Goldman Sachs and Abacus 2007-AC1: A Look Beyond the NumbersPublished: April 28, 2010 in Knowledge@Wharton
Goldman Sachs is the Wall Street mega-firm that people either love or hate -- or love to hate. Its money-making prowess leaves many impressed, envious or suspicious. To admirers, the revolving door between its executive suite and high government office shows commitment to public service. Detractors call it undue influence.
Now the firm's reputation is on the line, as it fights a fraud suit brought by the U.S. Securities and Exchange Commission (SEC) over a single deal in 2007, the sale of a complex "synthetic collateralized debt obligation" called Abacus 2007-AC1. The deal lost investors $1 billion but produced $1 billion in profits for Goldman's collaborator, Oregon-based Paulson & Company, a hedge fund betting the housing bubble would collapse. While Goldman says it did nothing illegal or unethical, the SEC says the firm withheld "material information" from the investors -- specifically, the hedge fund's role in selecting underlying securities.
The case raises important questions:
- Could Goldman's customers really have evaluated the Abacus risks for themselves, as Goldman claims?
- Do derivatives like synthetic collateralized debt obligations, or CDOs, serve any useful purpose?
- Does Goldman's defense -- that it had no obligation to alert investors to especially high risks -- undermine its claim to be a "client-centered" firm worthy of customers' trust?
"If you view your clients as adversaries, you're likely to have many fewer of them," says Wharton finance professor Richard J. Herring, describing Goldman's dilemma as it tries to be a trusted advisor while also investing for its own benefit. Adds finance professor Franklin Allen: "It looks very much like [Goldman] exploited the clients on the other side of the deal from Paulson. The problem is that the people who bought the securities at IKB Bank are not as sophisticated investors as hedge funds or as seasoned banks like Morgan Stanley or Deutsche Bank. It would have been helpful to them to know that Paulson had been involved in the design of the securities. That goes to the heart of the SEC's case." [Editor's note: While it is true that the SEC has charged Goldman with wrongdoing, the charges have yet to be proved in court.]
While Goldman maintains that its Abacus investors had all the information needed to evaluate risks for themselves, Herring says synthetic CDOs are very opaque. "They are so complicated that, in practice, it's virtually impossible with publicly available information to dig down to the underlying securities -- mortgages, credit card loans, etc. -- that need to be valued.... My impression is that, other than hedge funds -- and perhaps Goldman Sachs -- virtually no other players took the trouble to do it. They merely traded based on the credit rating bestowed by the credit rating agency."
Goldman insists it did nothing wrong in the Abacus trade. In a statement issued on April 16, after the SEC filed its case, the company said: "The SEC's charges are completely unfounded in law and fact and we will vigorously contest them and defend the firm and its reputation." Since then, the controversy has broadened with the release by the Senate Permanent Subcommittee on Investigations of boxes of Goldman emails and documents related to other transactions. In daylong hearings on April 27, some senators said the papers show Goldman made a practice of deliberately luring customers into money-losing deals, while Goldman secretly bet against them. The Goldman case has spurred Democrats' efforts to rein in trading of complex derivatives that contributed to the financial crisis.
"The evidence shows that Goldman repeatedly put its own interests and profits ahead of the interests of its clients," Sen. Carl Levin, D-Michigan, said at a press briefing on April 26.
In a statement opening his testimony before the investigations subcommittee on April 27, Goldman chairman and CEO Lloyd C. Blankfein said: "While we strongly disagree with the SEC's complaint, I also recognize how such a complicated transaction may look to many people.... We have to do a better job of striking the balance between what an informed client believes is important to his or her investing goals and what the public believes is overly complex and risky."
The SEC charges that Goldman illegally withheld material information when it did not tell the Abacus buyers that mortgage bonds underlying the CDO had been selected with the help of Paulson & Company, one of the world's largest hedge funds. Paulson wanted to bet that the housing and mortgage markets would collapse. To do that, Paulson needed a CDO based on mortgage bonds likely to fall in value when homeowners stopped making their payments. Paulson was not included in the SEC complaint and has not been accused of any wrongdoing.
A synthetic CDO transaction requires two parties taking opposite views. The "long" party profits if the underlying securities rise in value; the "short" party profits if they fall. Each side places a bet and, in effect, the loser's losses become the winner's gains.
In the Abacus deal, completed in April 2007, Paulson took the short side and two major investors took the long side: IKB, a large German bank, and ACA Capital Management, a New York-based investment firm. Paulson worked with ACA to choose the 90 underlying mortgage-backed securities. But there is dispute about Paulson's exact role. The SEC claims Goldman led ACA to believe that Paulson was taking the long side -- that he would bet the securities would rise in value -- when Paulson was actually taking the opposite view. This, according to the SEC, led ACA to believe Paulson thought the securities were safer than they were, and that its interests and Paulson's were the same. Goldman, however, says it "never represented to ACA that Paulson was going to be a long investor."
IKB, the German bank, did not know Paulson's role, according to the SEC, which states that this role was material information that would have alerted IKB to the high risks.
