What to do about CDO ratings?

One of the many causes of the financial crisis was excessive reliance on excessively generous ratings of novel financial instruments. There are proposals to change the incentives faced by the bond rating agencies to prevent them from being so lax. I don’t think that any such reform is likely to be successful. I think it would be better to rely on something other than the ratings. I explain my thoughts after the jump.

There is no doubt that part of the reason for the unreliable over generous ratings was conflict of interests. The bond ratings agencies also consulted earning fees for helping to design the instruments they went on to rate. This is unacceptable and should be banned. Beyond that, many instruments were rated by only one of the big three – Moody’s , Fitch and S&P . This gave each an incentive to be generous in order to attract business. Maybe, it is possible to eliminate this incentive by requiring clients of any of the three to pay for ratings from all three. Since this is good for Moody’s, Fitch and S&P it wouldn’t be an unconstitutional bill of attainder (such a provision would be challenged and the Supreme Court would make the final decision).

As I argued here, the agencies’ incentives would still be unacceptably distorted. They have to power to decide if whole sectors of finance exist or not. If they had procedures such that no CDO tranche were ever rated AAA, then the volume of CDOs would have been a small fraction of what it was. Even without competing for business, the agencies have an incentive to help new types of structured finance grow, since more assets means more assets to rate and more income.

I think it is clear that procedures such that no tranche of any CDO got AAA were and are appropriate given the definition of AAA (almost certain full payment on time) and uncertainties concerning CDOs. Basically if an agency doesn’t fully understand the properties of a new instrument (which can only be determined with, you know, data on the instrument) and is 99% sure it should be rated AAA, then the agency shouldn’t rate it AAA. 99% is much too low a number to be associated in any way with AAA.

With my total lack of authority and expertise, I declare that no instrument should be rated AAA unless an instrument which was identical (except for maturity dates) paid in full and on time through the last two recessions. That means no AAA structured financial instruments. I think it is also clear that ratings agencies will never adopt the procedures which I think are reasonable, since that would kill the goose that gives them golden eggs.

The problem is simple. The ratings agencies were providing an immensely valuable service for very modest fees. When I first learned of their existence I was amazed that their power hadn’t corrupted them. We were very very lucky to have reliable rating’s agencies. In particular, when the ratings were explicitly given legal force by, among other things, Basel I, impossible strain was put on the rating agencies. Then the rating mattered even if sophisticated investors didn’t trust it, since it mattered because of a simple formula in a regulation.

Obviously you can’t expect a modestly paid institution with such power to resist temptation. The ratings agencies were like, say, bureaucrats in the Russian ministry of privatization. We should consider it normal that we don’t trust them and not hope to get reliable ratings for almost nothing. We were very lucky to have reliable ratings for decades, but our luck has run out.

So I think that we can’t rely on ratings. This means that capital adequacy standards and prudential regulations must be based on something else. I propose that we put our trust in the Vampire Squid et al – in the broker-dealers which did much more than any ratings agency to destroy the world economy. This is not a joke.
First assume that the CDS market is transparent with full disclosure and mandatory trading on exchanges. We learn a lot if an investment bank is willing to write CDS charging say 10 basis points per year vs the investment bank is only willing to write the CDS if it can get 100 basis points per year. The CDS writer has skin in the game. The price of a CDS will be wrong, markets aren’t efficient, but it won’t be as systematically biased in favor of new products as the ratings.
I think that the prices of CDSs are better indicators than ratings provided that counter party risk is modest. That is an investment grade asset must be an asset whose CDS written by a bank with adequate capital costs less than x basis points per year. Hmm I suppose that this creates interesting opportunities for market manipulation. Well something like that.

The point is that we can’t rely on the ratings agencies which have no skin in the game and which put only their reputations at risk. Their reputations are shot. They have nothing left to lose. They must be replaced by agents with skin in the game in prudential regulations.