Understanding Analysts Earnings Expectations PDF Download

This thesis examines the determinants of analysts' reactions to firms' earnings management. I present a model showing that analysts revise their forecasts according to their forecast errors revealed by earnings announcements and reporting biases embedded in reported earnings. The model further demonstrates that the relationship between forecast revisions and reporting biases can be affected by analysts' forecasting ability, the inherent uncertainty of whether reporting biases have occurred, as well as analysts' incentives. To empirically test the model's prediction regarding analysts' forecasting ability, I use analysts' firm-specific experience, size of their brokerage firm, and the number of industries they follow as proxies. Consistent with the model's prediction, I provide evidence showing that well-experienced analysts adjust more for earnings management while analysts following a greater number of industries adjust less for earnings management. Sensitivity analysis using analyst's historical firm-specific forecast accuracy as an alternative measure of forecasting ability further supports the hypothesis that analysts with better forecasting ability adjust more for earnings management. Moreover, analysts adjust less for earnings management when the inherent uncertainty of the reporting bias is greater. Specifically, analysts adjust less for earnings management when: (1) the past volatility of discretionary accruals is high; and (2) the firm has a marked propensity to smooth earnings. There is little evidence that affiliated analysts adjust less for earnings management than unaffiliated analysts. However, analysts adjust more for earnings management in the post-Reg FD period than in the pre-Reg FD period, which is consistent with Regulation FD achieving its objective of strengthening analysts' incentives to issue unbiased forecasts.