Earlier, Megan posted a section from a suggestion from Joe Wiesenthal at Clusterstock on how to fix the rating agencies. He suggests a group of 10 agencies, arbitrarily chosen to rate deals. Since I have some pretty strong opinions about the rating agencies, I thought I might give this proposal some thought.
Here's Wiesenthal's solution to the rating agency problem:
You create a pool of 10 companies licensed to rate debt. When an issuer wants to bring a security of some sort to market, they tell some central body, and a rater is selected at random from the 10. There's no changing it once a name is selected. Thus the debt issuer can't go ratings-agency shopping if they're worried about what kind of ratings they can get.
If a debt issuer isn't happy with who they got, then, well, too bad. Over time, you'd give companies that showed a good track record a heavier weight in the pool, so that they're selected more often. Their only goal would be to increase market share by being accurate. Pandering to either buyers or sellers would be 100% impossible.
Now granted, it wouldn't be perfect. Performance measures would be backwards looking, and you'd probably end up with companies that had gotten lazy, and stuck to old ideas about how to rate debt, but that's just life. They'd lose their weighting in the pool, and eventually you could even put companies on probation if they got bad enough.
Nothing's going to change the fact that incumbents grow dumb and slow -- but at least they'd have an incentive to avoid that, whereas currently they don't (have the top raters lost any market share? No.)
I think to evaluate this solution we should consider whether it would have solved the problem that we ran into the past few years. I don't believe it would have.
As Wiesenthal admits, the performance measures would be backwards looking. Prior to 2007, the three rating agencies actually had a very good track record. There was no reason to believe that their ratings would suddenly collapse. So how can we have any certainty that those agencies with the highest market share out of the chosen 10 would have done any better than Moodys, S&P and Fitch?
I also think it's generally agreed (or should be) that it was not all fraud at the agencies. Some of the failure in their ratings was good, old fashioned, poor judgment. There's little doubt that most of these rating analysts actually believed in the ratings they gave. That could be due in part to the fact that the ratings agencies tend not to attract the brightest minds on Wall Street. Certainly, there are some very smart individuals that choose to work at these firms, but the smartest of the bunch flee to investment banks, where they can do essentially the same work and earn several times more pay. As a result, I think that the agencies currently have trouble staffing three agencies with the necessarily talent. So you could imagine how difficult it would be to create 10.
The incumbent problem is a serious one. I would worry that we'd just end up with the group earning the most market share becoming too cozy with bankers and don't bother to challenge the status quo with their ratings -- just like before. Those earning less market share would consequently have less money to hire better talent that might be in a position to pose such challenges.
I think as long as you have a sort of rating agency club, whether three or 10, many of the same oligopoly problems will still persist. I think this proposal is a step in the right direction, but doesn't quite go far enough. That's why I think the best ways to really reform the system would involve opening up the market for ratings. Wiesenthal notes that the big problem with this is capital requirements. They depend on ratings. If there aren't standardized ratings, then you can't have standardized capital requirements.
I don't think that's entirely true. You would just need to change the way capital requirements are defined. Instead, define them based on formulas utilizing collateral attributes (for secured debt) or firm metrics (for unsecured debt). For example, maybe you could instead require capital based on underlying asset quality and credit enhancement for asset-backed securities. That's just one suggestion, but there are certainly ways other than pure ratings to structure capital requirements. In fact, recent shenanigans involving re-remics seem to support the call for changing how these requirements work.
I still like the idea that investors ultimately need to take the most responsibility for understanding the risk of the bonds they purchase. They have the most incentive to get it right. That means they would naturally be the first to begin to question a security if something fishy is going on.