Sell-Side Research Independence in Theory and Practice: Analysts and Mortgage Traders Clash in 2007

The “theory” part of Laurie Goodman’s story describes regulatory attempts to improve the independence of sell-side research analysts. The “practice” part of her story takes place in 2007, during the subprime mortgage credit crisis. UBS analysts making bearish calls on mortgage securities clashed with traders holding those securities in inventory.

Research Regulation and Incentives

Ideally, research produced by security firms, aka “sell-side research,” enlightens investors and helps them make buy/sell decisions. Ethical problems arise when a research analyst’s opinion is unhelpful to his firm. This happens when an analyst has a poor opinion of a new issue his firm’s bankers or structurers are bringing to market or a poor opinion of a security his firm’s traders hold in inventory. The analyst’s dilemma is whether he expresses his true negative opinion of the security, or instead says something positive to help the firm that pays him.

In 2002, the Securities Exchange Commission (SEC) launched an investigation into equity research practices, specifically around initial public offerings. Corporations issuing equity, especially for the first time, naturally want the equity to sell at a high price. A high stock price maximizes proceeds to the corporation and to corporate insiders selling shares. So, in selecting an underwriter, corporations favor firms whose research departments express a positive view of the corporation’s stock.

The SEC accused Henry Blodget of Merrill Lynch and Jack Grubman of Salomon Smith Barney of skewing their research to help win equity underwriting mandates for their respective firms. To prove its point, the SEC cited instances where bearish views expressed by the analysts in private emails were at odds with the bullish opinions expressed in their published research. Henry Blodget paid $4 million in fines and Jack Grubman paid $15 million in fines. Both men were permanently barred from the securities industry.

Blodget and Grubman had tremendous incentive to help their firm’s bankers obtain profitable equity underwritings. Both men were well-rewarded for being instrumental in winning underwriting mandates for their firms. Merrill paid Blodget $18 million over 1999-2001 and Salomon Smith Barney paid Grubman $68 million over 1999-2002.

The SEC’s investigation of equity research practices led to a settlement between it and ten security firms in 2003. Besides incurring financial penalties, the ten firms agreed to equity research reforms. The theme of these reforms was “separate:” separate reporting lines, separate legal and compliance staffs, separate budget processes, and even physical separation. Equity analysts were barred from participating in underwriting pitches and bankers were barred from a role in analyst compensation. The most visible sign to investors of equity research reform is an 84-word testament that analysts must append to all research. Chief among the declarations is that the views expressed in the research accurately reflect the analyst’s own personal view of securities or issuers.

Later, equity research rules were applied to fixed income research, ignoring important differences between the two sectors. Equity analysts can avail themselves of publicly available information on corporations and equity prices are readily accessible. Since fixed income markets trade over the counter, fixed income researchers are dependent on traders for price information and market color. Securitized product analysts are also dependent on structuring desks for information on security structure. If traders and structurers freeze out a fixed income analyst, the analyst risks becoming ineffective. So, the link between fixed income research and trading and structuring desks cannot be completely severed.

In creating rules to make sell-side research independent, the SEC seemed confident that requiring research to have a separate reporting line from sales and trading would be effective. But research is a cost center at security firms and research analysts are paid from the profits that others create. Someone, somewhere in the firm, must estimate how much a research department or a particular research analyst contributes to profitability and compensate accordingly. Research analysts understand this and it influences their behavior. It is simply unrealistic to think otherwise.

On the other hand, research analysts must look after their own careers and protect their own reputations. If they give bad advice that harms investors, researchers will lose credibility, making themselves ineffective and eventually out of a job. It’s also true that good calls make an analyst’s reputation. Henry Blodget’s reputation was made on his seemingly unlikely but correct prediction that Amazon stock would rise 67% within a year. All these considerations were on my mind during the subprime mortgage credit crisis of 2007-08.

The Subprime Credit Crisis

As early as 2004, Tom Zimmerman, our non-agency mortgage research analyst at UBS, was concerned about the run up in home prices and its effect on mortgage credit quality. In the four years 2002-05, home prices appreciated 55% according to the Case-Shiller index. High appreciation made residential mortgage lending seem very safe. If a homeowner got in financial trouble and could not make his mortgage payments, home price appreciation often allowed the distressed homeowner to sell his house, pay off his loan balance, and most of the time at least recoup his initial down payment in the house and often incur a gain on the transaction. In the boom years of home price appreciation, lenders rarely foreclosed on properties. Rarer still was lenders taking a loss on a mortgage loan.

But home price appreciation also made traditional loans unavailable to many would-be borrowers. They did not have the higher down payment necessary for a home that had become more expensive or the higher income necessary to qualify for a larger mortgage loan with higher monthly payments. Lenders responded by creating “affordability products,” that required lower down payments and lower initial monthly payments. The rarity of default encouraged lenders to expand lending to those with lower FICO scores, higher debt-to-income ratios, and even to those with “stated,” or unverified, income. Lenders also gave loans to speculators, who assumed that continued home price appreciation would allow them to “flip” a house for a quick profit. Easy loan terms and speculation propelled home prices still higher.

Until one year, home prices did not rise spectacularly. Over 2006, home price appreciation was only 1.7%, not even keeping up with 2.5% inflation. Mortgage pools started sustaining “early payment defaults” where a homeowner fails to make one of his first three monthly payments. At December 2006, an unusually high number of houses for sale were unoccupied, foreshadowing a rise in distressed sales. In February 2007, Tom looked at the pattern of loan delinquencies and loan defaults that had already occurred and predicted 6.4% losses on the subprime mortgage loans backing subprime bonds issued January-June 2006. He also named six subprime bonds he thought would default out of a 20-bond triple-B-minus subprime bond index.[1]

Research Independence in Practice, 2007

We used to have a “dance” at UBS on Tuesdays before distributing The Mortgage Strategist, our weekly research publication. We gave the relevant trading desk a draft of our articles and a chance to comment. Sometimes we had errors in those articles because we didn’t completely understand a nuance in a bond’s structure. Sometimes the desk was happy with our articles, sometimes the desk was less happy with our articles. We knew the desk’s inventory, and tried to avoid bad-mouthing a position the desk held in our publication. We might, for example say security type A is cheaper than security type B; rather than saying security type B is a really horrible deal. So, prior to 2007 conflicts were rare, but they did happen, and sometimes we scraped an article or cut part of an article in deference to the desk’s wishes. These conflicts became more frequent in 2007-08.

The desk head looked at Tom’s article, which was very, very well done. He said, “Oh, these losses are way too high.” I said, “I don’t think so, and it’s a really an important article.” He said, “You can’t publish it. This is devastating to our business. You just can’t publish it, and I don’t even think it’s right.”

If the head of a mortgage desk and I couldn’t settle our disagreement, which was often the case in 2007-08, we would end up going to the head of the entire mortgage business to arbitrate the issue. The desk head said, “You really don’t need to emphasize the fact that triple-B-minus subprime bonds will default. In fact, you can just not mention that.” I said, “But all the article’s data leads to that conclusion and we don’t have time to substitute another article. Tom has been working on this for a long time and it’s really good.” The head of the entire mortgage business allowed the article to go as written.

So here was a case where we published an article that was worse for the desk than some of the articles we scrapped in deference to the desk’s wishes. Tom’s article wasn’t about one narrow group of securities; by implication it was about all subprime bonds issued in the first half of 2006. That’s why it was so important. Because it was so important, I pushed to publish it.

It's worth mentioning here how we handled conference calls and in-person meetings set up by salesmen. We would let the salesperson know what we were going to say if asked for our opinion on certain topics. Specifically, we would warn them if we had a negative opinion on something they were trying to sell. Most of the time, salespeople were happy for us to give negative opinions to their customers. Salespeople had similar incentives to ours: they got paid for selling securities, but they didn’t want to lose a customer by selling him a bad bond. The most productive salespeople were those with long-term, enduring relationships with their customers; this relationship was more important than any single sale. Proving to the customer that the salesperson had their back was a necessary part of cementing this long-term relationship.

