Rating agencies grew up rating corporate credit risk. The nice thing about corporate credit risk is the failures happen at least every seven years. There was a sideline on municipal credit risk, but since munis rarely defaulted, it was not very relevant.
Guess what? Moody’s, S&P, and Fitch are very good at predicting corporate default. Some new players are better still, but their ratings change more rapidly, which has pluses and minuses. They are faster to identify failing entities, but they also have more “false positives” where they signal failure, and it does not happen.
Go back to the late ’80s. The rating agencies were trying to evaluate the newly popular 401(k) investment Guaranteed Investment Contracts [GICs]. Guess what? No GICs had ever failed. What’s the right rating for all the companies offering them? If the history is so positive shouldn’t everyone be AAA?
Though some of the larger companies had corporate debt that traded, GICs were not obligations of the holding company, but of the insurance subsidiary. Not only that, unlike bonds, GICs (in most states) were policyholder obligations, not debt per se. GICs were often super-senior obligations of subsidiaries of the company.
So, how to rate them? Since there were few historical losses, and the rating agencies lacked a forward-looking view of what might happen, they rated most GICs AAA or AA.
After a few GIC-issuing companies went into insolvency, those ratings changed rapidly over the next seven years, leading to many exits by marginal players who did not default, a few defaults (with almost no losses), and a much smaller GIC industry dominated by synthetic GICs that relied upon insurance company derivatives.
Because the regulators required ratings, the rating agencies were willingly drawn in to rating structured products. With the GSEs it was easy — there is no credit risk, so they are AAA. With the non-guaranteed whole loans, it was tougher — no GSE guarantee. How to rate? This was a new product, a new risk, and so the rating agencies looked at resident mortgage default rates in the past.
Very much like the error of health insurers in the ’60s, where they tried to calculate utilization of healthcare services from the non-insured and apply that to the insured, mortgages retained by originators defaulted less frequently than those that were sold to third parties.
The rating agencies could get the math right on securitization, but could not get the parameters right. All of their loss data came from an era where lenders held onto their loans. Selling loans, and having servicing as a separable function was totally unknown in the past.
Much as those who implemented securitization were relatively farsighted, they did not take into account the agency problems involved in “selling to securitize.”
The rating agencies did the best that they could in a competitive environment, with little relevant loss data to guide them.
I suspect that they will do better before the next cycle of failures transpires.
In short, my point is this: the rating agencies, blundered with ratings on securitization. They will be better in the future, because they finally have real data to work with. They are trying to be more forward looking on sovereign issuers, with some degree of pushback. My view is that those that object more should be downgraded further, within reason. I differ from the rating agencies, because I am more skeptical, and imaginative.
“Figures don’t lie, but liars will figure.” Issuers are not objective with respect to their own debts. Rating agencies should ignore the issuers, and work off of publicly available data, lest they be subverted by the issuers, as has happened.
I don’t fault the rating agencies much with respect to sovereign ratings. They seem mostly rational to me, though there is little real default data to guide them.
That’s the crux of the issue here — they don’t have many sovereign fiat currency defaults to guide them. Does that mean the odds are low? By no means. A larger model, including political motivations, and using game theory would indicate that there are possibilities where no coalition governs that debts will be paid.
I hate the simplistic models of the Keynesians and their bastard progeny. One has to think more broadly, and consider the wide range of what might happen, because (surprise!) people/institutions aren’t always rational as economists view them.
I leave you with this: imagine that you run a rating agency, and your clients ask you to rate debts that you don’t have a good model to use. You have competitors, and they are seeking advantage as well. Add into the mix that the issuers can choose who they want. How do you react to a new class of credit? The question is a hard one.
And thus, to those who trash the rating agencies because of bad past decisions, I say, “Could you have done better, you have the benefit of hindsight?”
The rating agencies are flawed but are mostly honest dealers who used bad models on structured credit, because no one knew better. They took the risk and put forth some bad ratings.
For this reason I say to the fools who argue against the S&P downgrade of the US, “How do you know?” For once a rating agency is trying to be proactive, and they get a lot of abuse. Watch the trading in the market, it will tell you a lot more than the downgrade.