In an April 16 statement, Goldman argued that the long investors had no need to know of Paulson's role, or that Paulson was taking the short side of the deal. IKB and ACA "were provided extensive information about the underlying mortgage securities," Goldman said. "The risk associated with the securities was known to these investors, who were among the most sophisticated mortgage investors in the world. These investors also understood that a synthetic CDO transaction necessarily included both a long and short side." Among the key questions is whether investors, even if very sophisticated, could accurately assess the portfolio's risks on their own.
RMBS, CDS, CDOs
The case involves four types of securities that played roles in the financial crisis.
The first are residential mortgage-backed securities, or RMBS. These are pools of mortgages converted into bonds that are sold to investors, who then receive income from homeowners' monthly mortgage payments. Typically, the bonds come in a variety of grades, or tranches. Owners of the safest have first rights to the income from the pool. Owners of the riskiest are last in line, making them the first to suffer losses if homeowners fail to make payments. As compensation, they earn a higher interest rate -- and they get a bigger share of the income if homeowners do pay their mortgages.
Next are credit-default swaps (CDS), a kind of insurance policy that pays off if a debtor fails to make its payments. An investor, for example, could buy a CDS that would pay off if a company fails to make interest or principal payments on its bonds. The CDS becomes more valuable as this default risk rises or if the bond's rating is lowered, since that improves chances of a payoff. The speculator buying a CDS does not have to own the debt security being insured.
Third is the collateralized debt obligation. While there are many types, they are pools created from other securities with shares that are then sold to investors. Many CDOs, for example, were created by assembling portfolios of risky, low-rated tranches of mortgage-backed securities. The CDO is then tranched as well, just as the mortgage securities are.
Finally, there are synthetic CDOs. These are much like ordinary CDOs except that instead of owning real securities, investors own credit-default swaps on real securities. In Goldman's Abacus deal, the CDOs owned credit-default swaps that would rise or fall depending on the fortunes of a specific list of residential mortgage-backed securities, mainly on subprime loans to homeowners who are considered risky.
Evaluating the risk and potential rewards of a synthetic CDO is essentially the same as evaluating an ordinary CDO, because both depend on the quality of the underlying mortgage-backed securities, says William Frey, president of Greenwich Financial Services, a Connecticut firm that specializes in mortgage securities but does not invest in CDOs. Can an investor evaluate a CDO by studying the underlying mortgage securities? "Theoretically, yes," says Frey. But, he adds, "It's theoretically possible to check the engineering designs of the George Washington Bridge before you drive over it."
Evaluating the CDO would require studying each mortgage security in it, altogether comprising many thousands of mortgages -- perhaps hundreds of thousands of them, Frey says. A thorough evaluation would require studying the loan-to-value ratios of the mortgages, the geographical locations of the homes, unemployment rates, local default and foreclosure rates, and other factors determining how many homeowners were likely to default.
"For all practical purposes, unless you have the most sophisticated software on the market, which few investors have, you rely on the ratings agencies," Frey says, adding that a CDO investor would primarily rely on the ratings given to the mortgage securities reflected in the CDO.
The ratings agencies themselves have been criticized for giving good ratings to securities that later collapsed in value, as those in the Abacus deal did. Critics say the problem was that the agencies' fees were paid by the firms that issue the securities, as they still are, and that the raters' computer models used past patterns that did not reflect conditions in this decade.
To further complicate any analysis, two securities with identical ratings may treat their owners very differently, says Wharton finance professor Marshall E. Blume. "Ratings are a uni-dimensional measure of risk, but risk actually takes on many forms. For example, you could have two bonds with the same probability of default, which the ratings service might give the same rating to. But one bond, if it defaults, might lose a lot more than the other bond. Therefore the ratings themselves don't convey all the information about risk."
In the build-up to the financial crisis, many AAA-rated CDOs offered higher yields than other types of AAA-rated securities, suggesting the marketplace understood the CDOs were riskier despite the identical ratings, according to Blume.
In many CDO deals, the parties involved have unequal access to information, adds Kent Smetters, professor of insurance and risk management at Wharton. "The problem is that the quality of the securities held within a CDO is better known to a seller than a buyer. As a result, riskier mortgages were often put into CDOs by banks, whereas mortgages that were not sold off by the bank generally performed better. This problem is known as adverse selection."
Because assessing such risks is so difficult, CDO buyers seek any information that could affect what they are willing to pay to take on the risk, Smetters notes. "As a sophisticated investor in a CDO, I would want to know if it contained an unusual amount of junk -- that is, assets that are likely to default. That information would determine how much I would be willing to pay for the pool of assets, even if I had a high tolerance for risk-taking. It is difficult for me as a buyer to see the underwriting of the securities. Instead, the seller should disclose information that I, as a reasonably prudent investor, would want to know in making the determination of price." The law, he adds, "puts the burden on the sellers to disclose material information."