After Tom’s article was published, I told the UBS risk management folks that covered the mortgage business, “I hope you saw that article last night. It’s one of the most important things we’ve ever done. It showed that triple-B-minus subprime bonds are going to take losses.” He laughed at me. I said, “No, no. Please look at the article, do your own analysis, but this is an issue.” Throughout the subprime crisis, UBS’ risk management group was worthless.

That incident occurred in February 2007 and from that point on, we became more and more negative on mortgage credit and increasingly at odds with the trading desk. Every few months, Tom updated and published his list of bonds in subprime indices he thought would default. He was the only sell-side analyst to do so. The focus of his research shifted away from finding good relative values among subprime bonds to warning investors to stay away from the whole sector.

At about the time we published Tom’s article, analysts at Bear Stearns wrote an article saying that a particular subprime bond index would never go below 90. Our subprime trader came over to me and said, “This is what a research department that likes its trading desk does for them.” I replied, “This is what a research department that will have no credibility in three weeks does for its trading desk. After losing their credibility, they will be useless to their trading desk.”

The trader laughed and I asked him, “Do you think Bear Stearns’s analysis is right?” He said, “No, the index is clearly going below 90. But I just wanted you to see what supportive research looks like.” I said, “I’m just not kind enough” and he replied, “I didn’t think you were.” Two weeks later, the index that Bear Stearns forecasted would never go below 90 hit 63.

A couple months later, our subprime trader asked, “Why can’t you say anything nice about our securities?” I said, “Okay, here’s the deal. You give me two securities you own that you think are great value. Give me securities priced at 60 you think should be priced at 80 and I’ll write a bullish report on them.” He looked at me and said, “I have securities priced at 60 that are worth 60, barely; I have securities priced at 60 that are worth 30. I don’t have any securities priced at 60 that are worth 80.”

I had written an article on Alt-A mortgage securities in January 2007 saying that they would follow subprime securities down in value and didn’t make sense on a risk return basis. The trading desk ignored it. I updated the article in May 2007, and predicted still further declines in Alt-A mortgage securities. The Alt-A trader, who I actually liked a lot, went berserk. He said, “Listen, you can’t write this. This is devastating to us. Our desk is loaded with Alt-A inventory.” I said, “I showed you the first article in January that said Alt-A securities didn’t make sense on a risk-return basis. You shrugged and thought it wasn’t a very interesting article. I am writing the same thing again, updating it with the last three month’s performance and using today’s prices.” He said, “Yeah, but the difference between then and now is now people are listening to you.”

The Alt-A trader repeated that the article would be devastating to his positions. I said, “Okay, here’s the deal. I’m going to hold the article one week and you can sell your bonds. But even if you don’t sell your bonds, the article goes out in a week. Okay?” He agreed to that deal. I held the article a week and published. But despite aggressive marketing, he still held a lot of bonds in inventory because he didn’t reduce prices. The trader was always an adult about research conflicts and recognized that we had a deal and didn’t complain. But his boss, the head of the non-agency mortgage area, fumed.

In August we got a new head of fixed income. The risk management group asked us to support their assessment that UBS would never lose more than $400 million on its residential mortgage-related positions. They wanted us to produce confirming analysis in a week so they could show it to the new head of fixed income. I said, “We’re not writing anything to support a view that is wrong.”

Later, when I met the new head of fixed income he asked me, “What do you think of risk management’s $400 million ceiling on mortgage losses?” I said, “I didn’t sign onto it because I thought it was crazy. We are going to lose much more.” He said, “Yeah, when I read their report, I thought it was crazy, too.” I said, “But I heard you were new to mortgage-backed securities.” He said, “You don’t need to know a lot about mortgage securities to know the $400 million number is fantasy.”

The final break with the trading desk came in September 2007 when structured investment vehicles (SIVs) began to have difficulty rolling over their commercial paper liabilities due to questions about the quality of their assets. The only person on our team that knew anything about SIVs was Greg Reiter, who unfortunately, isn’t with us anymore. He died in a tragic accident. He, Tom, and I talked about the systematic effect of SIV failures because SIVs were huge investors in triple-A-rated mortgage-backed securities, issuing asset-backed commercial paper to fund themselves. Greg said, “I’ll write an article and estimate the dollar amount of RMBS (residential mortgage-backed securities) that might be liquidated by failing SIVs.”

