Analysts' Conflict of Interest: How Are IPOs Affected?

Jason Draho, IPOadvisory

The recent settlement between the New York state Attorney Generals' office, the SEC, NYSE, and Nasdaq with a dozen major investment banks brings to a close a rather dark period in the history of Wall Street. The impetus for the initial investigation was the allegation of conflicts of interest plaguing research analysts at investment banks. Analysts had the nominal role of producing research reports and recommendations for buy-side clients. In reality, much of their time was spent trying to win new investment banking business, often with the enticement of positive research coverage. An obvious conflict arises when the analysts issue positive recommendations to appease sell-side clients on the one hand, but also give an honest and accurate recommendation for buy-side clients. The incentives are further distorted when the bulk of analysts' compensation is tied to the amount of underwriting business they can help generate.

The settlement between the regulators and the investment banks requires reforms that many hope will resolve the conflict of interest problem. Investment banks are to create a clear "Chinese Wall" between research and banking, analysts' compensation will not be linked to investment banking performance, and analysts can not participate in the pitch to win new investment banking business. In addition, the banks must pay into a fund to generate independent research that must be offered to clients, along with their own recommendations. Many observers have said that the reforms do not go far enough in resolving the underlying conflicts of interest. Only a complete separation of research and banking will suffice. An evaluation of the reforms and their likely effect on research quality requires an understanding of just how good or bad analysts' recommendations have been and how it might change in this new environment.

IPOs played a particularly prominent role in the conflict of interest investigation for two reasons. First, underwriting IPOs is a very lucrative business and the competition to win new clients is fierce. Analysts did have a significant role in winning new business by being part of the initial pitch to the company. Companies often chose a particular bank as the lead manager based on the reputation of the analyst who would be assigned to follow the newly issued public stock. It is easy to see how the financial pressure on analysts to issue biased research and recommendations for IPOs would be overwhelming. The second reason why IPOs figure prominently is that many newly public companies were the subjects of the email messages that formed the basis for investigations by the regulating authorities. It is now clear that analysts at Merrill Lynch, and potentially other banks as well, made negative and derisive comments in internal messages about companies on which they had public buy recommendations.

A brief retelling of the history of analyst research reveals how the investment banking industry evolved to produce the conflict of interest problem of today. Equity research departments within investment banks are self-supporting as long as the incremental trading profits generated by the research cover the cost of operation. When trading commissions were high, research could pay for itself. The decline in the real value of commissions that began in the 1970s has shrunk the trading revenues for these companies. At the same time the cost of running a respectable research department at a top tier bank has ballooned to $500 million, or even a $1 billion, a year. During this period, underwriting and advisory work has grown into a dominant source of revenue, and often subsidized the cost of research. A gradual and subtle shift in the focus of the research occurred-- from being exclusively geared towards buy-side investors to appeasing sell-side clients. Once this happens, the potential for conflicts of interest is unavoidable.

Who Can You Trust?

The crux of the conflict of interest scandal is the claim that analysts' issued biased research and recommendations on the stocks that they covered. Anecdotal evidence clearly implies a bias, but neither the regulators nor the media have provided strong empirical evidence demonstrating a consistent bias in analysts' research coverage. A large body of research on analysts' recommendations and reports, much of it pre-dating the bubble period of the late 1990s, can help us to sift through the various allegations to get a truer picture of the conflict of interest problem.

Analysts produce two quantifiable pieces of information, recommendations to buy or sell and forecasts of future earnings. An important distinction must be made between affiliated and non-affiliated analysts. A conflicted analyst is much more likely when he is covering a company that also has a relationship with the corporate finance department of his bank. The analyst in this case is affiliated because his bank is underwriting the company's IPO. Consistent differences in the recommendations and research of affiliated versus non-affiliated analysts are a good indication of a conflict-induced bias.

At first glance the pattern of recommendations paints a pretty strong picture of biased coverage. Affiliated analysts are more likely to issue a buy recommendation and begin coverage sooner after the IPO, as compared to non-affiliated analysts. An affiliated buy recommendation is often used to provide a 'booster shot' to companies whose stock price is lagging. The stock return immediately prior to an affiliated buy is much more likely to be negative than that of a non-affiliated buy. The stock returns for the one-to-two year horizon after the IPO are a further indication of an affiliated bias. Overall, stocks with affiliated buys significantly under-perform relative to those with non-affiliated buys. On the whole analysts have a better performance record for recommending stocks with which they are not affiliated, suggesting (but not proving) that they are less discriminating in recommending affiliated stocks.

The primary difficulty in concluding that analysts are indeed biased is that in general analysts are too optimistic about the stocks they cover. This positive bias is the consequence of a combined under-reaction to negative news and an over-reaction to positive news. For example, analysts often do not lower earnings forecasts sufficiently when a company issues a warning, but extrapolate a recent positive earnings result into the future without accounting for mean-reversion. Earnings forecast errors for IPOs consistently show positive bias. However, much of the bias is attributable to optimism about the IPOs' industry as a whole. After all, companies go public when investors and analysts are most optimistic about the industry's prospects.

