In most trades, the impact of the risk of different economic variables could be analyzed separately. An investor could first consider the effect of changes in interest rates, and then worry about currencies. Interest rates and currencies affected each other—they were correlated—but, in a typical trade, the correlation didn’t matter. If the correlation didn’t matter, it was safe to analyze the risks separately. For example, it was difficult to value Andy Krieger’s currency options or Bankers Trust’s structured swaps, but neither posed a correlation problem; the interest rate and currency risks of these trades were separated.
The Quanto’s risks could not be separated, because—unlike the payoffs of the other trades—the payoffs of a Quanto depended on the correlation of interest rates and currencies. In simple terms, the amount to be converted at maturity into U.S. dollars (the currency variable) depended on the value of, say, British LIBOR (the interest rate variable). Interest rates and currency rates affected each other, not only in the real world, but also in the Quanto. They could not be separated.
At the time, correlation risk was new, and could not be hedged or traded directly. So banks had to come up with ways of pricing the risk on their own. Bankers Trust developed some techniques; CSFP mastered them. The techniques were sufficiently complex that most clients had no hope of accurately evaluating Quantos. Worse, traditional risk measures understated the amount of risk in correlation-based trades, so that even a sophisticated client—trained in the state of the art of risk analysis at the time—might underestimate the risk of a Quanto.31
Investors weren’t likely to obtain much comfort by phoning other banks, as they might have for other structured trades, because many banks didn’t understand correlation risk any better. In 1991, the banks selling Quantos disagreed so much about what the trades were worth that their quoted prices differed by as much as half a percent, a huge amount. By late 1993, the differences had narrowed to one-fifth of that, but still were substantial.32 During this period, it took a typical bank an entire weekend to produce a report assessing correlation risk. (Today, relatively inexpensive software programs calculate these risks instantaneously.)
Investors didn’t seem bothered by correlation risk, and they bought Quantos in droves. Either they were oblivious to the fact that their investments now depended on correlations of various interest rates and currencies, or they didn’t care. CSFP sold over 100 Quanto swaps in 1990 and 1991, more than any other bank.
The Quanto was the perfect trade for Japan. Many companies believed interest rates outside Japan would fall, but wanted their investments denominated in Japanese yen. Similarly, Quantos also fit U.S. investors who sought higher yields, but did not want to receive a currency other than U.S. dollars.
Still, it seemed odd that investors would be so eager to buy Quantos, especially when they were unable to assess correlation risk. If currency rates moved the wrong way, the investor still would lose money; it wouldn’t help that the money paid in a Quanto was in U.S. dollars instead of foreign currency. Moreover, investors who wanted to bet on foreign interest rates could easily buy foreign bonds and exchange whatever foreign currency they received at maturity, thereby avoiding correlation risk. Yet they wanted Quantos anyway. Why?
The answer: legal rules, yet again. The true driving force behind Quantos was that they allowed investors who, by law, were not permitted to speculate in a particular currency, to do so indirectly, without being noticed. Many investors—especially U.S. insurance companies, major clients of CSFP—could not buy large amounts of non-U.S. dollar instruments. But they arguably could buy as many Quantos as they wanted, because their payments were denominated in U.S. dollars.
James M. Mahoney, writing in the Federal Reserve Bank of New York Economic Policy Review, put the issue clearly: Some individuals and institutions use derivative securities to circumvent (sometimes self-imposed) restrictions on holdings. For instance, the investment committee of a pension fund or insurance company may require all investments to be denominated in the domestic currency. While this rule would prohibit direct foreign capital market holdings, the managers of these investments could gain exposure to foreign debt or equity markets through correlation products such as diff swaps or quanto swaps.33
(“Diff swaps” were based on the difference between two interest rates; P&G’s second swap with Bankers Trust—the one based on U.S. and German interest rates—was a typical example.)
Eventually, the Quanto market, too, became competitive. Banks did billions of dollars of such trades. The U.S. government-supported agencies also got involved, as issuers. Pension funds, insurance companies, and other regulated institutional investors bought Quantos, and—as with other structured notes—did not disclose the risks. Investors had no way of knowing whether they owned companies that speculated using Quantos or similar instruments.
A fund manager considering Quantos faced enormous temptation. Consider this deal: in January 1992, the Student Loan Marketing Association sold $150 million of three-year notes with a Quanto-style interest rate. The rate was twice the Swiss franc version of LIBOR minus twice the U.S. version of LIBOR. As was typical of a Quanto, the interest rates were in different currencies, but payments would be only in U.S. dollars.34
Why would a fund manager be tempted to buy this structured note? The first interest rate was set at 7.6 percent, an incredible teaser, given that U.S. interest rates were very low and European interest rates were falling. The note had a three-year maturity; but, for a fund manager focused on the upcoming year, the first-year return would outperform just about any other fund (perhaps except funds that were buying similar notes). It wasn’t clear what would happen in years two and three. But, by then, the fund manager would have a stellar reputation and likely would have opportunities to leave for other firms.
Fund managers loaded up. And when they tired of correlation trades based on European currencies—which were converging as part of the planned move to a single currency—they began doing trades based on more exotic interest rates and currencies, including those of Hungary, Mexico, and Brazil. No computer program could model the correlation risks in those markets.