Because assessing CDOs composed of hard-to-evaluate mortgage securities is an inexact science, a prudent investor would want to know the views of anyone, such as Paulson, involved in selecting the underlying securities, says Herring. "I would certainly have been interested to know that one of the most successful hedge fund managers in history was helping to select assets for a portfolio that he intended to short. If I had been a customer, I would certainly have felt deceived and abused, even though Goldman Sachs may have been within the letter of the law. It's simply not the sort of business that a firm that wants to be known for its integrity should take on."
In its statements and Senate testimony, Goldman's view is that its customers are sophisticated investors who assess risks for themselves. At the hearing, Goldman CEO Blankfein repeatedly described customers as coming to Goldman seeking an opportunity to take on a particular type of risk, with Goldman merely complying. Several senators, however, argued that Goldman was often the initiator, using its sales force to encourage investors to buy securities Goldman no longer wanted to own. In many cases, committee chairman Levin said, Goldman encouraged customers to buy securities without telling them that Goldman was betting against the same securities by taking a short position. Blankfein said that as a market maker, Goldman has no obligation to reveal to customers its own view on the quality of any security it sells.
This view could damage Goldman's franchise, Herring says. "Goldman Sachs has made its reputation as a trusted advisor and superb investment manager. Many people would buy a Goldman Sachs mutual fund simply because they are thought to be among the best investors in the business. If [Goldman] really wants to play the caveat emptor game, however, they are giving up a valuable reputation that was hard won over decades." According to Allen, "they have this reputation of dealing on both sides and exploiting information.... I think it will hurt them.... Along the way I think we will see a lot of messy revelations. I don't think Goldman will come out of it very well."
Nonetheless, says Blume, the SEC's case is not a slam dunk. The SEC was split 3-2 on whether to file the Goldman case, and usually does not proceed unless the view is unanimous. It is often hard to define what constitutes material information, he notes.
Wall Street's Casino
The Goldman case comes as Congress is debating Democrats' proposals to rein in derivatives trading. Some lawmakers want to create a transparent centralized exchange, replacing the opaque over-the-counter system currently used, so that participants could more easily evaluate prices. Some also want trades to go through a central clearinghouse. With the clearing agent guaranteeing payment and delivery of securities purchased, there would be less worry about whether a counterparty would make good on its end of a deal.
Using an exchange and clearing agent could help make the derivatives market safer, Blume says. Because these systems, long used for stocks, reduce fees and spreads between prices bid and asked, they reduce profits for financial-services firms, which explains their opposition, he adds. A centralized system could encourage use of standardized products, which are safer because they are more transparent, though there would continue to be some use of customized derivatives traded outside the central system. Most users would opt for standardized products, just as consumers buy ready-to-wear clothing rather than have garments tailor-made because standardized products are cheaper to use, Blume argues.
Allen agrees that the customized market could shrink if a standardized derivatives market is established. "That may happen, yes," he says. "This is an interesting balance -- between how much you want over-the-counter markets with a lot of tailor-made securities versus just standardized execution on exchanges."
As the derivatives trading system is re-evaluated, some critics wonder whether certain derivatives serve any useful purpose or have merely turned Wall Street into a casino. Frey and the Wharton faculty members interviewed say derivatives linked to real assets such as mortgages do serve a purpose. The mortgage-backed security, for example, allows a lender to convert homeowners' future payments into immediate cash, so the lender can provide money to other home buyers.
Credit-default swaps allow companies and other market players to hedge against risks. A bond owner, for instance, can use a CDS to insure against the danger of not receiving principal and interest payments as promised. The problem, some critics say, is that the CDS buyer does not have to own the security that is being insured. That allows the swaps to be used for pure speculation, as if one took out 10 life insurance policies on a stranger, hoping to profit if the stranger dies -- a bet that would be illegal with ordinary insurance. Rampant speculation with credit-default swaps forced the $182 billion government bailout of American International Group. Collateralized debt obligations can serve a useful purpose when they repackage real securities, such as those backed by mortgages, says Smetters. In that case, they help supply money for homeowners.
But synthetic CDOs, like those in the Goldman case, do not pump money to people or companies with real needs, says Frey. Synthetic CDOs, adds Blume, are more like side bets among spectators standing around a craps table in a casino.
While the creation of ordinary CDOs is limited by the availability of underlying assets like mortgage securities, an unlimited volume of synthetic CDOs can be created because they are not tied to asset-based securities but to credit default swaps, which themselves can be created in unlimited numbers. Thus, synthetic CDOs satisfied a hunger early in the 2000s for investments with high ratings. Because the ratings were poorly done, many mortgage-related securities collapsed in value, and synthetic CDOs magnified the losses.
Says Frey: "In looking at securities ... I ask one simple question: Is there an economic reason to have this transaction? And if the answer is 'no,' what does this transaction do?" Synthetic CDOs don't pass his test, Frey says. "I don't really see any need for them. I don't see that there is a real underlying economic need for that transaction. What does that transaction really accomplish, other than to move money around? Moving money around is not an economically productive event."