The Bias of Wall Street Analysts
Historically, stock analysts’ recommendations have been swayed by business relationships between the analyst’s employer and the target company, says Professor Mark Bradshaw. Have recent SEC reforms helped?
If it's one lesson the individual investor learned the hard way from the collapse of Enron, it is that the recommendations of Wall Street stock analysts can be influenced by much more than purely objective research. Just look at the large number of analysts who kept positive ratings on the company up until the point (and even after) the company tumbled off a cliff.
Indeed, HBS professor Mark Bradshaw and collaborators Scott Richardson and Richard Sloan found that pre-year 2000 forecasts and recommendations done by Wall Street research analysts universally tend to be more optimistic for companies their firms are issuing securities for—or hope to do business with.
The research, "Playing Favorites: Financing Options Sway Analysts' Thinking," was published in the June 2004 edition of Investor Relations Quarterly. Bradshaw discusses the findings and brings us up to date with recent developments.
His work examines how "sell-side financial analysts incorporate accounting information in their earnings forecasts, common stock valuations, and investment recommendations." He also analyzes management reporting of modified GAAP earnings figures to the stock market, and has begun investigating the impact of accounting method choices in cross-border investments.
Ann Cullen: After the scandals of the past few years, the potential for brokerage research deception is obviously something the Securities and Exchange Commission is well aware of and trying to confront with the recent regulations imposed on the industry. But given the amorphous nature of some of the trends you document, do you think these will be effective?
Mark Bradshaw: The entire function of the sell-side analyst is now up in the air, so a brief history should put things in perspective. The large brokerages are working on repairing the tarnished image of the sell-side analyst profession while at the same time reconfiguring the economics of that function. Up until brokerage commissions were deregulated on May 1, 1975 (known in the industry as 'May Day'), the cost of the sell-side analyst function was recouped through a schedule of commission rates on stock transactions by the firm's clients. These commissions were regulated by the exchanges along with the Securities and Exchange Commission. The result was high commission rates and a proliferation of research reports that were funded through the commissions. After deregulation of rates in 1975, commission rates predictably dropped, and the brokerages were left with a fixed cost of sell-side analyst operations and a shortfall in revenues to fund these activities.
Clearly, the brokerages did not get rid of the sell-side analysts at that time, but somehow the cost of keeping them on board has been justified. The existence of an analyst who provides research on a particular company is attractive on several grounds. The primary source of revenue that is arguably related to the sell-side research is investment banking fees. Trading commissions, albeit lower, still exist, but they represent a relatively minor source of revenue for the big brokerages compared to investment banking fees. Either way, the cost of sell-side research, estimated at over $8 billion in total, must be recovered somehow. In the wake of the $1.4 billion conflict of interest settlement, many brokerages such as Bear Stearns, Morgan Stanley, and J.P. Morgan Chase are outsourcing a lot of the number crunching activities of sell-side analysts to places like India, where labor costs are just a fraction of those for junior analysts in the U.S. Thus, the brokerages have signaled that they are not getting rid of these operations.
In addition to the $1.4 billion conflict of interest settlement, the self-regulatory organizations that govern brokerages have issued new rules. The National Association of Securities Dealers (NASD) issued Rule 2711, "Research Analysts and Research Reports," and the New York Stock Exchange (NYSE) issued Rule 472, "Communications with the Public." Both were effective in May of 2002, which is subsequent to the data used in our study, but seem to have noticeably affected the overall optimism of research reports. For example, whereas sell recommendations were virtually unheard of previously, they now constitute 10-20 percent of recommendations. In addition, the rules place limitations on interactions within the brokerage between sell-side analysts and investment bankers, as well as regulations on interactions between sell-side analysts and managers at the firms they cover.
Thus, the effectiveness of recent SEC and exchange regulations appears to have impacted the observed level of optimism. To the extent that overall optimism is more conducive to deception by such analysts, this is a step in the right direction. These subsequent regulations and general trends imply that the impact of investment banking conflicts that we document for affiliated and unaffiliated analysts might have been attenuated, but it is difficult to envision them being eliminated.
The fact remains that the brokerages have to somehow recoup the cost of the sell-side analyst function, regardless of whether the costs are lower due to outsourcing. Even if you completely resolve the conflict of interest between sell-side analysts and the investment banking function, there are other conflicts of interest that persist. For example, the trading commissions that remain are largely driven by big institutional investors who have brokerages execute their trades. Sell-side analysts have incentives to hype stocks to generate trading business through these large investors, which may still have a detrimental impact on the quality of research. Moreover, once a sell-side analyst has prompted an institutional client to take a large position in a stock recommended by the analyst, the analyst faces a disincentive to ever downgrade that stock, as doing so would directly impact the value of the position held by the institutional client.
