations. Finally, the results are not driven by factors related to the new economy industries.