The Subprime Market and Derivatives Once Again...

A correspondent makes three points:

• First, Moody's response to requests to rate highly complex structured mortgage derivatives should have been: "We don't have enough historical data on securities like this in a housing price environment where rent-versus-buy ratios are this low. We cannot rate them to our usual standards."
• Second, if you are collecting a fee for rating things, and if your fee is based on your past reputation for providing good ratings based on long-term historical data, and if you use the same classification scheme to make ratings that are not based on sufficient historical data--then you do have a problem.
• Third, the fact that there was an ABX index and thus an easy way for people to bet that the mortage-backed securities market would crash probably cut short the bubble--the true hedge funds were stabilizing speculators; the destabilizing speculators were (i) the funds that were long CDOs and (ii) the banks and other issuers who retained the CDOs because their portfolio managers believed their marketeers. A world without derivatives but with mortgage-backed securities would probably be a world in which we have a bigger problem than we have now.

This third point is very much Sandy Grossman's point about the 1987 crash: that the problem with "portfolio insurance" was that it was a trading strategy rather than a derivative security traded by a market, and so nobody knew how large the demand for it was, and so those planning to implement the trading strategy in a market downturn had no way of assessing how expensive implementing it was going to be. On Black Monday everybody found out.

Comments

[Bunbury]

Your correspondent has not exhausted the limitations that might have prevented Moody's producing effective ratings, even the ones that they should have known, but where does the "should" come from? How would the suggested response have maximised profits?

The second bullet suggests an answer but it's those ratings that people would pay for and there is no reason why they shouldn't have applied the same kind of rating to such products. There isn't a clear line between different kinds of bonds: if you wouldn't rate a CDO, would you rate an MBS? If you wouldn't rate an ABS would you rate a bank? Is rating a bank really simpler than rating an MBS? If you don't know enough about MBS to rate a CDO, how can you rate an MBS? The ratings agency always has a problem if it makes a mistake. There is never enough historical data in financial markets. The calculation involves turning down money now to avoid a hard to quantify downside. There's competition so the situation is full of winner's curses and tragic commons.

The implicit suggestion that there was an obvious deficiency in historical data is a red herring. There was far more data and transparency for mortgages and MBS than there ever is for single name corporate data.

The third point is valid but not obviously the full conclusion. Did the ABX burst the bubble early or just lubricate the way down? Don't really know. Hedge funds are on both sides of this balance sheet. Contrasting hedge funds favourably with investors willing to accept returns from a black box requires very nimble footwork.

The size of the securitised mortgage market is about as visible as it gets and the recovery from the 1987 crash was swift and left little collateral damage so following the Grossman reasoning current conditions are more favourable and the outcome looks worse so either the credit crunch is actually a triumph of the market or the search for a preventive for crashes will have to go on.

[Ken Houghton]

First, Moody's response to requests to rate highly complex structured mortgage derivatives should have been: "We don't have enough historical data on securities like this in a housing price environment where rent-versus-buy ratios are this low. We cannot rate them to our usual standards."

And that's where (s?)he goes astray. They're not usual securities, so you don't use the "usual" standards. And you don't look at macro changes as anything other than "what the market situation is." (Surely you, an economist, could tell your correspondent that if the ratios are that low, it is because markets are clearing at that level.)

The rating consideration is "given the market, how secure should the income streams of these securities be"? If the security is well overcollateralised, it would take a major tremor to place the payment in jeopardy. There is little choice, in those circumstances, but to run the models, view the stresses that would be necessary, and assign a AAA rating.

If you want to make the analogy complete, the second week of October of 1987 saw the pending launch of the first AAA rated "junk bond." It was, iirc, between 440-449% overcollateralised. The failure rate would have to have been over 75% for the payments not to be made.

People pretending that AAA means anything other than "this bond is highly likely, at the time of this rating (usually, on issue), to be able to make its scheduled payments on time"--without regard to HOW those payments will be made--are looking for a scapegoat, not a villain.

[dsquared]

I agree with the first commenter; point 1 is totally weak. Every "environment" has something unusual about it that can later be used as an excuse for mistakes of analysis. If Moody's were to say "we can give you a credit rating on any sorts of securities, except ones that have a risk of going bust" then why would they have a business?

[Gary]

In regards to the 3rd point, I think you're right on (I disagree with Felix on this one). http://shittingalpha.wordpress.com/2008/04/30/abx-and-the-market-for-rmbs/