Analyst research: the incentive behind the recommendation
Documented sequence. In 2003, a US federal court approved a settlement involving major securities firms and analyst conflicts. The SEC alleged that investment-banking influence created conflicts in research that firms had not adequately managed. The defendants settled without admitting or denying the allegations. The settlement included financial payments and structural measures intended to separate research from investment-banking influence. This case describes that historical action; it is not a statement of the rules currently applicable to every research provider.
SEC · Federal Court Approves Global Research Analyst Settlement, 2003
Original analysis: two customers. Imagine a firm serving both investors who read research and companies that pay for capital-market services. These activities can coexist, but the audiences may want different things. An investor needs a candid assessment of price and risk. An issuer may prefer enthusiastic coverage. An analyst can face pressure without receiving an explicit instruction to change a conclusion: promotion, access, coverage assignments and remuneration can all signal what is welcome.
The useful question is therefore not whether a report looks professional. It is how the report was produced. Who controls the analyst’s budget? Who can change coverage? Is a negative conclusion publishable? Are assumptions and conflicts visible? These questions concern the conditions under which evidence reaches the reader. They do not allow us to infer that a particular recommendation is false merely because the firm has another commercial relationship.
Hypothetical worked comparison. Two analysts estimate the same company will generate £10 million of sustainable annual free cash flow. One uses a 10% required return and no growth, giving a simplified £100 million perpetuity value. The other uses 8%, giving £125 million. The disagreement can arise without different operating forecasts. A polished headline target hides a 25% valuation difference caused entirely by the discount-rate assumption. The model ignores debt and other complications so the reader can inspect one moving part.
An independent reading separates three layers: reported facts, forward assumptions and recommendation. Historical revenue can be checked against filings. A forecast requires examination of demand, margins and reinvestment. A recommendation adds price, timing and the author’s judgement. Even if the final recommendation is unpersuasive, some factual work may remain useful. Conversely, an accurate historical table does not validate an optimistic forecast.
A common misunderstanding is that disclosure alone makes every conflict harmless. A disclosed incentive is easier to evaluate, but the reader still needs the evidence and may lack the ability to verify it. Another misunderstanding is that institutional research is inherently worthless. The case supports scrutiny of conflicts, not indiscriminate rejection. The practical response is to obtain independent inputs and preserve the full record of predictions, including revisions and failures.
Takeaway. Ask what would happen inside the organisation if the analyst’s conclusion disappointed a commercially important client. The answer should involve actual decision rights and accountability, not only a promise of integrity. For your own stock research, keep the original thesis and its dates. Later success should not allow you to rewrite the assumptions you really used; later failure should lead to a specific correction rather than an excuse based on everyone else having agreed.
How could you use a conflicted report without simply accepting or discarding all of it?
Verify facts independently, expose the valuation assumptions and treat the recommendation as a separate judgement.
Explore related chapters