Separating Information About Cash Flows from Information About Risk in Losses

Prior literature interprets the weak earnings response coefficient (ERC) of accounting losses as a manifestation either of lack of forward-looking information in losses or of market mispricing of losses. Based on return decomposition theory, I predict that losses contain information not only about future cash flows (i.e., cash flow news) but also, about risk (i.e., expected returns and discount rate news). However, these informational components have offsetting valuation effects, resulting in a muted ERC. Consistent with the prediction, I show that, after controlling for information about risk (mainly expected returns), the ERC of losses becomes statistically significant with more negative returns for larger losses when returns are measured either annually or around earnings announcements. Moreover, loss firms will continue to have poor future earnings and operating cash flows, and larger losses are associated with more negative analyst forecast revisions in the loss-reporting year. I also document that losses provide more negative cash flow information when they are not because of research and development expensing, when they trigger operational curtailments, and when they are less likely to reverse to profits. Further tests confirm the robustness of my findings to considering future return drifts/reversals, alternative proxies for expected returns and discount rate news, alternative test portfolios, and alternative model specifications. Overall, my paper provides new insights into the information content of losses. This paper was accepted by Suraj Srinivasan, accounting.