Thursday, December 27, 2007

Some cautionary tales (about subprime loss estimates)

Not all 2007 subprime deals are performing poorly (Housing Wire)
It may seem at times as if every residential mortgage-backed security — particularly in subprime — is in the process of being downgraded (which is, to a large extent, exactly what’s been going on). You may be surprised, however, to learn that not all RMBS deals are collapsing under the weight of a housing market gone south.

Fitch Ratings said Wednesday that it had affirmed $1.01 billion from a single 2007 subprime RMBS deal — J.P. Morgan Acceptance Corp 2007-CH3. True mortgage nerds might be interested in the prospectus for such a trend-breaker.

The affirmation here by Fitch would seem on the surface to fly in the face of two poorly-performing, recently-downgraded WaMu subprime deals from 2007 that have been mentioned on this blog recently (first deal mentioned here; second deal mentioned here).

The loans in this deal are predominantly first-lien 2/28s and 3/27s, and a good chunk were originated via the wholesale channel at Chase Home Finance; most of the loans are in Florida, California, Illinois and New York. I could not find any material differences in disclosed underwriting standards or deal structure, either. In other words, there doesn’t appear to be much different here versus the loans pooled in the two troubled WaMu deals and originated via Long Beach Mortgage.

It’s interesting to ponder — for a minute — why Chase-originated subprime loans are performing (and are expected to perform) infinitely better than subprime loans originated by Long Beach Mortgage.

Are Subprime Losses Being Exaggerated? (Felix Salmon)

John Berry dedicates his Bloomberg column today to debunking exaggerated estimates of the magnitude of the subprime crisis. $300 billion, he asks? $400 billion? Pshaw.

A more realistic amount is probably half or less than those exaggerated projections -- say $150 billion. That's hardly chicken feed, though not nearly enough to sink the U.S. economy.
A loss of $150 billion would be less than 12 percent of the approximately $1.3 trillion in subprime mortgages outstanding.

Now Berry admits that the markets are valuing subprime securities as though total losses will exceed $300 billion, so he's basically saying that he's right and the markets are wrong. I'm generally suspicious when people say that, but Berry's math does make a certain amount of sense:

Most subprime borrowers aren't going to default. Suppose even one in four does and lenders recover somewhat more than half the mortgage amount. A fourth of $1.3 trillion in subprime mortgages is $325 billion, and a 55 percent recovery would mean a loss of about $145 billion.

Once you remember that a good $500 billion of that $1.3 trillion is in fixed-rate subprime mortgages with relatively low default rates, these kind of numbers do seem reasonable.

On the other hand, Berry assiduously avoids trying to tot up total losses from Alt-A and prime mortgages, not to mention the resulting losses in industries ranging from homebuilders to diswasher manufacturers. So while Tyler Cowen points to Berry's column as a reason why he's "not yet convinced by the economic pessimists," I don't find it nearly as compelling. The fact is that a lot of the USA's biggest and most important banks are in serious trouble, and when banks are in trouble, lending and growth invariably suffer.

And at the risk of sounding like a broken record, I'll repeat: no one is going to have a real handle on mortgage losses unless and until someone manages to get a handle on the percentage of mortgage loans which are non-recourse. If your house falls in value and you have a non-recourse mortgage, then it makes perfect economic sense for someone in a negative-equity situation to simply walk away – something known as "jingle mail". But given the amount of refinancing going on during the last few years of the mortgage boom, I suspect that the vast majority of mortgages are not non-recourse. (Refis are never non-recourse.)

If there are lots of middle-class homeowners out there suffering under the burden of enormous non-recourse mortgages which are worth more than their houses, we could easily find ourselves in a situation where total losses moved up into the $400 billion range. But that's a really big if.

The WSJ's cursory story on CDOs (Naked Capitalism)

In less than a week, we have had two lame front page stories at the Wall Street Journal that bear a strong resemblance to each other. I am wondering whether this is a function of holiday short-staffing or a new Murdoch template.

"Cursory story" is far too kind a description for this paradigm. The current offering, "Wall Street Wizardry Amplified Credit Crisis," and December 21's "Fraud Seen as a Driver In Wave of Foreclosures," both try to shed light on problemmatic practices by focusing on a single, extended anecdote with some facts and data to provide a framework.

