Floyd Norris has an article in today’s New York Times, “Market Shock: AAA Rating May Be Junk,” that is enough off the mark to be annoying. The problem with the article isn’t so much inaccuracy as superficiality.
Norris points out correctly that a lot of buyers are waking up to the unpleasant reality that that triple A paper they thought they had may be worth a heck of a lot less. That’s an appalling outcome because the deal with high rated paper is you accept a relatively skimpy yield so as to have a safe asset. And these securities are coming unglued awfully quickly.
But Norris does his reader a disservice by treating this phenomenon in too broad brush a fashion:
Those securities were nothing like the bonds issued by companies with triple-A or double-A ratings. Such bonds almost never plunge in value because the companies borrowing the money are financially solid.
But the money invested by the hedge funds went to finance mortgage loans to subprime customers, borrowers as close to being a triple-A credit as Moscow is to Maui as a beach resort.
By the magic of securitization, sow’s ears could become silk purses, or at least look like them. Most subprime mortgages would never default, went the theory, and rising home prices would minimize losses when there were defaults. So if a security was protected from the first 10 or 20 percent of losses in a mortgage portfolio, then it was as safe as a loan to General Electric. Such securities got AAA ratings.
Why do I take issue with Norris’ reasoning? The argument is that because the underlying assets were junk, there was no way you could construct a triple A security. The reality is more complicated.
The problem with subprimes has more to do with lack of adequate and relevant data than the quality of the assets per se. You can’t model something unless you have comparable data over long enough time periods (to see how it performs in stress periods) in sufficient quantity to make reasonable analyses. You didn’t have that with subprimes. They have been around since the mid 1990s, but they had very small market share then relative to FHA loans. They grew in the later 1990s, fell in the tech bust period, and picked up again in 2002. But this is not a long history, it certainly isn’t a continuous history, and the 2000-2001 period was not terribly unkind to real estate (it came to be viewed as a safe alternative to stocks).
Now readers will note that even in this discussion I am making the very charitable assumption that it’s even meaningful to apply the same term “subprime” to the paper that was created before, say, mid 2005 and that which followed. Industry participants have noted that there was an explosion of activity in the latter part of 2005 and all of 2006 in which whatever standards there had been went completely out the window. By all accounts, the proportion of really terrible paper, no-docs, high LTV, dubious-to-fraudulent appraisals, increased dramatically. In other words, everyone talking about this situation has a nomenclature issue: what do you mean when you say subprime? Do you mean the market in general, or the really terrible recent vintages? Now Norris meant the awful recent paper talking about, I could agree with him, but he seems to be talking about the general premise of taking a pool of lower quality assets and tranching it so you get some triple A out of it.
So in my view, even if the recent subprimes were of comparable quality to the older ones, I still question whether there was sufficient data to model them (there wasn’t enough of the right sort of history to assess how they’d behave under stress). And the newer subprimes were simply not comparable. There’s a reason the ABX indicies have been plummeting. No one has any basis for judging how the underlying assets will fare.
Norris missed an opportunity in this piece. It’s evident that this securitization (or in the case of CDOs, resecuritization) technology has been used in ways that destroyed rather than created value. How much of it was due, per the discussion above, to analytical failings that should have been caught but weren’t, due to overspecilzation of roles and/or greed?
For example, the paper “Where Did the Risk Go? How Misapplied Bond Ratings Cause Mortgage Backed Securities and Collateralized Debt Obligation Market Disruptions,” by Joshua Rosner and Joseph Mason argues that applying a rating scheme developed for corporations and applying it to static pools of assets with the performance characteristics of mortgages (note we are talking about all mortgages, not subprimes in particular) inherently leads to more frequent and bigger downgrades (they use a lot of convincing math to support this claim). The implication is that the rating agencies should have used a different system.
Now the Rosner/Mason argument might seem to support Norris’ argument, but it really doesn’t. Their beef is about mortgage paper and CDOs generally, not subprimes in particular. Remember, prime mortgages went into CDOs as well.
Finally, there is a place for someone to do a study on “under what circumstances is the type of structuring done in asset backed securities successful?” The point that I think Norris may have been trying to get at, but didn’t articulate, is that the process of taking low credit quality assets and getting some triple A paper does seem like alchemy and is therefore suspect. There are clearly market conditions under which it is more of a stretch than others (the cost of the credit enhancement has to vary over a market cycle, the relationship between the credit spreads for the various ratings makes a huge difference, etc.). How much more often do ABS deals done in frothy times unravel? Any why do they unravel? Is Norris right, that the technology is too often applied to assets that are just plain too lousy? Or is it that the deals instead were structured too aggressively (meaning too much high grade paper sold relative to the support, or to look at it another way, the credit enhancement was inadequate)? And did the overly aggressive structuring wind up tainting all deals to some degree? The latter seems to be the reason for widespread uneasiness in the credit markets.
Regardless, any analysis is likely to conclude that more deals were done than makes sense in the cold light of day. But understanding in a more granular way why this process went afoul, and whether the problems were specific to subprimes or also affected other types of ABS (hopefully to a lesser degree) is crucial, not just for the sake of the investors but the dealers as well.