Thursday, December 21, 2006

Tanta: On Hybrids, Teasers, and Other Mortgage Guidance Problems

by Bill McBride on 12/21/2006 11:12:00 AM

According to the LA Times, our friends in the mortgage business have once again been asking the regulators to let this cup pass:

A coalition of nine banking and housing groups, including the American Bankers Assn., asked six federal and state bank regulators to abandon the idea of adding so-called hybrid adjustable-rate mortgages, or hybrid ARMs, to guidelines they put forward in September. . . .

Lawmakers and consumer groups have called on regulators to incorporate the mortgages in the guidelines, which banks use to set underwriting and disclosure standards. The guidelines currently apply to interest-only and payment-option adjustable-rate loans. . . .

In a Dec. 7 letter to regulators, six members of the Senate Banking Committee, including incoming Chairman Christopher J. Dodd (D-Conn.), asked that the hybrid mortgages be included because they "have a number of the same risky attributes" as the mortgages already covered by the guidelines.
I haven’t gotten my hands on a copy of either of these letters yet, so I really have no way of knowing for sure if they’re as stupid as the Times reporter makes them sound. But since the question of the “2/28s” keeps coming up in the comments, I thought I’d at least define a few terms as I understand them, in the na├»ve hope that this might help us move the ball forward about an inch.

First of all, a “hybrid ARM” is called a “hybrid” because it is, basically, a cross between a fixed rate and adjustable rate mortgage. Before the early 90s, an “ARM” basically meant a one-year ARM. The initial interest rate was set for one year, and the rate adjusted every year. The only real variations on this theme involved shortening the adjustment frequency: you could get an ARM that adjusted every six months instead of one year.

Around the early 90s, the “hybrid ARM” was introduced. It had an initial period in which the rate was “fixed” that didn’t match the subsequent adjustment frequency: this is the classic 3/1, 5/1, 7/1, and even 10/1 ARM. The whole idea of the hybrid ARM was to provide a kind of medium-range risk/reward tradeoff for borrowers and lenders. With a fixed rate loan, the borrower gets the security of knowing the payment will never rise, with the potential risk that market rates will go down and that secure fixed payment will be expensive relative to market rates. With an old-fashioned one-year ARM, the borrower gets to take advantage of an initial interest rate that is lower than the fixed rate, with the potential for a falling interest rate in the future, at the risk that rates will rise rather than fall. From a credit underwriting perspective, FRMs have always outperformed one-year ARMs by a wide margin, precisely because payment stability always reduces risk. The trouble for a lot of borrowers was that increasing numbers of them didn’t feel the need for 30 years’ worth of interest rate protection, as the average tenure in a home became around 7 years, while occasionally shocking spikes in short-term rates meant that a one-year ARM was too often not enough rate protection. So the hybrid offered the middle way: enough rate protection to more or less match expected loan life, a start rate that was discounted less than a one-year ARM but was still cheaper than a fixed rate, plus the security of not being forced to refinance at the end of the fixed period if things didn’t go as planned and you were still in the home you bought—the other then-popular alternative to the ARM was the fixed-rate balloon, which had a term of five or seven years and then required a lump-sum payment of the remaining balance. With a balloon, you have to either refinance or sell at maturity, and take the refinance costs and potentially much higher market interest rate. A hybrid ARM, on the other hand, just turns into a one-year ARM after the first adjustment, which may not be what you wanted, but it doesn’t force you to do something by its very structure. Whether you are forced by affordability issues is, of course, another matter.

The immediate question these products raised for lenders was how to underwrite them: are they ARMs or FRMs for purposes of determining credit risk? The position taken by the GSEs from the beginning—and also adopted by the mortgage insurers—was that you could treat a hybrid as a FRM for underwriting purposes if its initial fixed period was longer than three years. If it was three years or less, the loan had to be treated as you would a one-year ARM. And that, by the way, meant qualifying the borrower at (usually) the maximum first adjustment rate: if your start rate was 5.00% and the first adjustment cap was 2.00%, then 7.00% was the maximum rate you would face at your first adjustment and thus the rate used to qualify you. The mortgage insurers used the more expensive ARM premiums for these loans, and allowed those with 5-year or greater fixed periods to use FRM premiums. All of this made perfect sense to Tanta at the time, by the way, and still does. All other credit practices being equal, true hybrids (fixed period of 5 years or more) very slightly underperform FRMs, but only by a hair. The 3/1 has always performed only very slightly better than the one-year ARM. Historically speaking, it’s not a dumb place to make the cut.

Then, of course, we had to get creative, adding interest-only periods to the initial fixed rate period, which creates enormous payment shock when and if the loan does reset; moving from a more-or-less standard 5/1 ARM with a 2.00% first adjustment cap to a more-or-less standard 5/1 ARM with a 5.00% first adjustment cap (so that the initial rate could be more reasonably discounted from the fixed, creating the classic “teaser” problem), changing the adjustments from one-year indexed to the treasury to 6-month indexed to the LIBOR, and, worst of all, deciding that the “hybrid” was the perfect product for subprime borrowers. When hybrids hit the subprime market, they not surprisingly mutated a bit, the initial fixed period on the majority of loans shrinking to 2 years with a 6-month reset, often with interest only for two years and a deeply-discounted start rate. This made the “subprime hybrid” an even worse performer than the old one-year “true ARM,” but certain folks continued to delude themselves (and their investors) that the “hybridization” of these products actually improved the risk characteristics, never mind that the loan was less a cross between a FRM and an ARM than it was a high-risk ARM with a single drop of FRM in it two generations ago.

