Editorial - Actuarial Aspects of the Recent Financial Market Crisis
Actuarial Aspects of the Recent Financial Market Crisis
Securitization of mortgage pools was one of the great financial innovations of modern times. Collateralized mortgage obligations (CMO), which took this innovation one step further, are a classic example of what we understand today by the term "financial engineering." Together, these instruments brought liquidity to the real estate mortgage market by attracting new sources of funding for real estate loans. Securitization allows any investor to readily get into and out of the mortgage market, gain the benefits of diversification of large mortgage pools and select their degree of risk. In turn, this has reduced the cost of borrowing for tens of millions of homeowners and expanded the wealth of Americans.
As actuaries, we are familiar with the notion that the value of a financial instrument depends on the "level" and "timing" of its cash flows. For debt–type instruments, "level" generally pertains to credit risk, i.e., the likelihood that the borrower has sufficient means to repay principal and interest per the terms of the contract. "Timing" pertains to uncertainty in when the cash flows will occur, due to "flexibility" that is contractually granted to the borrower.
Normally, a mortgage is subject to both risks, i.e., with respect to "level" as well as "timing." But traditionally, mortgage backed securities (MBS) and CMO did not contain "level" risk, since the underlying collateral in almost all these pools is guaranteed by the federal government or a federal agency. The government sets strict requirements for the mortgages that can be securitized, mitigating the credit risk to itself. Thus, for buyers, these securities are primarily a "timing" play, involving what we call "market" risk with "credit" risk removed from the equation.
In the early years, not all buyers understood the nature of "market" risk, incurring substantial losses in some cases. Over time, they became comfortable with pre–payment probabilities and interest rate models, resulting in a robust market for these securities. While pre–payment probabilities have an actuarial flavor, for the most part, the valuation of these instruments involves concepts from modern finance theory.
More recently, the innovation in securitization has involved underlying contracts that have a substantial amount of credit risk, including so–called "sub prime" mortgages. These pools were carved up into CMO–type trenches dictating the pecking order in which cash flows are applied. The valuation of the securities generated by this process has more to do with "credit" risk, i.e., the probabilities of default, than "market" risk. The appropriate techniques to analyze and construct these derivatives resemble actuarial methods rather than option valuation methods taught in finance courses. An actuarial assessment would have revealed, of course, that the trenches that received AAA credit rating were not quite as free of default risk as the rating might suggest. In a market where spreads were hard to come by, buyers hungry for yields eagerly lapped up these securities, with predictable consequences.
Indeed, actuaries routinely construct "securities" of this type in their daily work. Aggregate stop loss layer reinsurance is a classic example of this "carving up" process. We know the sensitivity of losses to deviations from the assumed frequency and severity of the underlying portfolio of insurance contracts. We are familiar with the change in the volatility of losses to changes in the attachment points used to define the stop loss thresholds. A direct writer who wants to avoid any volatility in losses would need to set the attachment point conservatively low, and pay the corresponding higher premiums. For the reinsurer, this is risky stuff, requiring high returns.
As actuaries, we would be especially cautious in pricing the above reinsurance when the underlying insurance coverage is a "new" product, with no credible experience, as is the case with "sub prime" mortgages. We would carefully consider ruin probabilities, value–at–risk and contingent tail expectation (CTE). We would review anti–selection, i.e., the underwriting methods and channel/distribution used to sell the product. Cognizant of agency risk, we would take the rosy picture painted by the direct writer with a grain of salt.
Aware that the individual insurance contracts being reinsured are not always independent, actuaries analyzing aggregate stop loss coverage would identify sources of dependency. This often leads us to evaluate cash flows including the impact of consumer behavior under a range of stochastic economic scenarios. In the case of sub–prime loans, this would include the risk of a rising interest scenario coupled with a soft housing market where limited income homeowners would be unable either to afford the higher costs of their adjustable mortgages or avoid foreclosure by selling their homes.
It is not uncommon for a reinsurance market to "freeze," when supply totally dries up in the face of catastrophic losses, even when reinsurers understand the nature of catastrophe and go into their deals eyes–wide–open. This is especially the case when the loss distribution of the underlying contracts is perceived to shift, such as due to global warming affecting hurricane patterns, epidemics affecting contagion risk or terrorism affecting property damage. Reinsurers draw into a shell and reassess the situation, during which time it is impossible to obtain coverage at any economically feasible price. It is no surprise that at the time of writing of this editorial, the capital markets are in a deep freeze, its principals having gone into their deals eyes–wide–shut.