Posts Tagged ‘stress tests’

What to stress?

The coincidental timing of the release of stress test results and Fannie Mae’s 1st quarter statement makes a tempting target for comparison — look at the stress test assumptions, compare them with Fannie Mae’s actual results, and show how accurate the Fed’s “most adverse” scenario for credit defaults actually is and how much scarier things actually are in the real world.

The New York Times’ Gretchen Morgenson gives the calculus a go in her column today, and botches it.

Losses recently seen in Fannie Mae’s portfolio support this view. In the first quarter, its subprime loans had average losses of around 68 percent; the Fed expects two-year losses in subprime to be, at worst, 28 percent.

Alarming, right? But Morgenson appears to be comparing two totally different numbers. The stress tests projected loss *rates* — that is, what proportion of a portfolio would end up in default.* For subprimes, it anticipated loss rates on subprime betweem 21 and 28 percent. The Fannie numbers Morgenson cites show loss *severity* — that is, how much Fannie will lose on those loans that do go into default, not its losses on the entire portfolio.

The Fannie statement does include some ominous indicators — of Fannie’s small subprime portfolio, between 22 and 47 percent of mortgages are more than 60 days late on payment — but the “serious delinquency” rate right now for Fannie subprime, indicating the loans likeliest to go into foreclosure, currently stands at 17.95%, well within the stress tests’ projections.

* Clarification: The stress tests projected the “loss rate” — that is, the projected total proportion of the holdings’ value that will be lost. The Fannie loss severity numbers don’t show that.

Sunday, May 10th, 2009 | Tags: , ,
Posted in A Lot More | No Comments »