The New York Times understands how to do this sort of article, and more often than not, they use several long illustrations rather than hang the piece on only one. And generally, those stories are also ones that have real human interest (example from Wednesday's paper: the tale of a woman suffering from Alzheimers used to frame a discussion on advances in testing).

Today's topic, collateralized debt obligations, isn't exactly brimming with gripping personal stories, but the Journal tries nevertheless. It focuses on a single CDO, Norma, and the man who nurtured it, one Corey Ribotsky, who runs a firm called N.I.R. Group LLC in Long Island. The company comes out of the penny stock netherworld and its affiliates are being sued by three companies for share price manipulation. Norma is a mezzanine CDO, one that held instruments with an average rating of BBB-.

Ribotsky came in contact with Kenneth Margolis who later became co-head of Merrill's CDO banking business. Margolis put Ribotsky with two ex-Wachovia bankers to form a CDO management firm.

Aside: there may have been a vastly more important story here, namely, "Merrill and others recruited lowlifes with broker/dealer licenses to act as CDO managers." Oh, but that would have taken more work.

The story is so preoccupied with the recounting of this sorry deal that it obscures the fact that the information it gives about CDOs is often incomplete, or misleading. Some examples:

1. Inadequate definition and market data. The authors Carrick Mollencamp and Serena Ng never define clearly what a CDO is, save via a cute interactive that is many ways is less informative than one at Portfolio. com (that got across the idea of the cash flow waterfall). Nothing on the size of the overall market or how it has grown explosively since 2005. Similarly, the story fails to say how significant these Norma-type CDOs were (proportion of market? Losses sufffered by investors?)

2. Way way too much breathless prose and exaggerated claims:
In its use of newfangled derivatives, Norma contributed to a speculative market that dwarfed the value of the subprime mortgages on which it was based. It was also part of a chain of mortgage-linked investments that took stakes in one another. The practice generated fees for a handful of big banks. But, say critics, it created little value for investors or the broader economy.

The "newfangled derivatives" were credit default swaps, which have been around for a decade and are mainstream. The second sentence is unclear. It implies cross ownership, when all the article discusses is serial securitization (pieces of CDOs going into other CDOs, for instance, the previously discussed "CDO squared" and "CDO cubed").

3. Horror! Derivatives multiply risk! As the authors tell us:
In principle, credit-default swaps help banks and other investors pass along risks they don't want to keep. But in the case of subprime mortgages, the derivatives have magnified the effect of losses, because they allowed bankers to create an unlimited number of CDOs linked to the same mortgage-backed bonds. UBS Investment Research, a unit of Swiss bank UBS AG, estimates that CDOs sold credit protection on around three times the actual face value of triple-B-rated subprime bonds.

If you want to worry about the level of derivatives in relationship to the underlying cash market, three times isn't a high level compared to other markets (and even though CDS are called derivatives, they are actually insurance, but go under the "derivatives" rubric so as not to be subject to insurance regulations). For example, I believe that the volume of credit default swaps on corporate bonds is ten times the underlying credit.

Now having three times as much credit protection written on subprime bonds might indeed be a very bad thing, much worse from an economic standpoint than the level of protection written on corporate credits. But the Journal doesn't make that argument.

4. Presenting old news as news:
The use of derivatives "multiplied the risk," says Greg Medcraft, chairman of the American Securitization Forum, an industry association. "The subprime-mortgage crisis is far greater in terms of potential losses than anyone expected because it's not just physical loans that are defaulting."

Sorry, Greg, but the Financial Times' Gillian Tett wrote about this very topic repeatedly, starting in January, based in part on unsolicited information. People in the market were sufficiently concerned about the explosion of CDOs, the teeny amount of underlying equity, and the fact that they were often sold to hapless investors that they contacted a journalist. But the American Securitization Forum is a lobbying group, so it isn't surprising that they claim this problem was a surprise.
There are lots of other quibbles, such as imprecise use of terminology. But the big flaw is the whole premise of the story, that a single deal will tell you everything you need to know about CDOs. Were it was so easy.

No comments: