Infectious Greed (68 page)

The concentration and disclosure problems Greenspan cited were a double whammy. Just as Long-Term Capital Management unraveled when its lenders finally learned of the daisy chain deals that enabled that firm to borrow so much money at favorable terms, a derivatives dealer's network of contracts might similarly unravel if counterparties—or even shareholders—learned the truth about the dealer's mysterious web of deals.

Greenspan wanted banks to become transparent, so that shareholders, analysts, and counterparties could understand their activities. He also wanted more competition. In the long run, greater transparency and competition should lead to greater market discipline. In the short run, they might lead to some spectacular failures, as everyone finally learned how credit default swaps have been used to shift risks around the financial world.

 

 

T
he second type of credit derivative—the Collateralized Debt Obligation—posed even greater dangers to the global economy. In a standard CDO, a financial institution sold debt (loans or bonds) to a Special Purpose Entity, which then split the debt into pieces by issuing new securities linked to each piece. Some of the pieces were of higher quality; some were of lower quality. The credit-rating agencies gave investment-grade ratings to all except the lowest-quality piece. By 2002, there were more than half a trillion dollars of CDOs.

The latest innovation—the
Synthetic CDO—
was the ultimate in financial alchemy. A Synthetic CDO was like a standard
cash-flow
CDO, except that a bank substituted credit default swaps for loans or bonds. In other words, the “assets” of the SPE were credit default swaps. As a result, the companies whose debts formed the basis of a Synthetic CDO had
no relationship
at all to the deal; most likely, the companies would not even know about it. Neither the investors in the SPE, nor the banks, ever had to touch the companies' loans or bonds.

Synthetic CDOs might seem like unusual or esoteric side bets, but by
2002 they were a mainstay of corporate finance. In 2001, banks created almost $80
billion
of Synthetic CDOs. During 2002, even after the bankruptcies of Enron, Global Crossing, and WorldCom—companies whose debts were referenced in the credit default swaps of numerous Synthetic CDOs—financial institutions still were continuing to do these deals.

Credit default swaps had pushed risks to unknown places, and now CDOs were moving them underground. Howard Davies, the British financial regulator, relayed a comment, from an investment banker, that Synthetic CDOs were “the most toxic element of the financial markets today.”
⁸⁷

In conceptual terms, CDOs resembled the earlier Collateralized
Bond
Obligations pioneered by First Boston (and described in Chapter 3). To understand CDOs, imagine that a chocolate bar represents a typical corporate bond: a decent blend of ingredients; medium richness; almonds, if you wanted. Some chocolate bars contain artificial—or even accidental—ingredients that consumers would prefer not to think about.

Now imagine melting 100 chocolate bars of different types in a pot, and separating their constituent parts. What would it look like? A slice of Maison du Chocolat-quality ganache, some cocoa and milk, a few almonds, and a small pile of remains you might prefer not to eat. Do you think it would be possible to make money by selling the parts separately? You certainly could find buyers of the high-quality pieces, but who would want the remains? Moreover, given that someone already had gone to the trouble of making the chocolate bars from the various ingredients, is there any reason to think you might be able to deconstruct and then resell the parts for more than the cost of the bars themselves?

Substitute corporate debt for chocolate bars and you understand CDOs. In a CDO, a bank would gather a portfolio of, say, 100 different bonds or credit default swaps, and separate the portfolio into tranches of varying qualities, to be sold to investors. Typically, there was a AAA-RATED senior tranche, one or more investment-grade-rated mezzanine tranches, and then the remains, which were called anything from the junior piece to preference shares to nuclear waste.

The question was: how could a bank make money doing such an elaborate reconstruction? In economic theory, the Law of One Price said that similar assets should have similar values. If they didn't, someone would buy low, sell high, and earn a riskless profit. If a bank could make money repackaging corporate debt, it must mean there were inefficiencies in the
corporate-debt market. But if some of the debt in the portfolio were mispriced, why weren't people trading it in the market until prices were accurate? There weren't the same restrictions on selling-short bonds or credit default swaps as there were for stocks, and the manic individuals who drove up prices of Internet and telecommunications shares were minor players in the corporate-bond and credit default swap markets. Inefficiencies in these markets were unlikely to last long. Why, then, did banks think the whole—sliced in new ways—was worth more than the sum of the parts?

