More than 100 people gathered in New York on Wednesday for an excellent conference organized by Structured Credit Investor. I was a panelist for a session on risk management in the structured products business. Because the conference was “off the record,” I am sorry that I can only pass on my own remarks and not the more intelligent insights of my fellow panelists.
The first point I made in my remarks was to focus on the root causes of the credit crisis. The root cause of the crisis was not the shrinkage of the securitization market or increased mortgage defaults—it was the drop in home prices that caused both of those “symptoms” to occur. As the quote from Vik Pandit to the left indicates, Citi’s problems in fact stemmed from their unintended large exposure to the macro factor “U.S. home prices.” By implication, that means that Citigroup did not accurately measure its “delta” or exposure to home prices, did not have limits on this exposure, and therefore had too much exposure by the time the problem became apparent. For this reason, we need to focus on macro factors as the root cause of risk in the structured products market and on the balance sheet as a whole.
Next, we pointed out that this issue was not just one of academic interest. The slide from the Economist magazine’s “Hall of Shame” in August 2008 showed that six of the CEOs of financial institutions with the ten largest losses had been sacked as of that date. As of today, an additional three CEOs have been sacked. In previous crises, it was often sufficient for the CEO to sack the chief risk officer in order to save his own job. The magnitude of the current crisis, however, was the equivalent of an American pro football head coach recording a record of 0 wins, 17 losses. Sacking the offensive coordinator and the defensive coordinator would not be enough to save the head coach’s job.
The implication of this 90% sacking rate for bank CEOs with large losses is that risk management, perhaps for the very first time, has become critically important to the life time income stream of the CEOs of large financial institutions. More importantly, CEOs now realize that this is a fact. My friends working in risk management are ecstatic that they now “matter” for the first time.
Then, apologizing for my cynicism, I posed the question that we addressed in this blog on www.kamakuraco.com on May 1, 2009:
Is it possible to be profitable in the business of rating new-issue structured products? I suggested that the structured products business works like this. Investment bank A assembles securities with a market value of X. They tranche them in the hopes of selling all the tranches for a total of X + Y, for a profit of Y. This is only possible if the tranches are misvalued. The best way for that to happen is to get a rating that is “too good.” So Investment Bank A goes to the rating agencies and shops for a rating. If the rating they assign is not “too good,” the structured product never comes to market. If this cynical assessment is correct, the ratings on newly issued structured products will only be “too good.” Rating agencies can only get paid for ratings that are “too good” until investors realize their bias, and then the business of rating structured products ends. It’s a lose/lose business for the rating agencies, who are on the losing end of a game of “gotcha” with Wall Street. For that reason, sooner or later, investors have to do their own risk assessment.
For that reason, it’s important to go back to Vik Pandit’s quote on the Charlie Rose show and re-emphasize the point made in the slide at the left. All crises are created by macro factor movements. From the upper left, we see movements in oil prices, correlated corporate defaults, movements in interest rates, commercial real estate prices, and home prices. These macro factor movements cause a destruction of asset values. Liability providers, even at the retail level, perceive this and they dramatically shrink the supply of liabilities to the firm. The case of Northern Rock is particularly enlightening. EVEN though Northern Rock had never reported a loss and EVEN though the Bank of England declared its support, retail depositors staged a run on the bank on September 15, 2007. As subsequent events confirmed, their assessment of a sharp decline in the value of Northern Rock assets was 100% correct.
Because every crisis follows this pattern, what investors really need is a “credit risk CAT scan” that looks through the superficial descriptions of counterparties and the nature of what they owe you (i.e. FICO score, loan to value ratio, industry, rating, type of loan, etc.). What we need is exactly what Vik Pandit said that Citi didn’t have: an accurate measure of how much the bank made or loss in market value (and net income) terms when macro factors like home prices, oil prices, the economy, commercial real estate prices, and interest rates changed. Using ratings or FICO score as a proxy for these “true deltas” is the medical equivalent of taking the temperature of a patient with heart disease. We need to focus on the cause of the crisis, not the symptoms alone.
In addition to these insights from the 2007-2009 crisis, it was also apparent that investors didn’t realize the true nature of the outcome of the tranching of structured products by first loss, second loss, third loss, and so on. It’s now obvious with hindsight that, of say 10 tranches in a credit structured product, only one of the 10 tranches will suffer a partial loss of principal. The others will either lose 100% of principal or lose nothing. Tranches by the order of credit losses are “heaven or hell” bonds. These CDOs are inconsistent with the frequently heard rating agency assertion that “An AA-rated CDO tranche has the same risk characteristics as an AA-rated bond.” For senior debt at least, it’s very rare to get zero in recovery and suffer a total loss. In the CDO world, that’s one of three possible outcomes!
We think this has some very clear implications for risk management in the structured products business going forward. First, ratings have been a “high ease of use” substitute for investor-driven credit risk assessment. Clearly, even if ratings are necessary, they haven’t been sufficiently good in this crisis. What investors need is the direct ability to assess the sensitivity of a structured product’s value, cash flows and financial accruals to changes in all macro factors, of which home prices have been front and center during the 2007-2009 crisis. In order to do this, investors need direct electronic access to loan-level or transaction level collateral data. Up until now, this access has been ferociously blocked by trustees and investment bankers. If one can’t get this access, don’t buy the product.
In addition, investors need to be able to run their own risk management tools on this data and to use a multiple models approach to risk assessment, instead of blind faith in one tool like the copula method, which took down so many market participants in the CDO market. We note here that the need for loan level data by investors may be in conflict with privacy restrictions at the banks doing the securitization. For example, consider a mortgage loan in my condominium building. Even if the investor were given the address, but not the unit number, they would be unable to assess whether a $1 million mortgage loan was prudent or not. If the unit were on the second floor, the loan would be imprudent. If the unit is the penthouse, the loan would be extremely low risk. The upshot of all of this is that investors must do their own risk assessment. If they do not, they’ll be taken to the cleaners (again). Those who have the time and the money to do the risk assessments will be larger institutions that trade in large blocks. Smaller institutions and traders in smaller blocks who don’t have the time or money to assess risk have two choices: stay out of structured products, or go in and get taken to the cleaners.
Since the former choice is by far the wiser, we think the upshot of all this will be much smaller volumes, simpler structures, and fewer and bigger investors in structured products. That’s what’s necessary in order to know in advance that the wheels are correctly fastened before one takes a leap on this motorcycle:
Donald R. van Deventer
New York, November 4, 2009