Atrios is concerned that there will be a second wave of foreclosures when the next wave of option ARMs reset:
Option ARM rates are going to be recasting soon and in increasing numbers. That’s the magic moment when people can no longer make minimum payments, when they can longer make interest-only or neg-amortization payments.
Ryan Avent points out that most of the defaults associated with resetting mortgages happen before the reset occurs:
He notes that if you chart 2/28 and 3/27 mortgage defaults over time, you don’t see the kink at the 24 or 36 month mark you’d expect if the reset was doing most of the work in generating foreclosures. What they found was that defaults were happening surprisingly quickly — basically, underwriting standards were so terrible that borrowers couldn’t even afford the initial payment.
As the bust has proceeded, by contrast, defaults have mainly resulted from the combination of negative equity and some kind of income shock — job loss, death or illness, divorce, and similar.
And Megan Mcardle concurs and offers follow-up:
The problem is not principally people who can’t pay their mortgages because their interest rates have reset–people will cut back on a lot of other things to keep their house, and if you can’t afford a 1% rate increase even with drastic lifestyle cuts, you probably have too much house. Rather, the main problem is people who have an income shock.
The initial point about resets is absolutely correct. The people I’ve talked to at the Boston Fed and elsewhere who have dealt with this data have all been completely surprised by how much the resets don’t matter. But I want to expand on it, because the point is slightly different – these loans were designed never to be paid off and to be not active at the time of the reset.
For subprime mortgages, Megan is overstating that the rate reset were minimal. Check out these rates:
(For this entry, I am using data and graphs from the excellent – Subprime Outcomes: Risky Mortgages, Homeownership Experiences, and Foreclosures, a Fed working paper. full paper here, pdf slides here. Last version is May 2008, so it is looking at the height of the subprime market, 2004-2007.)
Rates go from about 8% to above 11% after the two year mark. In general, the numbers I’ve heard say that mortgage payments increased about a third after a reset. This is fine if the first, cheaper, payments is much less than what you can afford to pay per month. However most buyers could barely afford the 8% rate (which was high compared to prime mortgages). So the model was to refinance. All the time. Every two years, refinance. Get close to the jump, maybe a payment or two into it, and then refinance. The banks knew this, so their strategy was to get their cut from the prepayment penalties they could legally charge (I need to go back and actually write out that model for banning prepayments), while always making refinancing available until the credit crunch came.
Instead of being a business based around collecting interest off loans, it became a business based off collecting high fees off loans. It became, though the Boston Fed is loathe to use this analogy, a credit card company business model.
Let’s look at some empirical evidence:
This chart is hard to read, but take a minute to grasp it. This is from December 2007. The lines under Still Active reflect how many of the subprime loans are no longer active (in 12/07) percent wise, by time from origination. So for subprime loans originated in 2004, 81.6% were no longer in existence by month 30. They either defaulted or refinanced. Roughly ~18% of these mortgages defaulted by month 24 across all years. The rest were refinanced. If their FICO got high enough during that time period, they refinanced into a prime loan. For most, they refinanced into another subprime loan, another spin around the wheel. The evidence leads us to believe that if they couldn’t refinance, they would have been in a very risky situation.
We ended up in a situation where a product, a mortgage, that should be designed to survive a 360 month time horizon, were gone 80% of the time within 30 months. 10% of the mortgages don’t even make it 6 months! Given that it takes a few months to end the contract, that’s one out of ten not being able to make more than two payments. I can’t get a $60/month cable package if the cable company doesn’t think I own $120. Where were the underwriting standards!?!?
(If you can read heartbreak in math, look at the additional jump in months 25-30, where it goes from ~67% to ~81%. That’s some poor struggling household opening their mortgage bill and seeing their amount due jump up a third. They forgot, and their friendly neighborhood shadow bank sure as hell wasn’t going to remind them. The bank then makes sure they can squeeze some juice out of the jumped mortgage rate before sending them back to month 1 on the chart with a refi. Also note the last line in 2005 and 2006 – note how they drop off. That reflects, in December 2007 when this chart was compiled, people calling in for a refi and the bank responding “Sorry, there’s no more money.”)
So when Ryan points out that there isn’t much to worry about since most of it is has been decided by the date of the reset, that’s true, but that’s true conditional on functioning mortgage markets. It’s especially true conditional on people saying “we can refinance that crappy mortgage that is falling apart into a brand new crappy mortgage that resets the clock.”
As for the loans Atrios is alluding to, I am not in the doomsday camp on them, but I think there is real worry that they are going to put the resets into play in a way we haven’t see before in the data. I don’t know the specifics on where the markets stand, but if there isn’t a wave of mortgage liquidity that can handle the jumps in refinances in months 19-24 and 25-30 you see above, we are in uncharted territory in crap mortgage land. Hold onto your hats.