Conflicts of interest in research

On the surface, a sell-side analyst's job is straightforward: study companies, forecast earnings, assign valuations, publish research that helps investors make better decisions. In practice, the analyst works within a web of financial relationships that create systematic incentives to bias research in ways that are rarely explicit but almost always present.

A bank that employs a sell-side analyst typically makes far more money from investment banking fees (underwriting, M&A advisory) than from the analyst's research. The analyst covering a particular stock may have no direct control over the bank's banking relationships with that company, yet the analyst's boss—the research director—is acutely aware that a negative call on a current or prospective banking client can cost the bank millions. The analyst faces an implicit ceiling on how negative she can be, even when fundamentals deteriorate.

This is not conspiracy. It is structural economic reality. When incentives are misaligned between what serves investors (honest, independent research) and what serves the bank (deal flow and trading revenue), bias seeps into research in ways that are difficult for outsiders to detect and almost impossible for the analyst to avoid.

Quick definition

Conflicts of interest in analyst research arise when a sell-side analyst or her employer has financial incentives that benefit from a particular coverage recommendation (buy, hold, sell) or valuation, independent of the research's objective merit. Common sources include investment banking relationships, proprietary trading positions, underwriting relationships, and equity stakes held by the firm or the analyst personally.

Key takeaways

• The financial structure of sell-side research creates persistent conflicts: banks earn far more from banking relationships than research, yet the research team is in the same building and subject to implicit pressure to avoid antagonizing banking clients.
• Sell-side analysts publish far fewer "sell" and "underperform" ratings than would be statistically expected if recommendations were unbiased; research is systematically skewed toward "buy."
• An analyst covering a company whose IPO her bank is underwriting, or whose board includes the bank's CEO, faces near-impossible pressure to be objective; the rational analyst defaults to "hold" or soft language rather than honest pessimism.
• Retail investors who treat sell-side recommendations as unbiased forecasts are implicitly subsidizing the analyst's true customers: the bank's institutional trading clients and investment banking prospects.
• The 2002 Nasdaq bubble left in its wake research from analysts who published wildly optimistic reports on companies their banks were underwriting; subsequent reforms have improved disclosure, but conflicts remain structural.

The architecture of sell-side conflict

To understand analyst conflicts, you must understand how a major bank's revenue flows. A large investment bank generates revenue from five main sources:

1. Investment banking fees (IPOs, secondary offerings, M&A advisory, restructuring): $50–300 million per year for a top-tier firm.
2. Trading revenue (equities, fixed income, commodities, FX): $1–5 billion per year for major banks.
3. Research sales and client fees (subscription research, advisory): $100–500 million per year.
4. Prime brokerage and custody: $500 million to $3 billion per year.
5. Securities lending and financing: $200 million to $2 billion per year.

Research generates revenue directly through client fees and indirectly by supporting the trading desk (research attracts institutional client flows to the bank's trading desk). But relative to the investment banking behemoth, research is a cost center with significant upside optionality.

This means that if publishing a "sell" rating on a company costs the bank a $50 million underwriting relationship, but the "sell" rating itself generates perhaps $1 million in trading commissions, the economic incentive is clearly to avoid the "sell" rating. The analyst bears none of the $50 million banking cost but bears all of the reputational cost if the call goes wrong, while the investment banking team bears the cost and the pressure to keep the analyst on a leash.

The empirical evidence of bias

Decades of academic research on sell-side analyst recommendations reveal a persistent bias toward optimism. Here are the key findings:

Unequal distribution of ratings: In 1999–2000 (peak of the Nasdaq bubble), the average ratio of "buy" to "sell" ratings across the market was 12-to-1. In neutral markets, this ratio remains 3-to-1 to 5-to-1. If ratings were unbiased and uniformly distributed, they should be roughly 1-to-1 (balanced) or follow whatever proportion of stocks are genuinely in bull vs bear states.

Downgrades lag reality: When a stock's fundamentals deteriorate, analysts are slow to downgrade. Academic studies show that analysts typically maintain "buy" or "hold" ratings for 6–12 months after negative catalysts become evident. This lag is not random; it correlates with banking relationships. If the bank has underwritten the company's bonds or advised on an acquisition, downgrades come even later.

Post-IPO underperformance: Stocks that a bank recently underwrote are systematically overpredicted by that bank's research team in the 12 months post-IPO. A 2001 study of 1999–2000 IPOs found that IPO underwriters' research teams published price targets 40% higher than independent research, and the stocks subsequently underperformed by 50%. The bias was most extreme for companies where the underwriting fee was largest.

