Not recommended.

Citi Analysts Thought Everyone Knew 'Hold' Meant 'Sell'

Matt Levine is a Bloomberg View columnist. He was an editor of Dealbreaker, an investment banker at Goldman Sachs, a mergers and acquisitions lawyer at Wachtell, Lipton, Rosen & Katz and a clerk for the U.S. Court of Appeals for the Third Circuit.
Read More.
a | A

A major function of an investment bank is to get people to buy and sell securities. The way this works is pretty much that the banks' employees get on telephones or airplanes or Bloombergs and tell clients, "Hey here is a security you should buy," and tell other clients, "Hey here is a security you should sell."

Should they? The clients, I mean? I don't know, that is a hard question to answer in the general case. But there seems to be a popular view that the investment bank should mean it when it says that clients should buy or sell a particular security. But this is a very high bar. An investment bank is primarily an intermediary. To make its money, it needs to convince some people to buy a security, and other people to sell the same security. How can it mean it both ways? The more sincerely and adamantly it believes that people should buy a particular security, the less plausible is its simultaneous belief that other people should sell it. The safest course would be to buy and sell only securities about which the bank is entirely indifferent, but it is hard to market indifference with much enthusiasm.

Citigroup Global Markets Inc. research analysts did their best:

For example, at an idea dinner that occurred in July 2011, a CGMI equity research analyst identified a stock as a "short" pick. Before the dinner, however, the analyst had upgraded that stock from Sell to Hold and reiterated his Hold rating on the company in the last report that he published prior to the dinner. Thus, the analyst's characterization of the company as a "short" pick was inconsistent with his published research on the company at the time. Moreover, the analyst identified other companies, on which he had Hold or Neutral ratings, as "short" picks at six subsequent idea dinners between October 2011 and June 2013. These picks were therefore inconsistent with the ratings identified in the analyst's published research.

That's from the Financial Industry Regulatory Authority, which fined Citi $15 million today for that and other misbehavior. Even more so than the rest of the people at an investment bank, research analysts have to worry about sincerity. They have to certify in writing that they believe what they're saying. When an investment banker or salesman or broker recommends a trade, there may be some sort of background expectation that she is being sincere, but that expectation is tempered by knowledge of market realities, and she's not really held to it. But when a research analyst recommends a stock, he has to also separately say, "No, I really do like this stock." I guess this guy had to certify, "No, I really am indifferent to this stock." But even that wasn't true! Or it was, and his short picks at those idea dinners were insincere? Who knows. Finra doesn't say how they panned out. And even if it did, that wouldn't prove sincerity or lack thereof. In addition to being insincere, people can be wrong.

The problem is that no one seems to know what a research analyst's job is, or ought to be. One thing that analysts do is publish occasional reports about the companies they cover, with big Buy or Sell or Hold recommendations slapped on the top. This takes a certain amount of time and effort, but not that much time and effort. The rest of the time, this is what analysts do:

CGMI equity research analysts engaged in frequent communications with the Firm's clients as well with CGMI sales and trading personnel, whom equity research analysts viewed and treated as internal clients. These frequent interactions took place by email, over the phone and in-person, and at meetings, social events and other functions hosted or attended by CGMI equity research analysts and the Firm's internal and external clients.

On the one hand, everyone -- by which I mean, everyone who is entitled to get Citi research, meaning basically its brokerage clients -- can read the analysts' reports. On the other hand, some people are e-mailing and telephoning and meeting and going to "idea dinners" with Citi research analysts, and some people aren't. What distinguishes the first group from the second is that Citi likes the first group more. The clients who meet a lot with Citi analysts are the clients who are buying and selling a lot of stock. Citi wants to please those clients, and providing them with differentiated access to analysts apparently pleases them. And Citi explicitly pays the analysts for pleasing them:

Voting by clients and sales personnel was a significant factor in CGMI's determination of its equity research analysts' compensation. CGMI paid each equity research analyst based upon, among other things, the analyst's relative rank on a "scorecard." Approximately half of each equity research analyst's "scorecard" rating was related to interactions and feedback with both internal and institutional clients.

Imagine being an analyst. You go to a meeting with an important firm client, whose opinion of you determines your compensation. She wants to know what you think about Company X. You hand her the report on Company X that you published a month ago, saying that everyone should Buy Company X. She says, "Yes I read that a month ago; now I want to hear how your views have changed and what new information you have." You say ... "Wait for my next report in a month"? Really? That doesn't even make any sense. Why have the meeting if all you can say is stuff that's in your report?

