Why didn't the sophisticated, computerized pricing models that Wall Street firms use to predict returns and risk for complex derivatives save them from the sub-prime mortgage mess? The short answer is: Fund and portfolio managers rarely use them.
The long answer, of course, is more complicated. Many factors have contributed to the crash of hedge funds that invested in collateralized debt obligations (CDOs), including the irresponsible lending practices of mortgage brokers that made no-money-down, balloon and adjustable rate mortgages to people who couldn’t afford them; the reassuringly high ratings on instruments that have up until recently been provided by Moody’s, Fitch and Standard & Poor’s; the highly leveraged nature of many hedge funds; the unrelenting pressure to obtain high returns; and the herd mentality -- if everyone else is doing it, including seasoned veterans who have managed winning funds for years, it must be OK. As Igor Hlivka, director of the rates trading group at Mitsubishi UFJ Securities, observes, "The market is driven by fear and greed. It is not driven by realistic assumptions."
Igor Hlivka, Cohead, Quantitative Analytics Group, Mitsubishi UFJ Securities
Photo by John Rogers/Getty Images
According to experts, the mathematical models that are used extensively by the sell side to calculate the prices of complex instruments are often overlooked by the buy side, which reaches for double-digit implied returns without investing too much time and money on boring old math. When the buy side does use models, it’s typically to confirm prices rather than challenge them. Sometimes the wrong models are used, sometimes they’re misunderstood and other times the wrong assumptions cause the wrong data to be input into models.
Although economists have been issuing warnings for years of a looming sub-prime mortgage crisis that could potentially have been factored into financial models, few chose to do so until it was too late. After all, financial models don’t replace market savvy or risk policies; they merely help execute those already in place.
With this in mind, what are the possibilities -- and limitations -- of credit derivatives modeling? And how can modeling technology help investment firms avoid getting hit so hard by major market shifts in the future?
Complexities of CDO Pricing
If you wanted to determine the price of a fruit salad, you could find out the cost of each type of fruit and multiply by the quantities you need. But gauging the price of a slice of blueberry-peach torte to be served at a five-star restaurant four years from now would require many more factors to be considered, some of which would be nearly unpredictable (Will the local economy be strong? Will blueberries and peaches be in favor?). Imagine the peaches and blueberries represent slices of aggregated loans and the restaurant is really an issuer of CDOs, and the difficulty of the task becomes more clear.
A CDO is essentially an insurance contract through which a buyer takes on the risk that a portion of a pool of debt will default. To create a CDO, a firm takes in loans or mortgages and artificially, through financial engineering, mixes them and slices them, serving them up as tranches of synthetic bonds, which are then assigned a rating and considered equivalent to another corporate bond of the same rating. Thus the CDO issuer passes the credit risk of a portfolio of debt securities over to investors, who in return receive steady income until expiration or default. (A synthetic CDO is written on credit default swap contracts instead of actual debt.) Not only are these CDO tranches thinly traded, which means there’s little market data to mark to, but it is difficult to accurately predict when or whether the borrowers of a selection of loans will default.
In a typical scenario, an aggressive salesman at the CDO-issuing firm offers CDO tranches with an implied rate of return of perhaps 15 percent or more. The portfolio manager, experienced though he may be, doesn’t have the wherewithal to evaluate this investment because there’s so little reference data by which to go. And committing a financial engineer to spend two or three months to build a model that reflects fair pricing is often deemed too expensive.
In fact, industry observers say Bear Stearns, whose CDO-laden hedge funds were recently wiped out in the sub-prime mortgage market downturn, and other hedge fund managers don’t tend to base decisions on sophisticated statistical analysis. Rather, according to reports, Bear Stearns’ hedge fund managers relied heavily on credit ratings and hedging strategies. It has been reported that Ralph Cioffi, the 22-year Bear Stearns veteran who managed the two funds, hoped to minimize risk by investing only in the top-rated portions of CDOs, which have traditionally received AAA ratings from Moody’s, Fitch and Standard & Poor’s credit-rating agencies. "In general, hedge funds either don’t use modeling tools or they just use them to prove to themselves that they will get a certain return using very conservative functions and interest rate options," says Rajeev Seth, president of Beat Index, a quantitative investment research consulting firm.