The second major innovation at CSFP—after structured notes such as Quantos—involved structured finance, the repackaging of financial assets to reallocate risks and obtain higher credit ratings. Structured finance generated great benefits for many institutions, enabling banks to repackage and sell off their exposure to home mortgages, credit-card loans, and other assets. Before the mid-1980s, companies had done factoring transactions, in which they sold their receivables—rights to payments owed by other parties. Examples included uncollected debts or the rights to various lease payments. Structured finance was simply a modern version of factoring.
Although economists applauded structured finance, the practice was—at its core—contrary to economic theory. Why would investors pay more for repackaged receivables than the amount the companies holding those receivables thought they were worth? Did investors really prize these new financial assets so much that they would value them more highly than the company that was owed the money, which invariably was in a better position to manage the risks associated with the loan? Economists long had argued that the lowest-cost manager of risk would (and should) bear it.35 Why would an individual investor buy the rights to car-lease payments from General Motors, when General Motors obviously was in a much better position to be sure that the people leasing cars continued to pay? According to economic theory, if General Motors was in a position to extract the greatest value from leases, it would (and should) hold and monitor them.
Yet investors, nevertheless, began buying deals that enabled them to take on the risks associated with receivables owed to other companies. The companies were happy to rid themselves of the risks, and investors got a new set of investments, potentially with lower risk and higher returns than the available alternatives, such as government bonds. For example, banks began issuing securities backed by auto loans or credit-card payments. Lea
sing companies sold their risks to investors. The deals were complex, but essentially they involved transferring the rights to payments to a newly created company or trust, which issued bonds backed by the rights. First Boston was a leader in these markets, and did the first deal with a financial affiliate of a car manufacturer, General Motors Acceptance Corporation (GMAC), in 1985.36 Volvo later did a $300 million lease deal with First Boston.37 Sperry Lease Finance Corp. sold securities backed by a portfolio of equipment leases.38 And so forth.
Instead of buying bonds, investors could now buy shares of a trust backed by credit-card payments. A bank would create the trust, and payments by people who held that bank’s credit cards would flow into the trust, instead of to the bank. The legal documents establishing the trust provided for a cushion of credit-card payments, in case people defaulted or delayed payment for several months. Unlike the payments of a bond, which were backed by the issuer of the bond, the payments of the trust were backed only by the trust’s assets. In addition, investors in the trust were protected from the bank’s financial problems: even if the bank filed for bankruptcy protection, the trust’s assets would be secure. Because the trust was isolated from the bank, it was more attractive to investors.
One of the main reasons investors rushed to buy these new structured-finance instruments was that the trusts typically received very high credit ratings. For example, GMAC’s bonds were rated in the second highest category by Moody’s and S&P. That meant most institutional investors could buy them, and because the bonds paid a higher return than comparably rated investments, most institutional investors wanted to buy them. They were a new form of investment with a high return, and the demand for these instruments outweighed the fact that investors were not in a good position to monitor the various obligations. Even if General Motors was in the best position to ensure that car leases were repaid, the investors perceived greater value from a new, highly rated investment than the benefit of having General Motors monitor the leases.
The process of transferring receivables to a new company and issuing new bonds became known as securitization, which became a major part of the structured-finance industry. After 1985, the race was on to securitize every asset bankers could find.
First Boston had entered this market years before Allen Wheat arrived. The investment bank securitized credit-card payments, equipment-lease payments, auto loans, commercial mortgages, and other instruments. Other banks also did these deals, which became more exotic over time. Eventually, banks would securitize every contractual right imaginable, from repackaged Latin American debt to David Bowie’s rights to royalty payments from his rock albums.
One of the most significant innovations in structured finance was a deal called the Collateralized Bond Obligation, or CBO. CBOs are one of the threads that run through the past two decades of financial markets, ranging from Michael Milken to First Boston to Enron to Lehman Brothers and AIG. CBOs would mutate into various types of credit derivatives—financial instruments tied to the creditworthiness of companies—which would play an important role in the aftermath of the collapse of numerous companies. Although it wasn’t apparent during the early 1990s, even the relatively primitive CBO deals developed at First Boston posed serious regulatory questions, especially related to accounting.
In simple terms, here is how a CBO works. A bank transfers a portfolio of junk bonds to a Special Purpose Entity, typically a newly created company, partnership, or trust domiciled in a balmy tax haven, such as the Cayman Islands. This entity then issues several securities, backed by the bonds, effectively splitting the junk bonds into pieces. Investors (hopefully) buy the pieces.
In a simple CBO, there were three pieces, divided according to risk, and payments on the bonds held by the SPE flowed to the various pieces in order of seniority, just as credit-card payments had flowed to investors in the GMAC deal. The senior piece was paid first, and had a yield of, say, one percent more than government bonds. Once the senior piece had been paid in full, the mezzanine piece would be paid next, and had a yield of perhaps three percent more than government bonds. The bottom piece would be paid last—after both the senior and mezzanine pieces were paid in full—and received whatever was left, an unspecified yield that would be very high if the junk bonds didn’t default, but could be zero if they did.
The top two pieces were easy to sell, because the rating agencies gave them favorable ratings. Typically, the senior piece would receive one of the very top ratings; the mezzanine piece would receive a lower, but still respectable, investment-grade rating. These pieces had higher yields than any comparably rated investment, so fund managers who could only buy highly rated bonds could make more buying CBO senior pieces than in almost any other way.
The lowest-rated piece was the most difficult to sell. Sometimes, the owner of the junk bonds would keep this junior piece. But more often, an investment bank would need to sell it. This was where First Boston’s sales force stepped in. If a salesman could sell a junior piece, that sale would drive a huge CBO deal, which was very profitable for the bank. One deal typically generated several million dollars in fees.
CBOs were the brainchild of Fred Carr, the head of an insurance company called First Executive Corp., and one of Michael Milken’s biggest customers during the 1980s. CBOs were based on the same logic Milken had applied to the junk-bond market: low-rated bonds outperformed less risky bonds, in part because of the legal rules that made low-rated bonds less attractive, and therefore cheaper.
In a CBO, the sum of the pieces was more than the whole, like a nuclear reaction that released energy by splitting apart atoms. Investors were willing to pay more for the three pieces separately than they would pay for the underlying junk bonds on their own. The credit ratings “added value” to the bonds, by enabling institutional investors to buy pieces of them, which they otherwise could not have done. CBOs were a kind of high-finance fission: top credit ratings made the two safest pieces so attractive that a bank could unlock hidden value in junk bonds simply by dividing them.
The first CBO was TriCapital Ltd., a $420 million deal sold in July 1988. There were about $900 million of CBOs in 1988, and almost $3 billion in 1989.39 Notwithstanding the bad press junk bonds had been getting, analysts from all three of the credit-rating agencies began pushing CBOs. They were very profitable for the rating agencies, which received fees for rating the various pieces.
Fred Carr did the most ingenious CBO of all in 1989. When regulators began requiring First Executive to keep financial reserves to back all of its bonds, Carr responded by converting his junk bonds into CBOs. But instead of selling the pieces for more than the bonds’ value, First Executive kept all of the pieces.
Why? Once the top two pieces were highly rated, Carr could argue that First Executive no longer needed to keep financial reserves to back them. If only the most junior piece appeared to be risky, only that piece required full reserves. It was as if the owner of a three-story house had claimed it was really three pieces with only the ground floor subject to property tax. Yet the regulators bought the argument, and Carr saved First Executive $110 million of reserves .40
Of course, First Executive denied that the purpose was to avoid regulatory requirements. It was doing CBOs simply because the market valued the pieces more highly than the whole. But that begged the question: should the bonds really be worth more packaged in a CBO than they were worth on their own?
Fred Carr’s idea spread to Thomas Spiegel, the founder of Columbia Savings & Loan, a client of both Michael Milken and First Boston. Columbia, based in Beverly Hills, held billions of dollars of junk bonds. When Spiegel became embroiled in the savings-and-loan fiasco of the late 1980s, he resigned, and began working with First Boston on a CBO of Columbia’s junk bonds.41 (Meanwhile, Spiegel was delighted when a federal jury in Los Angeles acquitted him—he even hugged his attorneys.)
First Boston gained expertise on this deal, and then began setting up more CBOs based on a wide range of financial assets, including Brady
bonds—the restructured debt of several Latin American countries. CBOs were good for First Boston, not only because they generated fees, but also because the bank made money buying and selling the junk bonds that went into them. In 1990, First Boston raised $255 million for a CBO called Delaware Management Co., and used the proceeds to buy junk bonds, buoying the market.42
Not surprisingly, some CBOs didn’t do well, especially during 1990. After several of the fifty junk bonds held by one deal, CBC Holdings, either defaulted or were downgraded, Moody’s announced it was considering a possible downgrade.43 But the rating agencies were reluctant to downgrade these deals, because the consequences were so dire. For CBC Holdings alone, $200 million of bonds carried high Moody’s ratings, and if Moody’s downgraded the bonds too much—and investors had to sell them—the CBO and junk-bond markets might enter a death spiral. (In 2002, several corporate defaults would lead to just such risks; more on that in Chapter 11.)
With the various types of structured-finance deals, a trend began of companies using Special Purpose Entities to hide risks. From an accounting perspective, the key question was whether a company that owned particular financial assets needed to disclose those assets in its financial statements even after it transferred them to an SPE. Just as derivatives dealers had argued that swaps should not be included on their balance sheets, financial companies began arguing that their interests in SPEs did not need to be disclosed.
The Financial Accounting Standards Board began debating the SPE question during the late 1980s. The issue was controversial and was referred to the group’s Emerging Issues Task Force, which decided in 1990 that if an outside investor made a substantial investment in, and controlled, the SPE, then the other assets and liabilities of the SPE did not need to be included on a firm’s balance sheet. In 1991, the acting chief accountant of the SEC, concerned that companies might abuse this accounting standard, wrote a letter saying the outside investment had to be at least three percent.
Infectious Greed Page 10