Consensus earnings estimates, prominently quoted at major media outlets such as CNBC, Bloomberg, and the Wall Street Journal, play an important role in capital markets by providing investors with a proxy for earnings expectations. The common belief is that they are formed by simply taking an average of all analyst forecasts. However, to what extent is this true? Unsophisticated investors know little about the process, while sophisticated investors typically know only what consensus vendors such as Institutional Brokers’ Estimate System (I/B/E/S), whose estimates CNBC uses to identify earnings misses, state in their marketing materials. Specifically, I/B/E/S claims it “removes’ stale estimates (i.e., those that are not updated after a material event like an earnings announcement) and those that use a different accounting basis than the majority of analysts covering the company (majority basis).
In a recent paper, we conduct the first large sample empirical test of the determinants that I/B/E/S uses in removing an earnings forecast from the consensus. We find that I/B/E/S’ marketing materials are materially misleading about the process it uses to form the consensus. While I/B/E/S states that it removes stale estimates, it retains about half of such estimates in the consensus. The system is similarly selective about the enforcement of the majority basis policy. In private communication, I/B/E/S acknowledges that, while its marketing materials suggest that it uses an objective standard to assess which estimates to include, in actuality, it makes a subjective determination about the quality of the estimate.
Interviews with both investor relations officers and I/B/E/S reveal that, in reaching the subjective determination, I/B/E/S uses input from companies. Given firms’ incentives to maintain a pessimistic consensus in order to meet or beat the benchmark, this reality raises the question of whether firms are selectively supplying information to cause I/B/E/S to remove optimistic estimates from the calculation of the consensus. We find evidence consistent with this conjecture. Controlling for the “rules” stated in marketing materials, we find strong evidence that I/B/E/S systematically removes optimistic estimates. The likelihood of removing an optimistic estimate increases about 10-fold when removing the estimate would enable firms to meet-or-beat. The evidence highlights that these removals enable firms to accomplish their reporting objectives. Furthermore, we find that these effects are more pronounced when managers’ incentives to just meet or beat the consensus are stronger (i.e., higher subsequent insider sales or higher compensation sensitivity to stock prices), or managers have greater ability to influence I/B/E/S.
In additional analysis, we examine the effect of this subjective removal policies on consensus accuracy, a measure of the information quality of the consensus. While we might expect accuracy decreases because firms are selectively supplying information, we find these removals actually improve accuracy. Our explanation is that this finding is consistent with firms’ selectively supplying information about optimistic estimates when those estimates themselves are of low quality. Consequently, the provision of such information allows I/B/E/S to deliver a high quality consensus to its clients by removing the low quality estimates and, as a byproduct, enables companies to accomplish their reporting objectives (i.e. meet or beat the consensus). Our explanation for the accuracy improvement, even when firms have incentives to be opportunistic, is that repeated interactions between firms and I/B/E/S likely constrain managerial opportunism and encourage managers to help identify low-quality optimistic estimates for removal, resulting in a high-quality consensus.
This post comes to us from professors Zachary Kaplan and Xiumin Martin at Washington University in St. Louis’ John M. Olin Business School and Yifang Xie at the University of British Columbia’s Sauder School of Business. It is based on their recent paper, “Truncating Optimism,” published in the Journal of Accounting Research and available here.