Is sell-side research more valuable in bad times?

In bad times, uncertainty is high, so that investors find it more difficult to assess the prospects of the firms they invest in. Learning models suggest that in such times investors should, everything else equal, value informative signals such as analyst forecasts and recommendations more than in good times. However, the higher uncertainty in bad times and career concerns stemming from troubled employers may make the task of analysts harder, so that analyst output is noisier and hence less valuable in bad times. Consequently, whether analyst forecasts and recommendations are more valuable during bad times is an empirical matter. We examine a large sample of analyst output from 1983 to 2011. We find that analysts work harder in bad times, but their earnings forecasts accuracy is worse and that they disagree more. Despite more inaccurate earnings forecasts, revisions to earnings forecasts and stock recommendations have a more influential stock-price impact during bad times as predicted by a learning model.