Analysts research coverage also includes forecasts for current and next year earnings and forecasts for long-term growth rates in earnings. The difference in earnings forecasts between affiliated and non-affiliated analysts is positive, but the difference is not as clear-cut as for recommendations. Two factors could account for this difference in measured bias. First, earnings forecasts are difficult to fudge and provide an objective test for analyst accuracy; evaluating the quality of recommendations is much more subjective. Second, recommendations are discrete in nature. Analysts are effectively limited to initiating coverage with a buy or a hold, giving an affiliated analyst little choice but to issue a buy. The finer partitioning of earnings forecasts should reduce the upward bias unavoidable with recommendations.

Analysts' optimism increases with the horizon of the earnings forecast. A positive bias in the projected long-term growth in earnings may have less to do with a conflict of interest than self-selection. Analysts tend to initiate coverage on stocks that they like, and consequently will provide a buy recommendation. A similar logic may explain the positive bias of affiliated analysts for IPOs. An investment bank making its pitch to a company during the "bake-off" will offer a target valuation for the shares. The price is based on the projected long-term earnings of the company. The bank selected to be lead manager is likely to be the most optimistic about the company's prospects. The initial positive recommendations and forecasts of affiliated analysts could be due to a self-selection bias that is reversed after new information comes to light. The stock returns of a company with an initial affiliated buy, later reversed to a sell, shows no signs of under-performing non-affiliated buy recommended stocks. Rush to judgment, rather than conflict of interest, may account for affiliated analysts recommendations.

The evidence on analysts' recommendations and forecasts for IPOs leads to three general conclusions. First, affiliated analysts do exhibit a bias in their recommendations that is probably due to possible conflicts, but a self-selection bias could also contribute. Second, analysts will, on average, have a positive bias in the stocks they recommend, even if research departments are completely separate from investment banking, again because of self-selection. Third, investors can still get valuable information from analyst recommendations and forecasts. However, it is important to look at all the information on a stock collectively, and to understand the incentives of the different analysts.

A major caveat to these conclusions is that this analysis was conducted with data prior to the mania of the late 1990s. The usefulness of the findings should not be discounted because the data is out-of-date. It is probably safe to conclude that the biases became even more extreme during the stock market bubble. Reports that there were approximately 4,000 buy recommendations compared to 50 sells lend credence to this conjecture. In the post-mortem of the bubble it appears that analysts have returned to providing more balanced coverage of stocks, no doubt partly motivated by investigations into their activities. If that is the case the findings based on the older data may still be applicable to the current environment. Whether the bubble period was an aberration will only be resolved with additional analysis of the data.

Breaking Up is Hard to Do

The settlement with the major investment banks requires a clear and enforced separation between the research department and investment banking. Critics have argued that the reforms do not go far enough. As long as analysts are under the same corporate umbrella as banking they will be influenced by a culture focused on winning new sell-side clients, even if there is no contact between the two groups. Disagreements about how best to reform the banking industry are likely to continue as the required changes are implemented. Lost in the debate about reforms is any rigorous analysis of whether retail investors, the group the regulators have been most concerned with, will truly be better off under any of the alternative recommendations.

Analyst research consists of more than just recommendations and forecasts. Analysts possess tremendous industry knowledge and expertise, and their ongoing research reports on companies contain nuggets of information that are useful to the public in arriving at a valuation for the stock. If research departments were forced to operate independently from investment banks, and prove to be not viable economically, the result will be a reduction in the production of information.

Investors and companies would suffer directly from a reduction of analyst research for two reasons. First, research produced by analysts and distributed to investors will be impounded into the stock price through the course of regular trading. The more informative the stock price is about company value, investors will be less concerned with trading in a mispriced stock. When information is abundant, trading volume increases, and with it liquidity. A direct measure of the increased liquidity is a narrower bid-ask spread and a lower cost of equity capital. Investors benefit from analysts' research by paying lower transaction costs, and companies have a lower cost of capital, and a higher valuation. The second cost to less analyst research is the reduced monitoring of companies and their management. The production of research has proven to be an effective tool for evaluating a company's prospects, and can help to spur on management to improve performance. Without the publicity of analyst recommendations, managers will face less external discipline, which could lead to poorer decision making and lower company value.

The debate on how best to reform Wall Street to avoid future potential conflict of interest scandals is likely to be ongoing. Any proposed reform must be evaluated based on both costs and benefits to investors. The pre-settlement institutional structure suffered from conflicted analysts, which hopefully will be resolved with the reforms. However, investors know now, if they did not know before, how and why biases arise and how to adjust for them. Analysts also produce a great deal of valuable information. Any further structural reforms may resolve the conflict of interest problem, but could also lead to new, unknown problems. Sometimes it can be better to stick with the devil whose tricks you know and understand, rather than switch to the one you don't.

About The Author: Jason Draho founded IPOadvisory, an educational service on IPOs that provides investors access to the information produced by the most up-to-date academic research. He has written many articles on IPOs, and is author of the forthcoming book IPOs: Motivation, Preparation, and Performance.