Then there is the “teaser rate” problem. There is no exact or absolute definition of a “teaser” rate. The concept was traditionally applied to a start rate that was so deeply discounted relative to the fully-indexed rate (index plus margin) that its clear purpose was merely to rope in the borrower, and which had a pretty fair chance of creating severe payment shock down the road. The problem is, of course, that almost all ARMs—hybrid or not—are “discounted” relative to something (either the fully-indexed rate or the prevailing 30-year fixed rate). And there’s absolutely nothing wrong with that; in fact, if you the borrower are willing to take the future risks involved in an ARM, you’re getting screwed if your lender doesn’t give up a little yield in the here and now in exchange. The whole reason that the 7/1 and 10/1 ARMs go in and out of fashion isn’t that people wouldn’t like to have them sometimes, it’s that in certain rate environments it’s hard to price them with any real discount to the 30-year fixed, and why take one if you don’t get a rate discount? The point here is that conflating the terms “teaser” and “discounted start rate” is a big problem if you’re talking about regulations designed to combat the use of teaser rates, since if you outlaw any discount of the start rate you just basically increased the credit costs to prime borrowers for whom a 5/1 ARM with reasonable caps is often a great deal.

Of course, you can try to come up with some formula to define the difference between “teaser” and “discounted,” but you’re likely to fall into the silly pit that a lot of regulations have ended up falling into when they do things like this. The better approach—and for once, some regulators seem to have tried it—is not to focus on what makes a rate a “teaser,” but to focus on what makes a teaser a problem: how to calculate the payment used to qualify the borrower for the loan. The idea, in short, is that you can give the borrower any old rate discount you want, as long as you ensure that the borrower can afford the inevitable payment increase when the discounted rate goes away. Here, however, we run into the problem of deciding what is a reasonable time period over which to force the borrower to qualify at his or her current income, and that gets us back to the distinction between the “true” hybrids and the ARMs masquerading as hybrids. It is not obvious to me, at least, that prudence always and everywhere requires a mortgage borrower to qualify for the worst possible mortgage payment she might face in five years while assuming that her income will never increase. It does, however, strike me as prudent that a mortgage borrower should be able to carry the worst possible payment she might face in three years at her current income. Arbitrary? Sure, but you really do need to make a cut somewhere, unless you want to de facto eliminate the true hybrid ARM offering for prime (or even near-prime) borrowers, and why would you want to do that?

So this is why I’m so frustrated about all this noise regarding whether the 2/28 should “count” in the Nontraditional Mortgage Guidance or not. To give the lender lobby its due credit for a minute, they aren’t being as hysterical as they might sound by objecting to any regulation that codifies any and all “hybrid ARMs” as a problem product by definition. In my view, what the Nontraditional Guidance needs to do is just declare a set of rules for when a hybrid is an ARM-hybrid and when it’s a FRM-hybrid, require all ARM-hybrids to be qualified at the maximum interest rate possible at the start of the fourth year of the loan, and require lenders who offer interest-only ARMs to either offer the IO period for at least three years longer than the initial fixed period or qualify the borrower at the amortizing payment. (Going out three years past the adjustment means that you smooth out the payment increases, letting the rate rise over time to fully-indexed before you add in the recast to amortizing payments. The recast can still hurt, but by now you’re six to ten years into the loan, and chances are still reasonable that appreciation and income growth will bail you out if a refi won’t.) Beyond that, simply requiring lenders to chose either a prepayment penalty or a subprime credit score, but not both, would solve most of the 2/28 problem. (Why? Because you can’t afford to give that “teaser” rate to the borrower without throwing in a prepayment penalty to keep the loan from paying off before it adjusts, thereby increasing your expected yield over the life of the loan. If the investor didn’t get the prepayment penalty, the investor would require a more reasonable yield in the first two years.)

The rest of the problem can be solved most practically, in my mind, by mandating risk “classifications” for things like interest only, negative amortization, CLTVs over 90%, 5% first adjustment caps for the 5/1 ARM, stated income, etc. “Classifications” mean that a loan portfolio has to set aside higher loss reserves for loans in certain classes than for others; this costs the portfolio some yield, effectively putting a disincentive to risky lending in front of an institution that can’t afford decent reserves, while letting an institution that is strong enough offer a reasonable number of higher-risk loans. After all, it isn’t just borrowers whose income, stated or verified, can be a bit iffy. I’m all in favor of making the bank verify rather than state its income, and qualify for its loan portfolio based on its actual equity position, collateral quality, and generally responsible debt management behavior. But I’m old-fashioned enough to think that corporations shouldn’t necessarily get a better deal than consumers get.