One reason, again, was legal rules: CDOs, like credit default swaps, were driven by banks that wanted to remove loans from their balance sheets, just as Fred Carr—one of Michael Milken's best clients—had repackaged a portfolio of junk bonds to avoid regulatory requirements in 1989. Now that banks were being permitted to engage in the securities business, they wanted to free up their capital to use in attracting new clients. Banking regulations required banks to set aside capital for their loans; if they could get rid of those loans, they could use the capital for something else.

Another reason involved the all-important credit-rating agencies. Beginning in the 1970s, regulators had given up trying to keep pace with modern financial markets. Instead of making substantive decisions about which securities financial institutions should be permitted to buy and sell, the regulators had deferred to the credit-rating agencies, by passing rules and regulations that depended on ratings. Recall from Chapter 3 that, as a result of these rules, there was a sharp divide between the prices of investment-grade and non-investment-grade bonds. Michael Milken of Drexel had spotted this inefficiency, and had profited during the 1980s by buying portfolios of non-investment-grade bonds. The rules regarding credit ratings hadn't changed since his time, and the three major agencies still had an oligopoly lock on the ratings business, as regulators refused to approve any competitors. Not surprisingly, their business boomed. Moody's became a free-standing, publicly traded company worth more than $5 billion.

Anyone looking closely at the credit-rating agencies would find it difficult to justify their importance. The analysts at the three rating agencies were perfectly nice people, but they were not—to put it charitably—the sharpest tools in the shed. Banks snapped up the best analysts, and investment funds hired the second best. Based on their recent track record, the remaining employees would have done a better job if they had simply
followed the business section of a daily newspaper. Not only had the rating agencies given Orange County and Pacific Gas & Electric their highest ratings just before those entities became insolvent, they more recently had given high ratings to Enron, Global Crossing, and WorldCom—and stuck to those ratings until just before the companies filed for bankruptcy.

Yet it was the credit-rating agencies' inaccurate ratings of corporate debt and the tranches of CDOs that made the parts worth more than the whole. To see this, suppose the agencies rated only two bonds on a simplified scale, from A to F. Also suppose legal rules required that regulated institutions buy bonds with a rating of A or B. If there were two bonds in the market—one rated A and the other rated F—the regulated institutions would buy only the A bond.

Now suppose a bank repackaged both bonds into a CDO with one tranche of new securities. If the rating agencies gave these new securities a rating of C—the average of A and F—regulated institutions still wouldn't be able to buy it. But what if the rating agencies bumped up the new securities to a rating of B? Now regulated institutions effectively could buy
both
bonds. The new securities of the CDO would be worth more than the bonds sold separately, because now there were more buyers. More demand meant a higher price.

This logic applied equally to more complex structures, with dozens of bonds and multiple tranches. Essentially, the credit-rating agencies made the parts worth more than the whole by
overrating
the CDO tranches. The computer models used to assess CDOs were merely complex ways of justifying higher ratings for the tranches than the overall ratings for the underlying debt in the market. Many of these models had been created by the banks doing the CDO deals, and their nuances were not understood by some rating-agency employees. For example, the ratings generated by a Synthetic CDO model depended on variables—such as diversity scores, weighted average rating factors and spreads, and various “overcollateralization” tests—which determined the scenarios under which the rated tranches of a CDO would lose money. Bank employees privately admitted they could tweak these models to make a CDO deal appear to add value. By the time a rating agency employee understood the bells and whistles of these models, the banks doing CDO deals would hire him or her, at a significantly higher salary.

