by Tanta on 7/24/2008 09:37:00 AM

Well, yesterday was quite the odd day. I was having a late afternoon nap when I suddenly awoke, heart pounding and skin crawling, with that horrible spooky sense of being watched. I decided it was a bad dream, made a cup of tea, and wandered over to the computer, only to discover that my co-blogger had just a few minutes earlier put up a post letting us all know that the FDIC is going to be keeping its eyes on bloggers.

There's only one thing for it, then. If the FDIC is going to be worrying about the bloggers, the bloggers are going to have to be worrying about the insured depositories. I don't know that that's an ideal setup, exactly, but someone has to worry about the banks and thrifts, not just about bad PR for the FDIC, and if Sheila Bair is going to ruin my naps I'm going to have time on my hands.

Which brings us to Downey Financial, who visited the confessional this morning. It was pretty ugly. What we got, for the first time as far as I can remember (I nap a lot these days, you know, or at least I used to), is some post-modification performance information on the infamous "retention modifications."

If you remember, Downey got everybody a little fired up back in January when it announced that its auditor was making it restate its Non-Performing Asset (NPA) numbers for the second half of 2007. Downey had put in place a program to offer "market rate" modifications to performing borrowers in its loan portfolio. These were, apparently, mostly Option ARM borrowers whose rates had adjusted to pretty high levels. Downey modified them into amortizing ARM loans at the same interest rate that a new ARM borrower would have gotten. Because there was not a below-market rate given to these borrowers, and because they were current at the time of modification, Downey decided it did not need to count these loans as "troubled debt restructurings," which would mean including them in the NPA category. However, KPMG told Downey that the loans did indeed need to be considered "troubled debt restructurings," for a very specific reason: Downey did not re-underwrite these loans at the time of modification to verify and document that the "market rate" given to the borrowers was truly the rate that a new borrower of the same credit quality would have gotten. Downey agreed to count all of these "retention mods" as NPA until each borrower had made six consecutive payments under the new loan terms. Ever since then, Downey has been reporting separate numbers for total NPA including the retention mods, plus NPA without the retention mods. They clearly believe that having to include the retention mods in NPAs makes their NPA number look worse than it "really" is. Whatever that means.

In today's press release, we got some additional information about the performance of these loans:

To the extent borrowers whose loans were modified pursuant to the borrower retention program are current with their loan payments and included in non- performing assets, it is relevant to distinguish those from total non- performing assets because, unlike other loans classified as non-performing assets, these loans are paying interest at interest rates no less than those afforded new borrowers. At June 30, 2008, $548 million or 82% of such borrowers had made all loan payments due. Accordingly, the 15.50% ratio of non-performing assets to total assets includes 4.34% related to performing troubled debt restructurings, resulting in an adjusted ratio of 11.16%.

Through June 30, 2008, $347 million of loans modified pursuant to our borrower retention program have been removed from non-performing status because they met the six-month payment performance threshold. Of all loans modified pursuant to the borrower retention program, including both those classified as non-performing as well as those removed from non-performing status, 87% have made all payments due.

So. The "retention program" has been in place for one year now. If I am reading this correctly, a total of $1.015 billion in loans have been modified under this program. $347 million have made at least six consecutive on-time payments and are no longer included in NPA. Of the $668 million still in NPA, $548 million have made all payments due so far (that might be less than six, since some of these mods will be less than six months old).

This means that of a group of modified loans that are no more than one year old, 87% are performing. If you remove the oldest performing loans from that group--the ones that have had six payments actually due and have made those payments--you get 82% performing loans. A delinquency rate of 13-18% in the first year would probably be something to be proud of if these were "classic" troubled debt restructurings--namely, workouts of delinquent loans that required below-market interest rates to result in payments the borrower could afford.

But Downey tells us these "retention" mods were done for borrowers who were current on their loans at the time of modification, and who were given rates no better or worse than market. The very idea of a "retention" program, of course, is that (in theory at least) these are borrowers whom you don't want to refinance away from you with another lender, because they're good borrowers. You "retain" them by offering them a less expensive alternative to refinance, namely a modification that keeps the loan with the same lender and servicer.

And 13-18% of these "keepers" went bad within a year or less of a modification that supposedly improved their risk profile? Perhaps people are getting used to seeing such high delinquency rates on subprime and "worked out" loans that figures like this seem normal. But these are supposed to be prime loans, and that level of early delinquency or default in a book of "retention" mods is just awful. Downey says these borrowers got the same modified rate and terms that a new refinance borrower would have gotten. Were they expecting that 13-18% of their newly-originated refinance transactions would be delinquent in the first year?

If nothing else, I'd say this demonstrates a good call by KPMG. Downey called these "retention" mods--implying that they were the kind of high-quality borrowers you don't want to lose to the competition--and failed to re-underwrite the loans to verify that claim. The results of this program suggest to me that it was less a classic "retention" program than a pre-emptive strike: Downey made a big effort to modify as many of its then-current Option ARMs as it could into better loan terms, not because they were the kind of borrowers you necessarily want to "retain" but because they were probably not going to get a decent refi anywhere else and if they'd stayed in the Option ARM program the delinquency rate would likely have been worse than 13-18% a year later.

I don't actually think that a "proactive" modification program is necessarily a bad thing. But I think calling it a "retention" program is disingenuous at best, and I think that Downey's experience is proving the point that it really does matter whether you re-underwrite those loans before you modify them. But then, forcing outfits like Downey to call these programs "pre-delinquency workouts because letting them ride is too dangerous" programs rather than "retention" programs would probably spook people, and we know the FDIC doesn't want that.