That is doubly odd when one remembers the Enron saga. As Enrons creativity with its balancesheet blew up in its face, attention justifiably shifted to their accountants at Arthur Andersen, who had not responded to the problem in time; Andersen, while still nominally in business, has never recovered.
Have CRAs been hit half as hard No. True, a few months ago the SEC said that they had done a bad job rating complex investment vehicles, and announced new rules reducing funds reliance on them. The CRAs didnt raise a fussbecause the funds, with Vanguard in the vanguard, were perfectly willing to raise it for them.
How did the CRAs screw up In the 1970s, they shifted from charging buyers of their ratings to charging the firms they were rating. (This was simultaneous with, and probably caused by, US regulators making their function quasi-official.) That may be okay when hundreds of corporations are coming to you to rate their debt. When, however, their business changed and they started having to rate collaterised debt obligations (CDOs) for a small group of giant broker-dealerslike Lehman or Merrillit is difficult to deny the incentives to cozy up to the major source of your income.
The CRAs insist that they had processes in place to minimise the conflict of interest. The numbers, however, tell a different story. Considerable evidence exists that ratings downgrades arent timely that they are anticipated by credit-default swap (CDS) spreads. In other words, participants in those markets already know what CRAs are supposed to be telling them. One study found that 42.6% were for instruments already heavily traded in the CDS market the previous month.1 The CRAs claim that they are responding to a need in the market for stable ratings. Holes can be shot in this, however. For example, examining the actions of KMV, a small agency that relied on user-fees rather than being paid by those it was rating (and since taken over by Moodys) found that 75% of the S&P rating changes were significantly anticipated by KMV more than a year beforehand2. The bigger CRAs were clearly captured.
Even more, a giant gap appears to exist between solicited and unsolicited ratings. Solicited ratings are higher than can be accounted for by selection bias. Meanwhile, the agencies continued to ignore the vast differences between corporate bonds their traditional businessand CDOs. They did not change their mechanisms of evaluation; they did not actually examine the quality of the paperwork underlying the mortgages; and they ignored the fact that real estate, much more than other sectors, is prone to contagion. (The latter is now being justified as un-forecastable market risk, which must have most real estate analysts laughing in the aisles.) The only change they made was to quietly up the acceptable default risk for investment-grade CDOs. Joseph Mason of Drexel University quotes a Nomura handbook: Suppose you have a seven-year ABS rated AA with an idealized default probability of 0.168%. If we call the repackaged instrument a CDO, it ought to be rated AAA because the idealized default rate for AAA-rated CDOs is 0.285%. This sort of ratings inflation permitted those legally required to hold AAAs to pack their holdings with risker, higher-yield instruments. When Mason compared corporate bonds and CDOs rated BAA, he found that the CDOs defaulted eight times more. When we add that CRAs in any case took 3-7 weeks to update ratings of CDOs in response to a change in underlying ratings, we clearly have a combination of undisclosed risk, inefficiency and conflict of interest that should see us all howling for their heads.
The UKs chancellor, Alistair Darling, told the Labour partys conference this week that global reform of CRAs is necessary; the head of South Africas reserve bank told MPs that he now had no confidence in rating agencies. In India, some CRAs have already moved to forestall some criticism: although 45% of Icras income comes from non-ratings work, Care, for example, claims to have spun off its consulting business to subsidiaries. Meanwhile, Sebi has released guidelines that real-estate funds must be valued by two CRAs, and they must alternate ratings. Still, this is in anticipation of a domestic securitisation market that hasnt developed yet. One day it will. When that day comes, the lessons are clear: trust CRAs only if conflict-of-interest problems have been worked outand read the small print.
1 Hull, Predescu and White, The Relationship Between Credit Default Swap Spreads, Bond Yields, and Credit Rating Announcements, Journal of Banking & Finance, 28:11
2 Robbe, Paul and Mahieu, Ronald J., Are the Standards Too Poor, 2005