Q: Your research uncovered that most academic research in the past has benchmarked "affiliated" analysts—or those covering a company that is an existing client—and "unaffiliated" analysts against one another, rather than seeking out commonalities in their behavior. Why is this so?
A: The predictions in prior research were always that the affiliated analysts would be more optimistic than the unaffiliated analysts. This always seemed odd to me because unaffiliated analysts want to be affiliated so they can realize the benefits of associated investment banking fees and other ancillary services such as mergers and acquisition work, private placements, asset financings, and so on. The mixed results of prior research that we summarized in the Investor Relations Quarterly article are not surprising if you believe that unaffiliated analysts want to be affiliated.
This reasoning is why most prior research pitted affiliated analysts against unaffiliated analysts. Clearly, if the choice of brokerages by managers based itself on the optimism of the analyst employed by the brokerage that covers their firm, then it is not surprising that affiliated analysts might be those that are the most optimistic, on average. All else equal, managers would clearly favor hiring a brokerage with analysts that are the most optimistic about the firm's future prospects.
Analysts are persistently overoptimistic in their forecasts for the firms issuing debt and equity securities.
However, the documented differences, when they are significant, between the level of optimism for the affiliated and unaffiliated is hardly impressive, which is perhaps not surprising. If one thinks of the competition to get investment banking business as a race, the analysts that end up being affiliated might have "won the race" by just a nose, with their observed level of optimism being only slightly higher than that of the other competitors. Indeed, many previously documented significant differences between affiliated and unaffiliated analysts' research are statistically significant, but not economically significant. For example, one study found that the difference between affiliated and unaffiliated analysts' forecasts of earnings growth was statistically significant, but the forecasts were 21.3 percent for affiliated analysts and 20.7 percent for unaffiliated analysts. The difference of 0.6 percent is hardly anything to get excited about, especially given the wide variation in forecasted growth levels overall (e.g., 3 percent to 40 percent). We found similar evidence when we partitioned our analysts on the standard affiliated versus unaffiliated dimension. However, the real insight from our analysis was that this differential is minor relative to the differential when we look at the level of financing activities of the covered firms.
Q: You make the point in your research that "the figures clearly demonstrate that analysts' optimism peaks during the months of the external financing activity. It may be that in an attempt to garner investment banking business, analysts have strategically become overly optimistic as they sense that firms are considering capital issuances." Can you elaborate?
A: What we did was to alter the benchmarks. Rather than compare the optimism of affiliated versus unaffiliated analysts, we checked to see whether all analysts (regardless of affiliation status) were more optimistic for firms that were doing things that would generate revenue for the analysts' brokerage. So, we sorted firms on their relative levels of changes in net financing. The prediction was that all analysts would be more optimistic for firms that were issuing debt and equity securities than for firms that were either not engaging in such activity or were reducing their financing needs. An integral part of the forecasting function that analysts perform is assessing the need for external financing to fund operations. Thus, analysts are already intimately aware of what financing needs a company is likely to have. Accordingly, in periods leading up to a need to go to market for financing, we predict that all analysts would tend to increase the level of optimism embedded in their published research. This is exactly what we found.
Even more interesting is that the optimism embedded in analysts' forecasts differs depending on whether the firm is issuing debt or equity securities. For firms that issue securities, their concern is the initial pricing of those securities. The pricing of equity is largely determined by the long-term performance of the firm, whereas the pricing of debt is limited on the upside, with investors standing to recapture principal and receive interest payments. Accordingly, we see that analysts' optimism varies with the type of security being issued. For debt, the optimism is restricted to near-term earnings forecasts (i.e., the next two years); for equity, the optimism is concentrated in longer-term forecasts (i.e., growth forecasts, target price forecasts) and stock recommendations.
In our analysis, it should be noted that the trick to measuring the analysts' optimism is to measure the signed accuracy of the forecasts. To do this, we calculate a forecast error of all analysts' forecasts. For example, for a forecast of one-year-ahead earnings per share, the forecast error would be calculated as the actual earnings per share that ended up being reported minus an analyst's forecast; thus, optimism would be indicated by negative forecast errors. As predicted, we found that all forecast errors surrounding external financing activity were systematically negative and large, and peaked during these periods.