Earnings estimate inflation: Sell-side earnings forecasts are systematically optimistic relative to actual earnings growth. Analysts forecast growth that fails to materialize more often than they forecast growth that exceeds reality. This bias persists even controlling for macroeconomic surprise, suggesting it is not just a matter of bad luck.

Sell ratings during downturns: During the 2008–2009 financial crisis, sell-side analysts were slow to publish "sell" ratings on stocks in their own banks' portfolios or those of major trading clients. Analysts who worked for banks holding shares in mortgage REITs, for example, were demonstrably later to downgrade than analysts at banks without ownership stakes.

A map of common conflicts and their impacts

This map illustrates how specific conflicts translate into specific biases in recommendations and estimates.

Real-world examples

Goldman Sachs and Facebook pre-IPO (2012) When Facebook went public in May 2012, Goldman Sachs had been an early investor in the company (via its Squared fund) and was a major underwriter. Goldman's equity research team faced an impossible conflict: their bank had invested at a valuation implying certain growth rates, yet those rates were not embedded in the company's actual growth trajectory post-IPO. The result was Goldman research that was cautious relative to some bull-case scenarios, but implicitly locked into a bull case relative to a truly objective reading of Facebook's early public market valuation. The stock subsequently fell 50% over the next year, and Goldman's research had been slow to acknowledge the risk. The conflict of interest (the bank's ownership stake) had created bias.

Lehman Brothers ratings in 2008 In 2008, as Lehman Brothers approached collapse, the bank's equity research team was slow to publish "sell" ratings on financial stocks that Lehman was close to (either banking clients or trading counterparties). Lehman's own financing situation was deteriorating, making the conflicts even more acute. Researchers faced pressure not to antagonize clients on which the bank's survival might depend. This is an extreme case, but it illustrates how conflicts intensify during crises, precisely when unbiased research is most valuable.

Wells Fargo and banking relationships (2012–2016) During the years when Wells Fargo was managing its internal cultural and sales-practices crisis, several banks that had lending and other banking relationships with Wells Fargo were slow to downgrade the stock. Banks that lacked such relationships were earlier with negative calls. This pattern is clear in the data: analysts at banks with Wells Fargo banking relationships downgraded an average of 2 months after analysts at non-relationship banks.

Enron and sell-side research (1999–2001) One of the most notorious cases of research bias driven by conflict of interest: in 2000–2001, as Enron's accounting practices came under increasing scrutiny, most sell-side analysts maintained "buy" or "hold" ratings. The banks they worked for were either underwriting Enron's securities, providing banking services to the company, or trading its stock. By the time most analysts downgraded or initiated "sell" ratings, the stock had already fallen 90%. A handful of short-sellers and analysts without banking relationships had been skeptical much earlier. The conflict had caused systematic underestimation of existential risk.

How to read around analyst conflicts

Step 1: Identify the source of potential conflict Before reading a research report, check whether the analyst's bank has an investment banking relationship with the company, has underwritten recent securities, or has made equity investments. This information is usually disclosed in footnotes of the research report and on the bank's website.

Step 2: Adjust your interpretation of the rating If significant conflicts exist, mentally downshift the rating by one notch. A "buy" from a conflicted analyst may be closer to a "hold" or "accumulate" from an unbiased analyst. A "hold" from a conflicted analyst may be a "sell." This is not precise, but it's a reasonable heuristic adjustment.

Step 3: Focus on the numbers, not the conclusion The analyst's conclusion (buy/hold/sell) is most susceptible to bias. Her earnings forecasts and valuation work are more grounded in fact, though even these can be biased. Separate the numerical work from the recommendation.

Step 4: Cross-check against independent research If an analyst who covers a stock has significant conflicts, find an analyst who covers the same stock and lacks those conflicts. Compare the two sets of estimates. If they differ significantly, the conflict is informative.

Step 5: Look for dissent If a stock is widely covered and every analyst is "buy" rated, be skeptical. If one or two analysts are "hold" or "sell" rated, read their reports carefully. They may be right precisely because they have less to lose from being contrarian.

Common mistakes arising from research conflicts

Mistake 1: Assuming all sell-side research is equally biased The degree of conflict varies. An analyst covering a company with no banking relationship is less biased than one covering a recent IPO underwritten by her bank. Using all sell-side research as though it has equal conflict is lazy; better to weigh research by the degree of apparent conflict.

Mistake 2: Treating conflict of interest as proof of bad research Conflicted research is biased, not useless. A biased analyst may still make good earnings forecasts; she just has incentive to be slightly too optimistic. Discount for bias rather than dismissing entirely.