This attrition wouldn't prevent the rating agencies from continuing to make money—they would do very well. But it indicated that the product the agencies were selling was not their own expertise or even some well-constructed
strategy to exploit market inefficiencies. Instead, they were selling an inflated rating methodology that enabled regulated buyers to purchase riskier, higher-yielding investments. The banks controlled the inputs and told the rating agencies what they needed to get a deal done. As a result, the agencies could hire just about anyone to fill in the blanks—and still make money.

The “value” magically created by a CDO was parceled out among the various participants: the buyers of the highly rated pieces were paid a higher yield than on comparably rated bonds; the banks arranging the CDO received a fee; and the buyer of the junior piece had access to a new kind of investment, which otherwise was not available in the markets—a highly leveraged investment in corporate bonds. Essentially, the junior piece was borrowing money from the senior pieces to invest in the debt of the CDO. That meant that the junior piece was very risky and volatile, but also had the potential for high returns.

Of course, there were valid reasons for investors to buy diversified portfolios of corporate bonds, and there were occasions when financial innovation could bridge the gap between the needs of corporate borrowers and the needs of lenders. However, it wasn't necessary to create CDOs to fulfill these purposes. Investors could diversify on their own or through investment funds, and financial institutions had proven more than able to serve as intermediaries when borrower and lender needs weren't matched—the classic example was an interest-rate swap, which enabled a company that wanted to borrow at a fixed interest rate to borrow from a lender that wanted to lend at a floating interest rate, or vice versa. In contrast, CDOs—especially Synthetic CDOs—weren't necessary to fulfill any economic function. Instead, they made sense only because of foolish credit ratings and the legal rules that depended on them.

Put another way, the same thing would happen with chocolate bars if there were chocolate-rating agencies and legal rules that required supermarkets to purchase only approved chocolate products. If the Food and Drug Administration created incentives for people to buy particular blends of products contained in chocolate bars—or if there were mandatory chocolate-bar raters who gave inflated ratings to reconstituted bars—candy companies would engage in the same kinds of dysfunctional behavior as parties to CDOs.

No one had paid much attention to the first warning that CDOs threatened the health of the global economy. In July 2001—two months before Jeff Skilling had resigned from Enron, and long before investors
learned about the accounting problems at Global Crossing and WorldCom—American Express, the U.S. financial-services conglomerate, had calmly announced that it would take an $826 million pretax charge to write down the value of investments in high-yield bonds and Collateralized Debt Obligations. It all sounded much too esoteric to matter to average investors. The media brushed off the details by focusing on the junk bonds involved in the various deals, and commentators seemed to agree that these losses were just a minor consequence of the explosion in financial innovation.

American Express was a sophisticated financial institution, and it easily could handle a loss of $826 million. From an outsider's perspective, the losses were due to some complex workings within the company; but, the assumption was that someone within American Express understood what was going on.

Then there was the stunning public admission by the chairman of American Express, Kenneth Chenault, that his firm “did not comprehend the risk” of these investments. What?

During 1997, the financial-advisory division of American Express had begun creating CDOs, buying some of the highly rated pieces, and selling the junior pieces to clients. When the bond markets collapsed in 1998, along with Long-Term Capital Management, the buyers of junior pieces disappeared, and those risky pieces were viewed as nuclear waste. Without those buyers, American Express would not be able to create any more CDOs, which meant it would have to give up substantial fees.

At this point, American Express made its big mistake. Instead of exiting the CDO business, at least temporarily, it continued to do CDOs—and began buying the junior pieces for its own account. By March 2001, American Express owned several billion dollars of CDOs.

Unfortunately, there were some corporate defaults during 2001—even before Enron—and American Express sustained losses on its junior pieces due to these defaults. Moreover, as the junior pieces lost money, they provided less protection to the more senior pieces American Express held. Those pieces also lost value, and by the time Kenneth Chenault realized he had a problem, his firm had lost $826 million. The computer models used to calculate these losses were complex, and—at least according to Chenault—American Express had not properly understood them.