Analyst Rankings And Forecast Informativeness

Sanjay Banerjee
University of Alberta – Department of Accounting, Operations & Information Systems

June 30, 2016

Abstract:

Does competition for a higher reputational ranking induce analysts to reveal more information? Two analysts make simultaneous forecasts about a company’s future value to achieve a higher reputation relative to their peers, and are paid a tournament-style compensation. I show that the effect of reputational rankings on the information content of analysts’ forecasts depends on the convexity of their payoffs and the correlation between their private signals. Competition diminishes forecast informativeness under a convex payoff structure when their signals are positively correlated and the market uncertainty is high. Under a convex payoff, analysts expecting a positive signal correlation tend to differentiate from each other by forecasting against their own signals, revealing less information in equilibrium. For a concave payoff, competition improves forecast informativeness. Forecasts by more reputable analysts may not be more informative under a convex payoff structure. These results have important implications for Regulation Fair Disclosure.

Analyst Rankings And Forecast Informativeness – Introduction

“The real money comes from a pool put together by portfolio managers. They vote on the value of our stuff relative to that of other analyst teams, and allocate a proportionate share of the pool. It’s a tournament reward structure where the top takes in a lot.” – Equity analyst at a major bank (The Guardian, June 11, 2012).

Empirical evidence suggests that sell-side analysts are concerned not only about their own reputations, but also about their reputations relative to their industry peers. One of the top two components of sell-side analysts’ compensation is their rankings1 by portfolio managers at the buy-side firms (e.g., Groysberg et al., 2011; Brown et al., 2015). The pay differential between a top-ranked analyst and his non-ranked counterparts can be substantial. While Wall Street no longer has the late 1990’s Jack Grubman-type super star compensation,2 pay in the top percentile can be as high as 6 times the pay in the bottom percentile at leading investment banks (e.g., Groysberg et al., 2011). Given the prevalence of relative rankings among sell-side analysts and the tournament reward structure in equity research, a few natural questions emerge: How does a tournament compensation structure affect sell-side analysts’ forecasting behaviors? Does this reward structure encourage or discourage analysts to reveal their private information? It is well known that tournaments are useful in optimal contracting when there is a common shock that affects the performance of the entire peer group (e.g., Lazear and Rosen, 1981; Green and Stokey, 1983). What is the role of correlation among analysts’ private information in the presence of a similar tournament among analysts? What are the regulatory and policy implications?

I study how analysts’ rank-ordered reward structure affects analysts’ forecasting behaviors, and specifically, the information content of their forecasts. I show that, contrary to common intuition, competition for a better ranking can diminish, rather than improve, analysts’ forecast informativeness. When analysts’ payoff structure is convex in their reputational rankings and their signals are expected to be positively correlated, they tend to differentiate from each other by forecasting against their own signals, thereby communicating less of their private information in equilibrium. The loss in forecast informativeness is the highest when the market uncertainty is high, which is precisely when analysts’ research is most useful (e.g., Brown et al., 2015). This is a stark prediction, because intuitively, competitive pressure tends to improve the information content of an analyst’s forecasts. Competition has long been believed to improve service quality, reduce bias, and encourage innovation (e.g., Gentzkow and Shapiro, 2008; Hong and Kacperczyk, 2010). In contrast, I show that competition can decrease forecast informativeness.

I consider a communication game in which two analysts privately observe a noisy signal about the future value of a company, and simultaneously make forecasts about the future value to a decision maker (a representative investor, the “market”). The decision maker uses analyst forecasts to make investment decisions. Analysts’ signals may be positively correlated. The accuracy of an analyst’s signal depends on his type, good or bad. A good analyst receives a more accurate signal than a bad one. Neither the analysts nor the market know an analyst’s type with certainty. An analyst’s reputation is the market’s belief about his type. Once information related to the future value of the company is publicly disclosed, the market uses analysts’ forecasts to update an analyst’s reputation. Analysts have a tournament reward structure which depends on the rankings of their reputations. An analyst’s payoff structure is convex if the prize for being ranked higher is greater than the penalty for being ranked lower; the opposite is true for a concave payoff structure. An example of a convex payoff is a situation in which being right when everyone else is wrong substantially increases an analyst’s number of clients (e.g., Laster et al., 1999). Winner-take-all is an extreme case of a convex payoff structure. A concave payoff arises when an analyst is fired for having a sufficiently poor reputation compared to his peers.

Analyst Rankings, Forecast Informativeness

See full PDF below.