Our take on the arc of eventual subprime mortgage losses is simple: most estimates, particularly of losses on loans originated in 2006 and 2007, are significantly too high . The reason why they’re too high is simple, too. They assume that last year’s credit performance will persist far into the future. Only it won’t.

Three events in particular sharply skewed subprime credit performance last year. First, the decline in home prices, particularly in the hottest markets, flushed out both the overt and covert speculators. One servicer reports almost 50% of its foreclosures last year were investor-owned properties, though only 16% were listed as investor-owned on the original mortgage application. Once leveraged investors are underwater on their properties, they’re very likely to default quickly. It’s uneconomic to do anything else.

Second, the supply of mortgage credit dramatically tightened. That reduced the refinancing options for borrowers facing sharply higher monthly payments once their teasers reset, and reduced the demand for homes generally.

Finally, the borrowers in 2006 and 2007 who never should have gotten a mortgage of the sizes they did, but could anyway because of lax underwriting, couldn’t refinance their loans, and defaulted.

As a result of these factors, subprime credit quality got bad in a hurry last year, as the rate at which current loans became delinquent ballooned, and the rate at which delinquent loans moved from early-delinquency buckets moved into later-delinquency ones rose sharply.

One-two punch

This one-two punch of a higher inflows and deteriorating roll rates led industry observers, most notably the rating agencies, to dramatically increase their estimates for cumulative losses for loans originated in 2006 and 2007.  I think the agencies significantly overreacted to what happened last year. The monthly reports filed by the companies that service the loans provide support for my view.

Here’s a quick (and pretty intuitive) summary of how delinquency rates affect eventual losses: when new-delinquency inflows and delinquency roll rates rise, estimates of cumulative losses, particularly for recently originated loans, must rise, as well. They have too. More iffy loans at the front of the pipeline means higher chargeoffs several months down the road.  Similarly, if new delinquency inflows and roll rates are stable, so should estimates of cumulative losses.  But here’s the important part. If new delinquencies and roll rates decline, then, ceterus paribus, estimates of cumulative losses should decline, as well.

And so far in 2008, the inflow and roll-rate numbers have been improving. Monthly servicer reports have consistently shown a decline in the inflow of new problem loans and an improvement in delinquency roll rates. 

Just seasonal?

Skeptics argue that the improvement simply reflects seasonal factors. (Consumers receive their tax refunds during the early part of the year.)  Last year, they point out, servicer reports also showed that new problem loans fell in February, March, and April. But once tax season ended on April 15, new problem loans resumed rising in the May reports, and kept on going up for the rest of the year.

While clearly seasonality has helped, the most recent data we see show that the improvement represents a real improvement in subprime credit quality.

This year, as in 2007, the inflow of new problem loans fell in the February, March, and April reports. But then in the May report--the month in 2007 things started to deteriorate again--inflows of new problem loans kept on falling in three of the four ABX indices that represent originations in 2006 and 2007. That’s not seasonal!   Take a look at the comparison.

 
Don’t get me wrong, the inflow of new problem subprime mortgage loans, even at the May level, is still quite elevated. But the level would need to be even higher, based on my calculations, to justify the cumulative loss forecasts currently being thrown around by the rating agencies.

But that’s not all the encouraging news.  The roll rates of loans moving from early-stage delinquency buckets to later stage buckets, are improving, as is the roll rate at which loans over 90 days delinquent are “curing.”

Similarly, cure rates on foreclosed loans have begun to rise as well:

So delinquency inflow rates are improving, not deteriorating. If these trends continue for a few more months, and if the rating agencies are objective (ha!), then their estimates of cumulative losses must eventually come down, as well. This has important implications for the holders of RMBS, CDOs, and the financial guarantors.

What do you think? Let me know!