by Tanta on 8/13/2008 07:52:00 AM

My post yesterday featuring some rate reset charts from Clayton prompted a good deal of concern in the comments regarding the issue of Option ARMs and the differences between the Clayton chart and some others that have been published lately. Reader Greg kindly emailed me copies of two charts on Option ARMs that have been published in the WSJ and Business Week recently, with a request that I comment on the apparent differences between and among these charts in terms of the timing of "reset" problems.

As far as I'm concerned, a large part of the confusion here is that our friends in the media are not very careful about using the terms "reset" and "recast" consistently, like us UberNerds do. Take this chart from Business Week:



The chart title says "Reset Schedule," but the legends make it clear that what you have here is actually a "Recast Schedule." No wonder people are encouraged to use these terms interchangeably.

This chart from the Wall Street Journal doesn't use the term "reset" at all, which is good since it clearly explains that it is talking about "recast":



Do note, though, that the WSJ chart uses only "scheduled recast dates." The Business Week chart above contrasts "scheduled" recast with projected actual recast based on the rate of growth in actual negative amortization balances as of the chart date.

And, finally, we have our Clayton chart I posted yesterday that avoids the whole lingo problem by opting for the title "Loans With Rate Changes." Maybe the Clayton analysts got tired of the "reset vs. recast" confusion and just decided to go long-form. In any case, the Clayton chart, unlike the two above, includes but is not limited to Option ARMs; it is looking at the whole "Alt-A" pile which includes amortizing hybrid ARMs and lots of interest-only ARMs as well as OAs.

There obviously isn't perfect consensus here on terminology. All I can really do is make clear how I am using these two terms. I think my usage conforms to the way industry wonks talk, but I can't promise you that anyone outside the wonkosphere will be as careful with these distinctions. Caveat lector.

"Reset" refers to a rate change. "Recast" refers to a payment change.

On a normal fully-amortizing ARM, the interest rate resets on what is called the "Change Date" (five years out for a 5/1 ARM, three years out for a 3/27 ARM, each year thereafter for the 5/1 and every six months thereafter for the 3/27, etc.). The payment recasts exactly one month after the rate resets. Mortgage interest is paid in arrears, so first you reset the rate, then the following month you recast the payment. "Recast" is really just a shorter word for "reamortize": you take the new interest rate, the current balance, and the remaining term of the loan, and recalculate a new payment that will fully amortize the loan over the remaining term.

On an interest-only ARM with a rate change that happens during the interest-only period, the rate resets on the Change Date and then the interest payment is recalculated on the next payment date. I wouldn't tend to use the term "recast" here since with an IO, you aren't actually amortizing or "casting" a new payment, just adjusting the interest due given current balance and new rate. The big issue with IOs is the end of the IO period, when the payment has to be amortized over the remaining term. This date is what I would call the "recast" date of an IO. It may or may not coincide with the first interest rate reset date. Some 5/1 IOs, for example, reset and recast both at the end of five years. Some have a 10-year IO period, meaning they reset annually between years 5-10 but do not recast until year 10. If the rate resets to a higher rate in that period, the required IO payment increases, but not as much as it will when the recast hits and principal must also be repaid on a 20-year schedule.

On a typical Option ARM, the rate resets monthly beginning as early as the first month of the loan. The payment is adjusted, but not recast, annually; usually the payment increases by no more than 7.5% each year. It is that mismatch between rate reset and payment change that actually creates the potential for negative amortization; the "minimum payment" gets outstripped by the actual interest due because it increases much more slowly than the rate does.

Option ARMs do not "recast" until the sooner of 1) the loan reaching its balance cap or 2) the first "scheduled" recast date, which is usually 60 months from origination. What you see in the Business Week chart is the difference between the two: the recast projections come a lot earlier if you look at how close loans are actually getting to their balance caps, rather than just assuming they'll all recast on their five-year anniversary.

By and large, the biggest danger for Option ARMs and IO ARMs is the recast date, not the first or subsequent rate reset dates. However, for any ARM borrower who qualified at the highest possible debt-to-income ratio they could manage, any payment change, even one not quite as shocking as the recast on an OA or an IO, can tip the balance. As we are talking in this specific context about Alt-A, I for one believe that most of these loans did stretch too far in the beginning, and so even first rate resets on IOs or fully-amortizing ARMs will cause a marked increase in delinquencies in the absence of the borrower's ability to refinance at reset into a new discounted ARM, which will be the case for some time.

I hope that clears it up a bit, at least for the next week or two.