An article by John Berry at Bloomberg, “Fed Actions Defuse Subprime ARM Rate Reset Bomb,” is extraordinarily misleading, claiming that a Fed paper based on a single pool of MBS issued by New Century in 2006 shows that subprimes will work out much better than conventional wisdom says.
Let’ s start with Berry:
Many analysts and public officials have said that foreclosures of subprime adjustable-rate mortgages would soar this year as owners’ monthly payments jumped when interest rates reset to a higher level.
Not only is that unlikely to happen, this year’s resets of earlier vintages of subprime mortgages may even reduce some payments that increased in 2007.
The reason? The index to which many ARMs are tied is the six-month London inter-bank offered rate, or Libor, and that rate has fallen from more than 5.3 percent last fall to about half that level….
Unfortunately, most of the defaults and foreclosures that have wreaked havoc in financial markets haven’t been due to resets so far. Many borrowers simply bought a house or condo they couldn’t afford unless bailed out by rising prices, and lower rates alone won’t help them much.
Still, the big drop in Libor means there likely will be many fewer foreclosures than there would have been….
A new report, “Understanding the Securitization of Subprime Mortgage Credit,” by economists Adam B. Ashcraft and Til Schuermann of the New York Federal Reserve Bank, published this month, provides a wealth of detail about subprime mortgages. Much of its information is based on a pool of such mortgage-backed securities issued by New Century Financial in June 2006.
All but 12 percent of the loans in the pool were ARMs, either the so-called 2/28 or 3/27 variety. That is, they carried a fixed-initial rate for two or three years, respectively, so the former will first reset in June.
The average initial rate for the loans was 8.64 percent, set when the six-month Libor was 5.31 percent, according to the report. It was a teaser rate in the sense that once resets began, the interest rate would be based on Libor plus a spread of 6.22 percentage points.
Now the interesting thing is that the authors of the paper stress that they investigated only one pool used to illustrate how subprimes work and to try to understand how the product turned out to work so badly. The abstract and executive summary make no reference to the economics of subprimes. The abstract:
In this paper, we provide an overview of the subprime mortgage securitization process and the seven key informational frictions that arise. We discuss the ways that market participants work to minimize these frictions and speculate on how this process broke down. We continue with a complete picture of the subprime borrower and the subprime loan, discussing both predatory borrowing and predatory lending. We present the key structural features of a typical subprime securitization, document how rating agencies assign credit ratings to mortgage-backed securities, and outline how these agencies monitor the performance of mortgage pools over time. Throughout the paper, we draw upon the example of a mortgage pool securitized by New Century Financial during 2006
Fed economists no doubt would know better than to use on one pool of MBS issued by one issuer in one month for an economic when there are vastly more comprehensive data sources that are designed for precisely that sort of analysis. And a quick look at one suggests that Berry’s conclusions are quite a stretch.
The American CoreLogic databases as of March 2007 contained 38 million mortgages. Their extraordinarily detailed analysis of 8.4 million ARMS originated between 2004 and 2006 showed only 9.1 % with initial interest rates of 8.5% or higher (note that the paper claims an average of 8.64%)
There were more mortgages ate 2% and below (1,1 million) than above 8.5% (770 thousand). Without throwing in the intermediate levels, it’s obvious that the weighted average is well below 8.64% (the level in the New Century pool, which gave Berry the notion that there wouldn’t be much reset shock). Similarly, a March 2007 (admittedly now dated) paper by Chris Cagan deemed ARMs with initial rates of 6.5% or higher as not-very-vulnerable to reset shock.
ARMs with low introductory rates were never intended to reset; the assumption was that the would refinance. And recent pools are running at unheard-of rates before reset, with monthly default rates of 3.5%, which equates to a 34.8% cumulative default rate over three years. Thus the performance of later subprimes is horrendous independent of the issue of resets.
Finally, while Libor was a popular index for setting the reset rate, it’s far from the only benchmark. Others include the 11th District Cost of Funds rate, the Prime rate, the Monthly Treasury Average rate, the Constant Maturity Treasury rate. And some of these have not been affected by the Fed’s cuts:
Note that while prime has fallen, its level is not much below what it was in 2005 and 2006, which were the heaviest years for origination of dubious subprimes (while the 2007 vintage is worse in terms of quality, the volume issues was lower than in the two preceding years):