RealtyTrac announced today that foreclosures are up markedly. While the number of foreclosure filings, which includes default notices, auction sales notices, and bank repossessions, rose 8% in January compared to the previous month, the new figures represent a 57% increase compared to a year ago.
While the number of subprime mortgages, especially those that were written in 2006 when rational lending guidelines took a hiatus, is a major factor contributing to this increase, another trend that’s emerging is painting a disturbing picture.
A few days ago, Global Economic Analysis (GEA) posted a screen shot from a particular Washington Mutual Alt-A mortgage pool known as WMALT 2007-0C1. The screen breaks down the pool of mortgages into the typical categories, including delinquencies. Here are some of the highlights from the pool:
Weighted Average LTV = 78%
Fico Score = 705
Full Doc Loans = 11%
Geography = 48% California, 15% Florida
The chart breaks down performance by month, starting with July 2007. By most standards, 705 is a respectable credit score, which makes the delinquency numbers all the more surprising. In a period of 7 months, this pool is showing a massive foreclosure rate of 13.17%. Add REOs into the mix and the figure goes to 15%. Even the vintage 2006 subprime pools didn’t default at such a rapid rate.
GEA poses an interesting question as to whether the FICO system has lost its mind or if maybe there’s a larger issue at work. Although it’s hard to imagine borrowers with a 20%+ equity stake (albeit phantom like) and strong credit scores defaulting at a rate that would lead any servicing portfolio manager to jump out of the nearest window, the numbers seem to indicate that borrowers may be walking away when they are 30 or 60 days delinquent, not even waiting for foreclosure. In December 2007, the 90 days delinquent category stood at 3.79%. Even if every one of these delinquencies became a foreclosure, the figure should only double to 7.58% in January. Instead, the foreclosure figure is 13.17%.
A look at the details shows that nearly 93% of the pool was rated AAA yet almost 15% of the entire pool is in foreclosure or REO after 8 months.
What does it all mean? Until recently, I may have been one of the last holdouts on the FICO bandwagon. I’ve seen enough delinquency reports to make me believe in the ability of FICO to accurately predict performance. But something is terribly wrong with this picture. Credit scores north of 700 have not, in my experience, shown such poor performance levels so quickly. While it’s possible that a deterioration in the underwriting guidelines (e.g. reverses after closing) that we saw on the subprime side became part of the fabric in Alt-A lending, it doesn’t explain these numbers, even if most of them were stated income loans. Unless of course, these were mostly No Doc loans, meaning that most of the borrowers didn’t have jobs. It’s hard to imagine just what was going on in the underwriting department.
If there was ever a doubt that the phenomenon recently dubbed as “jingle mail” actually exists, wonder no more. It’s alive and well. Hopefully, it won’t be still be around around come Christmas time. But given the recent trends, that may be wishful thinking.
Richard Bitner
3 responses so far ↓
1 Rob Dawg // Feb 26, 2008 at 1:58 pm
LTV is not CLTV nor is it adjusted for new sales prices.
FICO 705 is not a respectable score.
WaMu was not lending uniformly across geographic regions. 48% in California is not surprising nor informative.
Part of the accumulating REO inventory is banks ramping up too slowly to respond to changing market conditions.
In summary, things are really really bad but no so bad that careful analysis cannot model them. The gnashing of teeth you hear is those whose models have failed not the failure of models.
2 Richard Bitner // Feb 26, 2008 at 8:10 pm
Rob,
Good catch on the LTV. I clearly overlooked that part when writing this. However, I disagree with part of your assessment. North of 700 is respectable. Not stellar, but certainly a strong enough score that for most lenders and most programs, it’s a fico band that doesn’t see a pricing hit. That’s what I use to judge whether or not a score is respectable.
While models have failed, I doubt there is a historical reference in which a pool of 700+ credit score loans would have an aggregate foreclosure rate of 15% in less than 9 months.
3 grover13 // Feb 28, 2008 at 10:39 am
RobDawg beat me to the punch- look at CLTV, not LTV. In fact….look at layered risk in general. The problem that the industry ran into with Alt-A mortages is it relied fully on the FICO to assess risk, and ignored layered risk. Surely you must understand layered risk from your days at Kellner.
Other important omissions that could help clear up your confusion on the performance:
- Further Doc Type breakdown- you alluded to this, but didn’t dig deep enough. What was the breakdown between SVA, SISA, No Ratio, NINA, No Doc? This tells a story.
- Given that 2/3 of the volume was in CA & FL- how many properties were Non-Owner Occupied?
- How many borrowers were first time homebuyers? Or the opposite- how many had multiple loans in the same pool?
- What is the average level of reserves the borrowers had?
- What was the mortgage product these were tied to? ARMs (or worse- Pay Option ARMs) are indicative of borrowers with a short term focus, and therefore added performance risk.
As a whole, FICO still rank orders risk reasonably well. But that doesn’t mean you can ignore all the other risk factors. I’d have thought you would have learned that by now.
Leave a Comment