This compensation structure, while not inconsistent with the regulatory framework developed in the wake of the 2003 Global Research Analyst Settlement, created an incentive for CGMI equity research analysts to engage in inappropriate communications with clients, including providing non-public research information to clients before the research was published.

I have added some emphasis there. Rather famously, during the last Internet boom, a bunch of research analysts recommended a bunch of stocks that they didn't really believe in, deep in their souls. We know this not because we looked into their souls, but because we looked into their e-mails, where they would call companies naughty words, or at least abbreviations of naughty words, and then issue Buy recommendations on those companies anyway. This was bad, and so the banks and their regulators reached a Global Research Analyst Settlement that they thought fixed what they thought was the problem.

The problem, they thought, was that the research analysts were too solicitous of investment-banking clients -- issuers of stock -- and so were too generous with ratings. So the solution -- besides requiring analysts to certify that they mean what they say -- was to cut analysts off from investment banking. So the new rules said that analysts couldn't "directly or indirectly" participate in "a road show related to an investment banking services transaction," or get paid for bringing in investment banking (i.e., stock offering) transactions. Analysts could only get paid for helping out investors, not issuers.

But that settlement didn't prohibit saying different things to different (investor) clients. Why would it? Providing better service to better clients is pretty standard, in the financial industry as in other industries. And really, if you'd told people in 2003 about this stuff, they'd have been thrilled. This is a story of research analysts working on behalf of investors, not pulling any punches, sharing their ideas openly and honestly with investor clients. I mean. Some investor clients. Not all of them. Not, you know, the consumers of equity research reports. But the consumers of those reports have had at least a decade to get used to the idea that they're not to be taken too seriously. At least here the conflicts of interest are all among investor clients. It's just that saying that an analyst has to be solely loyal to his investor clients doesn't really answer too many practical questions. Which investor clients?

Weirdly Citi also violated the global settlement rules around initial public offerings:

In 2011, CGMI violated NASD Rule 2711(c)(5) when one of its senior equity research analysts indirectly participated in road show presentations in connection with the IPOs of two companies.

On May 5, 2011, the analyst participated in a "Roadshow Presentation practice session" for one of the companies to provide input on the presentation and areas to emphasize and de-emphasize during the road show.

On July 6, 2011, the analyst received draft road show presentation slides from the other company. The CGMI research analyst then sent an email to company representatives in which he suggested that the company "amp up" discussions of certain topics in its road show presentation.

Unlike the idea-dinner stuff -- where Citi told some clients to Hold a stock and others to Short it -- this stuff was clearly against the rules. And yet. Comparatively speaking, it seems pretty harmless. This doesn't call the analyst's sincerity into question, or run the risk of deceiving anyone. The bank has an issuer client, and it has someone with the ability to help that client, not by lying to investor clients (bad!), but by providing stylistic tips on a roadshow presentation. What's wrong with that? The rules are designed to preserve analysts' honesty by cutting them off from investment banking, but this seems to go a bit overboard. Plus, cutting analysts off from investment banking isn't always enough to preserve their honesty.

  1. You know what I mean. As a market-maker, a bank is an intermediary in the medium term, though a principal at any particular moment. Even in pure prop trading, if a bank is really really convinced that it should sell (buy) a particular security, it is on shaky ground trying to get anyone else to buy (sell) it.

  2. This is not really true. A big big big function of modern finance is precise allocation of risks from people who want to get rid of them to people who want them, for reasons other than just difference of opinion. A bank can quite reasonably sell long commodity risk to a consumer and short commodity risk to a producer and feel good that it's leaving both sides better off. Or it can advise on a merger that produces enough synergies to leave both sides better off. Et cetera et cetera; surely a huge chunk of the work actually done by investment banks can be more or less plausibly justified on grounds like this.

    I'm not convinced that buying and selling individual stocks is frequently in that category, though.

  3. Fun examples:

  4. As my Bloomberg View colleague Megan McArdle has written here, "everyone" knew that these reports were worthless, and yet:

    And why were equity research analysts issuing hold ratings when they actually meant sell? Presumably because not everyone knew that all these things were meaningless. Some people were buying stocks based on these insider tricks. But we never talked about them. We never even thought about them.

    Still, I mean, there was a lot of reason to doubt.

    In this and other cases, there's a plausible defense that the investing time frames for the idea dinners and the research reports were different, so the analyst could have meant that a stock was a short-term Short but a long-term Hold, or vice versa, or whatever. This defense does not move me all that deeply. Presumably the nuances of this view were communicated more clearly to the idea-dinner clients than to the research-report readers.

This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.

To contact the author on this story:
Matthew S Levine at

To contact the editors on this story:
Zara Kessler at
Zara Kessler at