Greg came out with a huge number. I showed the article to the head of non-agency mortgages on our publishing day. He said, “Cut the number in half.” I said, “Why should we cut the number in half? This is Greg’s estimate.” He said, “Because this will really hurt us.” I said, “We’re not cutting the number in half.” He said, “Laurie, I’ll have you fired.” But his desk had been losing money all year, so I said, “You’ll have to start making money before you can get me fired.” The way it works is if you aren’t making money, you have no power.

Meanwhile, the head of the mortgage business was on vacation that week and not available to arbitrate our differences. The non-agency mortgage head thought of something else he could do, since we both knew he couldn’t get me fired. He said “If you run this as is, I will never give you or your team any information. I will never talk to you again.” I decided to take a chance and run the article as it was. True to his word, he never did talk to me again.

In October 2007, Shumin Li, another analyst on my staff, helped a Wall Street Journal reporter on an article about Countrywide Financial that was very negative. He gave her some numbers for her article, but because the non-agency head stopped talking to us, we didn’t realize that the desk held a large inventory of Countrywide bonds. The desk is supposed to post its inventory, and there was only one or two small positions posted. I was in Chicago on business when the head of non-agency mortgages called me screeching, “How could Shumin give this interview? Don’t you realize that we’re loaded with Countrywide bonds?” I said, “You stopped talking to us. Your desk stopped giving us information. Your inventory is posted, but it’s not your real inventory. We don’t know what the desk really owns.”

Resolution

Eventually, the market settled our disputes with mortgage traders and risk managers. On 30 October 2007, UBS admitted to $5 billion of losses and on 10 December 2007 $10 billion more. In April 2008, UBS acknowledged cumulative mortgage losses of $37.4 billion and in 2009, when the final write-down was taken, cumulative losses totaled $50 billion. An outside law firm, hired in 2008 to do an analysis of what went wrong, wrote “there appears not to have been sufficient discussion of, or actions upon, concerns surrounding subprime as an asset class until Q3 2007, even though UBS’s research team issued research reports on this area.”

But ultimately, UBS’ disaster shows that research analysts are not totally independent from their firm’s profit-making centers. Our mortgage-related CDO analysts had used Tom’s analysis to predict devastating CDO losses. With CDO issuance dead, the analysts were laid off in November 2008. When mortgage research was reduced to three people in December 2008, I resigned. Tom Zimmerman, the first of us to fully recognize subprime mortgage risk, only avoided being laid off because he was seconded to the UBS “bad bank” that held distressed assets on behalf of the Swiss government. He was able to bring four other team members with him.

Laurie Goodman was a sell-side research analyst for 30 years, covering US residential mortgage-backed securities (RMBS). Institutional Investor ranked her first in RMBS research for 11 straight years. At UBS from 1993 to 2008, she was also manager of US securitized products research, and from 2004-2008 head of global fixed income research. After leaving UBS in 2008, Laurie became senior managing director at Amherst Capital Management.

She stepped down from full-time responsibilities at Amherst in 2013 to found and become the Director of the Housing Finance Policy Center at the Urban Institute, a position she held for the next eight years. She remains at Urban Institute as an Institute Fellow, focusing on research. Laurie is on the board of directors of MFA Financial, Arch Capital Group Ltd., and Home Point Capital Inc. She is also on the Consumer Financial Protection Bureau’s Consumer Advisory Board and continues as a consultant to the Amherst Group. She has published more than 200 journal articles and has coauthored and coedited five books.

[1] “A Simple ABX Loss Projection Model,” in UBS Mortgage Strategist, 27 February 2007, page 32. For more on Tom’s predictions versus the credit rating agencies, see Getting Rid of Issuer Pay Will Not Improve Credit Ratings, But Maybe Investor Revenue Will.

To comment on a story or offer a story of your own, email Doug.Lucas@Stories.Finance

Copyright © 2022 Laurie Goodman. All rights reserved. Used here with permission. Short excerpts may be republished if Stories.Finance is credited or linked.