Do earnings estimates add value to sell-side analysts' investment recommendations?

Abstract

Sell-side analysts change their stock recommendations when their valuations differ from the market's. These valuation differences can arise from either differences in earnings estimates or the nonearnings components of valuation methodologies. We find that recommendation changes motivated by earnings estimate revisions have a greater initial price reaction than the same recommendation changes without earnings estimate revisions: about +1:3% (-2:8%) greater for upgrades (downgrades). Nevertheless, the postrecommendation drift is also greater, suggesting that investors underreact to earningsbased recommendation changes. Implemented as a trading strategy, earnings-based recommendation changes earn risk-adjusted returns of 3% per month, considerably more than non-earnings-based recommendation changes. Evidence from variation in firms' information environment and analysts' regulatory environment suggests that recommendation changes with earnings estimate revisions are less affected by analysts' cognitive and incentive biases.