"It could be structured by cows and we would rate it." Photographer: Krisztian Bocsi/Bloomberg
"It could be structured by cows and we would rate it." Photographer: Krisztian Bocsi/Bloomberg

A popular narrative of the ratings agencies' role in the financial crisis goes something like:

  • banks paid ratings agencies a lot of money to rate mortgage-backed securities,
  • so the ratings agencies gave a lot of those securities undeserved AAA ratings,
  • so a lot of investors were deceived into thinking those securities were safe,
  • so the investors bought a lot of those securities and eventually blew up.

This is the theory under which the investors, and the U.S. government, have gone and sued the ratings agencies, reasoning that the agencies' job was to tell the investors what not to invest in, and that the agencies did not do their job.

There are two objections to that story. The weaker objection is: Ratings agencies aren't just looking for a quick buck. Their whole business is reputational; the only reason their ratings are worth anything is because the market trusts them. If the agencies lost the trust of investors, then they wouldn't be able to charge for ratings, so they have every incentive to get the rating right even if their high-paying issuer is pressuring them to get it wrong.

Rating the Raters

If you believe that, email me, because I have an assortment of financial products to sell you. There is some theoretical truth to it, but pretty much anywhere you look you can find people damaging their reputation for short-term gain. The whole point of a reputation is that you can amortize it for short-term gain. In the long run we are all etc.

The stronger objection is: Ratings agencies didn't just give securities AAA ratings because the issuers of those securities wanted them. They gave them AAA ratings because the buyers of those securities wanted them. Buyers -- at least the big institutional price-setting buyers -- can figure out how risky securities are, either by doing their own credit analysis or by just looking at the yield. A thing that yields a lot more than Treasuries1 is riskier than Treasuries, even if it's rated AAA, because risk and reward go together.

On this theory, investors rely on ratings not to tell them how risky things are, but to tell them what things they're allowed to invest in. Investors with a charter mandate to invest in investment-grade securities have to invest in things with investment-grade ratings. Investors who are subject to ratings-driven capital regulation have to invest in things with ratings that satisfy their capital regulators. Add in a fairly straightforward set of principal-agent problems2 and you get the conclusion that investors want to invest in the riskiest thing that satisfies their ratings requirements.

This objection relies on the assumption that investors are not idiots, or at least, that the marginal investor is not an idiot.3 This assumption seems reasonable to me! But your mileage may vary.

Here is a new National Bureau of Economic Research working paper (free earlier version here) by Harold Cole of UPenn and Thomas F. Cooley of NYU Stern (and S&P!) that basically says that, but with math:

In this paper we show how credit ratings have value in equilibrium and how reputation insures that, in equilibrium, ratings will reflect sound assessments of credit worthiness. There will always be an information distortion because of the fact that purchasers of ratings need not reveal them. We argue that regulatory reliance on ratings and the increasing importance of risk-weighted capital in prudential regulation have more likely contributed to distorted ratings than the matter of who pays for them. In this respect, much of the regulatory obsession with the conflict created by issuers paying for ratings is a distraction.

I've been saying that for a while, but without the math. If you'd like the math, you know, be my guest. If you don't like this story, I don't know that the math will convince you; it's entirely theoretical. But I find it persuasive. The story implies that inflated ratings were driven by investor desire for riskier, but still highly rated, assets:

The increasing reliance on ratings in order to risk weight assets for prudential regulation purposes increased the demand for "safe" assets - assets that met regulators standards. ... Our model suggests that investors in this restricted class would have been interested in holding riskier securities that passed the threshold because they would provide ex-post higher returns. The experience of 2007-2009 suggests that restricted investors continued to invest in the claims rated inaccurately even after it became clear to them that some ratings did not perform as expected.

The authors conclude from this that the "issuer-pays" model of ratings is not a real problem,4 and that regulators should focus on eliminating reliance on ratings for regulatory purposes.

And why not.5 This theory also helps clear up some crisis-y mysteries. My View colleague Jonathan Weil has pointed out how odd it is that the government's lawsuit against Standard & Poor's blames S&P for defrauding some of the banks that were selling the securities that S&P allegedly mis-rated. And it is odd to look at the facts -- Citigroup creates bond, Citigroup asks S&P for a rating, S&P rates bond AAA, Citigroup buys bond itself, bond goes bust -- and conclude that S&P ripped off an innocent bank. But it's easier to conclude that Citigroup wanted to take more risks, but was constrained by regulators to invest mostly in highly-rated securities, and that S&P helped it get around those constraints by giving risky securities high ratings. A decent model of ratings agencies' role in the financial crisis is that they helped banks rip themselves off.


1 Now Aaa/AA+, but you know what I mean.

2 You know, if your returns are good, you get a big bonus; if your returns are bad, you don't have to pay it back, that sort of thing. Basically anyone managing money for anyone other than themselves has some sort of call option on the returns and so should want to maximize volatility, which is why there are rules (investment-grade mandates, capital and prudential regulation for insured banks, etc.) designed to limit volatility.

3 My favorite financial product, the CPDO, was a piece of pure ratings gaming that was marketed to poor hopeless Australian regional councils but was clearly designed for investors who wanted to game ratings. Its marketing documents said clearly enough (enough for me, but not for the Australian councils) that it was risky but AAA-rated.

4 Meh. Empirically there is some evidence the other way.

5 Oh I'll tell you why not. See footnote 2: Banks with insured deposits and limited shareholder/employee liability have incentives to maximize risk. You need a thing to limit that risk. That thing will be gamed. If the thing is ratings-agency ratings, or internal ratings, or internal models, or raw leverage ratios, or whatever, the gaming will be different, but it is not obvious that ratings-agency ratings are the easiest alternative to game.

To contact the writer of this article: Matt Levine at mlevine51@bloomberg.net.

To contact the editor responsible for this article: Tobin Harshaw at tharshaw@bloomberg.net.