Management acts while pools go fetid: ratings and dynamics of loss distributions

I have been reading a fascinating (if long) article Where Did the Risk Go? How Misapplied Bond Ratings Cause Mortgage Backed Securities and Collateralized Debt Obligation Market Disruptions by Joseph Mason and Joshua Rosner. There is an awful lot in the document, but here I want to concentrate on one issue they raise which I had not thought about before, namely the shape and trajectory of the loss distribution through time.

Consider a corporate bond. For a holder of the bond, the (hold to maturity) loss distribution has a big lump of probability in the 95%-100% return bucket corresponding to the likelihood that they will get their money back. Then there is a gentle bell curve lower down corresponding to the distribution of recovery values. Two observations:

• As the credit quality of the corporate declines, this shape moves but typically those moves are slow. The big lump gets a little smaller and default gets a bit more likely, fattening out the curve around the expected average recovery.
  
• One of the reasons that moves are slow is that the company has management. Default is bad for these folks so they try to avoid it by altering their strategy or the capital structure and/or by asset sales. They have strategic options which they exploit, often saving the company.

Now consider a typical tranched ABS backed by a pool of collateral. Here tranching and other credit enhancement means that default is unlikely for the rated tranches. However:

• The shape of the distribution is different. In particular, the average loss given default can be much higher.
• The time evolution of the distribution is different: most static ABS pools evolve so that for a given tranche, default becomes either certain or vastly improbably.
• Thus if the pool behaves a bit better than expected, most or all of the rated tranches will be money good. Mason and Rosner say this `wastes' credit enhancement, which I don't really see, but certainly even lower tranches can become risk free in some deals fairly fast.
• On the other hand, if the pool behaves even a little worse than expected, the impact on the lower tranches can be severe. Therefore ABS downgrades, when they come, are often multiple notch downgrades.
• Note that this is partly because most ABS has no asset diversification and no time diversification: unlike a corporate, there are no strategic options for the issuer to do something that doesn't lose as much money as their current approach.

None of this means that ABS ratings are wrong, necessarily, but it does mean that the users of ratings need to understand the key differences in the time evolution of credit risk between corporates and ABS.