One thing to note is that it is not unexpected that investors and analysts would have optimistic expectations for firms that are growing and need financing. One might argue that such firms (i.e., 'growth' firms) are characterized by greater uncertainty, making it more difficult for analysts to predict the future for these firms. However, there is no reason why analysts would systematically be inaccurate in one direction (i.e., too high) for these firms relative to other firms. However, this is exactly what we found. Analysts are persistently overoptimistic (i.e., too high) in their forecasts for the firms issuing debt and equity securities.
Q: Last summer as part of an SEC settlement, the ten largest U.S. brokerage firms were required to provide clients with a second independent source of research in addition to reports by the firms' own analysts. How do you factor in the role of independent or boutique research firms in your analysis of trends concerning overly optimistic estimate forecasts? Do you think they could have an impact on the skewed behavior of brokerage analysts that your research documents?
A: First, our analysis ended with the year 2000. As you note, a lot has happened since then. Thus, we have seen an increase in the amount of research provided by supposedly independent research firms. Our study reported results based on consensus forecasts and recommendations, which collapse all kinds of analysts into a summary forecast or recommendation. However, we also examined the data using individual forecasts and recommendations, partitioned by whether the analyst worked at a brokerage with investment banking activities or not, finding similar patterns.
Many believe that analysts now spend inordinate amounts of effort in obtaining guidance on what to forecast for the next quarter's earnings.
Although we cannot draw any conclusions about the impact of independent brokerage activity subsequent to 2000, our analysis prior to that time indicates that such brokerages were also characterized by persistent overoptimism. Also, current research in progress by colleagues here at HBS (Amanda Cowen, Boris Groysberg, and Paul Healy) and elsewhere (Brad M. Barber, Reuven Lehavy, and Brett Trueman) confirms that brokerages without investment banking operations are also optimistically biased. The difference is that while brokerages with investment banking operations are optimistically biased consistent with the lure of investment banking and other ancillary fees, brokerages without investment banking activities are optimistically biased due to the lure of trading commissions. In the context of our analysis of firms engaging in a lot of external financing, it is clear that these firms—typically characterized by strong growth—would be the same types of firms that brokerages could most easily expect to generate trading commissions.
Q: Due to the likelihood of inaccuracy acknowledged in your research, do you think there could be a time in the future when investors will disregard stock research entirely? And if they did, what other forms of analysis would they likely find more reliable?
A: This would be a great outcome indeed, but probably for different reasons than intended by the question. The demand for stock research is driven by individual specialization and constraints on time. For example, I am currently teaching Business Analysis and Valuation in the Elective Curriculum, and the students are learning how to do their own stock research. If we are successful in class, these students could indeed be able to disregard stock research entirely in the future. However, because stock research is a time-intensive effort, and we all have limited time, the more likely outcome is that educated investors like our students will consume sell-side stock research, but will be able to do so objectively, identifying inconsistencies or overoptimistic assumptions embedded in that research. The bigger problem is that not all investors are able to rigorously study financial statement analysis and valuation, and given constraints on time, there will always be a demand for the research of experts.
The good news is that there are a number of legitimately independent sources of research; the bad news is that this research is expensive, and there is always the possibility that even legitimately independent stock research will be affected by conflicts similar to those mentioned above, such as a desire to not reverse a previous recommendation.
Q: The Internet's almost real-time access to news [and rumors] has undoubtedly put extra pressures on the expectations of brokerage research analysts. Given these dynamics, do you feel this could be having an effect on their ability to do the thorough research necessary to provide accurate forecasts?
A: Clearly the show-me-now mentality of many investors has hampered the effectiveness of sell-side analysts (and perhaps managers too!). For example, the intense focus on the quarterly earnings reporting season is a relatively recent phenomenon. The idea of a consensus earnings forecast and earnings surprise relative to the consensus was not pervasive until the last decade. Increasingly, firms and analysts have engaged in a game whereby managers attempt to report a quarterly earnings per share figure that just meets or beats the consensus estimate.
Sell-side analysts, on the other hand, know that they can increase their stature and compensation greatly if they achieve recognition on one of the analyst rankings put out by the Institutional Investor or The Wall Street Journal. Given that earnings forecasting accuracy is one component of these rankings, it is clear why analysts seem to devote a lot of effort towards getting the right forecast for the next quarterly or annual earnings announcement. Many believe that analysts now spend inordinate amounts of effort in obtaining guidance on what to forecast for the next quarter's earnings. The belief is that analysts displace efforts away from objective valuations and recommendations; in place of these efforts, analysts may be mechanically issuing optimistic recommendations for all firms to curry favor with managers.
The dynamics of this game between managers, analysts, and investors experienced a shock recently, with declines in the stock market, corporate scandals, and the record analyst research settlement. The near future of the industry will be interesting indeed.