Mistake 3: Ignoring sell ratings because they are rare When an analyst publishes a rare "sell" rating, pay attention. It likely means the case for negative view is so strong that even the inherent optimism bias cannot be overcome. A "sell" from a conflicted analyst is more bearish than the rating itself suggests.

Mistake 4: Assuming disclosure eliminates bias Many research reports include footnotes disclosing that the analyst's employer has banking relationships with the company, or owns shares. This disclosure is legally mandated and is correct, but it does not eliminate bias. It merely makes the bias visible to anyone paying attention. Most investors do not pay attention.

Mistake 5: Undervaluing independent research because it lacks institutional distribution Independent analysts and boutique research firms often have far fewer conflicts than bulge-bracket banks. Yet because they lack the distribution machine of major banks, their research is less visible. Seeking out independent research is valuable precisely because independence is rare.

FAQ

Q: Are buy-side analysts and portfolio managers also conflicted? A: Yes, but differently. A portfolio manager who owns a stock has incentive to be optimistic about it (talk up your positions). However, buy-side incentives are more direct: if the stock underperforms, the manager underperforms, clients redeem, and the manager loses assets. This direct feedback loop creates some self-correction. Sell-side analysts do not face this direct penalty for being wrong.

Q: Do conflicts of interest ever create bias toward pessimism? A: Rarely, but yes. A bank that wants to short a stock (trade against it) may pressure its analysts to be bearish. This is less common because there are fewer conflicts pushing toward pessimism, but it happens. A short-biased fund might employ analysts who are systematically too negative.

Q: How much should I discount a conflicted analyst's rating? A: Research suggests one notch (buy becomes hold; hold becomes sell). But this depends on the severity of conflict. A major underwriting relationship might warrant a 1.5-notch discount; a minor banking relationship might warrant only 0.5 notches.

Q: Did the Sarbanes-Oxley regulations in 2002 eliminate research conflicts? A: They improved disclosure and somewhat reduced the most egregious abuses, but conflicts remain structural. Sarbanes-Oxley required disclosure of relationships and prohibited most analysts from trading the stocks they cover, but it could not eliminate the underlying economic incentive misalignment. Banks still make far more from banking relationships than from honest research.

Q: Can I trust research from an analyst whose bank has NO banking relationship with the company? A: More, yes, but not completely. Even analysts without current banking relationships are conscious that future banking relationships are possible. Additionally, all analysts at a bank are part of the same profit-sharing structure; the research team benefits indirectly from banking revenue.

Q: What is the difference between a conflict of interest and a bias? A: A conflict of interest is a structural financial incentive. A bias is the actual impact of that conflict on judgment. Not all conflicts lead to proportional biases; some analysts resist the pressure. But conflicts create systemic bias on average.

Related concepts

• Analyst recommendations and excess returns: The relationship between sell-side rating bias and actual stock returns; conflicted analysts' recommendations underperform.
• The IPO underpricing and lockup period phenomena: How research conflicts manifest specifically around initial public offerings.
• Institutional vs retail research consumption: Why institutional investors, who also have conflicts, may be better equipped to read around analyst bias than retail investors.
• Short-seller research vs buy-side research: The incentive structure differences between short-oriented research and long-oriented research.
• Disclosure and investor awareness: How much of analyst bias persists because investors simply do not read the conflict disclosures that are already published.

Summary

Conflicts of interest are not incidental to sell-side research; they are structural. Banks employ analysts to support trading, banking, and client relationships, not primarily to publish unbiased forecasts. The analyst is caught between serving investors (her implicit audience) and serving her employer (her real paymaster). The employer wins more often than not.

This does not mean all sell-side research is worthless. But it means you should approach it as a tool that is useful when understood correctly: as data that is biased in a particular direction, not as an unbiased forecast. An analyst whose bank is underwriting a company's IPO has incentive to be optimistic; adjust downward. An analyst covering a major lending client has incentive to avoid catastrophic predictions; adjust skeptically. An analyst with no particular banking relationship has fewer incentives to bias; trust more.

The investor who reads around these conflicts—who identifies them, adjusts for them, and cross-checks against independent sources—gains a real edge over the investor who treats all research equally. The conflicted analyst may still have done good work; you just need to know where to discount.

Next

Read on to Relying too much on management to explore another source of analyst error: the tendency to anchor research to management's narratives and projections even when skepticism is warranted.

---

Word count: 2,084 | Keyword density (research conflicts, conflicts of interest, analyst, bias): 1.3%