Courtesy of Charles Chao Kang, Kenneth J. Merkley, Roni Michaely, and Joseph Pacelli
The Global Analyst Research Settlement (“Global Settlement”) was a historic 2003 enforcement agreement between the SEC, NASD, NYSE and ten of America’s largest investment firms, aimed at addressing problems of conflict of interest within investment firms. In our recent article, we use the passage of the Global Settlement to show that disclosure requirements about analysts’ recommendation distributions incentivize analysts to manage their recommendation distributions to reduce concerns about perceived objectivity. Following the regulation, analysts frequently issue sell recommendations concurrently with buy recommendations, consistent with recommendation distribution management. Analysts’ propensity to provide concurrent sell recommendations increases as recommendation distributions deviate from historical benchmarks and when public scrutiny and visibility is high. Importantly, we find no evidence of such behavior prior to the Global Settlement, thus suggesting that the behavior is an unintended consequence associated with the regulation. This behavior has important implications for investors: sell recommendations issued concurrently with buy recommendations provide a weaker investment signal, as they exhibit muted return reactions and are less likely to be supported by downward earnings forecast revisions. Overall, our results highlight the importance of analysts’ distributional incentives in influencing the quality of recommendations.
Background
Security analysts are important information intermediaries in capital markets and numerous studies find that their recommendations have investment value. At the same time, analysts face significant incentives to optimistically bias their reports, and issue disproportionally more buy recommendations, thereby limiting the informativeness of their research (e.g., Michaely and Womack, 1999; Jegadeesh et al., 2004; Mehran and Stulz, 2007; Hong and Kacperczyk, 2010; Corwin et al., 2017). One way investors can assess analysts’ objectivity is by considering the distribution of an analyst’s recommendations. If an analyst’s portfolio contains “too few” sell recommendations, investors might conclude that this analyst’s reports are optimistically biased and discount the corresponding buy recommendations. This concern, in turn, can incentivize analysts to issue more sell recommendations to achieve greater portfolio balance and maintain credibility with investors.
Following the financial scandals in the early 2000s, achieving a more balanced and less biased portfolio of stock recommendations has become an important regulatory objective. New regulations (e.g., Global Settlement and FINRA 2241) now mandate that brokerage houses prominently disclose the distribution of outstanding recommendations in every analyst report. The objective behind this disclosure requirement is to improve the balance of analysts’ individual recommendation distributions and reduce optimism bias. This objective has at least partially been achieved: in aggregate, a buy to sell recommendation ratio of about 39:1 at the height of the tech bubble decreased to 6:1 after Global Settlement (based on data from our sample).
However, this requirement in turn may have strengthened the aforementioned incentives for analysts to “manage” their recommendation distributions in order to maintain a perception of credibility with investors. We examine how such distributional incentives affect analysts’ decisions to issue sell recommendations and whether acting on these incentives affects the informativeness of their recommendations. Because analysts likely face added scrutiny about having portfolios that are heavily tilted toward buy recommendations, our central prediction is that analysts issuing buy recommendations are more likely to concurrently provide sell recommendations to limit changes in distributional balance—even if such recommendations are less than fully justified. As these sell recommendations may be issued for reasons that are not purely based on expected future performance, we consequently expect such sell recommendations to be less valuable to investors.
Main Findings
We begin our analyses by considering the effects of the Global Settlement. We obtain stock recommendation data for all U.S. analysts from I/B/E/S. Employing the Global Settlement as a quasi-exogenous shock to analysts’ distribution concerns, we examine changes in the practice of issuing sell revisions concurrently with buy revisions. We find evidence of a significant increase in concurrent buy and sell recommendation issuance in the period immediately following the Global Settlement when disclosure requirements provide analysts greater incentives to manage their portfolio distributions. In contrast, we find no evidence of a positive association between concurrent buy and sell recommendation issuance in the period preceding the Global Settlement.
Given the importance of distributional incentives in the post-Global Settlement regime, we focus the rest of our analyses on analysts’ recommendation revisions in the period spanning from 2004 to 2015. We find that during this period, when an analyst issues a buy recommendation in a given month, the likelihood of the analyst issuing a sell recommendation in the same month increases by approximately 38% relative to the sample mean.
We next strengthen our findings by providing important empirical support for our assumption that having a more optimistically-weighted portfolio imposes costs on analysts. To do so, we examine whether the market’s response to buy recommendation upgrades varies with the average recommendation rating in an analysts’ portfolio (where a higher average recommendation rating suggests a more optimistically-weighted portfolio). We find that the market responds less to buy recommendations when analysts’ portfolios have a higher average recommendation rating, even after controlling for analyst and industry-month fixed effects. This suggests that investors place less weight on analysts’ recommendations when they maintain a more optimistically-weighted portfolio, which creates incentives for analysts to manage their portfolio distributions to reduce the appearance of optimism bias.
To further support this interpretation, we examine how the results vary with other factors connected to analysts’ incentives. We predict that portfolio concerns likely become more significant as analysts’ distributions become more positively tilted. Consistent with our expectations, we find that analysts are almost twice as likely to issue a sell recommendation concurrent with a buy recommendation when they deviate from historical ratios. We also predict and find that concurrent sell recommendation behavior is more pronounced for more visible analysts (i.e., All-Star analysts) and analysts working at larger brokerage houses, who likely face the most scrutiny and pressure to maintain a balanced distribution.
Building on the finding that analysts are more likely to issue sell recommendations concurrently with buy recommendations, especially when distributional incentives are higher, we conjecture and find that analysts’ conviction about such concurrent sell recommendations are lower. First, concurrent sells are about 23% less likely to be associated with downward earnings forecast revisions relative to the unconditional sample mean, which suggests that these recommendations contain significantly less negative performance expectations. Second, we also find that the short-horizon market response to sell recommendations issued concurrently with buy recommendations is about 21% less negative (i.e., more muted stock price reaction) than the response to standalone sell recommendations, relative to the unconditional sample mean. Overall, these findings support the interpretation that such sell recommendations are likely influenced by analysts’ distribution incentives as opposed to pure evaluations of performance expectations.
Conclusion
Our study provides one of the first investigations into the role of analysts’ distributional incentives. We contribute to the literature by providing evidence on how incentives across an analysts’ portfolio can lead to biased recommendations. Whereas prior studies focus on firm-level incentives (i.e., investment banking and trading pressures), our evidence indicates spillover effects across covered firms in an analyst’s entire portfolio. We also uncover an important incentive for analysts to issue sell recommendations; namely, to rebalance the distribution of their recommendations. Finally, these findings are also relevant to regulators and investors as recent disclosure rules require brokerage firms to display the distribution of outstanding recommendations in their research reports. Our study adds to several prior studies (e.g., Barber et al., 2006; Kadan et al., 2009) showing that the Global Settlement changed analysts’ recommendation distributions. We further this literature by demonstrating how the disclosure requirement introduced by the Global Settlement incentivizes analysts to manage their recommendations distributions, which has the unintended consequence of reducing the informativeness of sell recommendations.
