Saturday, February 16, 2008

LFKAJ* Update

by Tanta on 2/16/2008 12:28:00 PM

As you all know, the stimulus bill just passed raises the statutory conforming loan limit from its current $417,000 in all states (except Alaska and Hawaii) to a new variable limit, equal to the greater of the current limit or 125% of the area median home price, with a maximum of 175% of the current limit ($729,750). The new limit expires on 12/31/08, and is retroactive to July 1, 2007 (meaning that some jumbo loans closed by lenders from July 1 to passage of the new limit would be eligible for sale to the GSEs). This was supposed to kick-start the jumbo mortgage market and bring the spread between jumbo and conforming rates down.

So what's happenin'?

1. We do not yet know exactly what the new limits will be for affected areas, or even exactly how many areas are affected. The bill gives HUD the responsibility for defining and publishing the median home prices for purposes of establishing the new conforming limits, but many people do not realize that HUD has never published its own home price indices before. Will it use OFHEO data? NAR? Federal Housing Finance Board? Some combination thereof? Is an "area" an MSA or a county? We don't know; that's undoubtedly why HUD was given 30 days from enactment of the law to come up with the relevant numbers. My guess is they'll be using every one of those 30 days to work this out.

2. The bill does not require the GSEs to purchase the new higher-balance loans at the same terms currently in place for $417,000 loans. If they did offer the same loan parameters, that would mean--in theory at least--that the new LFKAJ could be made as high as 95% LTV/CLTV. However, a 95% LTV loan would require mortgage insurance. As a general rule, the MIs mostly limit purchase loans over $650,000** to 90% LTV. So unless the MIs relax their rules--which would surprise me--there wouldn't be a lot of point in the GSEs setting a maximum LTV that cannot qualify for the required MI. Realistically, the GSEs will have a lot of work to do hashing out what the allowable guidelines are for these loans, which will involve a lot of discussions with the MIs. As OFHEO Director Lockhart has been signalling nothing but disapproval of the whole thing since the idea was first floated, it is surely certain that any proposal the GSEs come up with will involve review and possibly negotiation with OFHEO. Plus, the GSEs have to update their AUS (automated underwriting), their pricing and delivery systems, and their selling and servicing guides to accommodate the new rules. You can confidently expect this to take a while.

3. Lenders will have their own systems and policy and pricing work to do once they know what the GSEs will accept. Doug Duncan of the MBA estimates this will take three to six months. There's a lot involved there, particularly in the pricing engines, but to give you a simple example: loan processing systems currently are programmed to validate the maximum loan amount on a conventional conforming loan simply by looking at the property state (AK/HI or not) and number of units (the limits are higher for 2-4 unit properties). Somebody will have to change that to MSA or county code, and build the new tables for the new dollar amounts. If I know my industry, this part of the process alone will require ten Gantt Charts, two task forces, thirteen meetings, seventeen dry-erase markers, four new white boards, and eleventy-jillion budget dollars allocated to IT consultants per lender. On the conservative side. The "stimulus" to IT investment will outweigh the mortgage dollars put into circulation by a factor of probably five, but hey! Stimulus is stimulus. If it turns out that the layoffs in the back room we've been hearing about for the last year included too many of the people who understand how the systems interface, multiply by five. (Do I sound jaded about how long it takes a huge corporation to turn on a dime once it has overloaded itself with multiply-interfacing automated systems and laid off too many cubicle dwellers? So I sound jaded. So does the MBA with this three to six month estimate.)

4. SIFMA has already announced that these LFKAJ won't be allowed in TBA pools. The short version of that: this pipeline will have to be segregated from the CCNJC*** and priced as "specified pools." Housing Wire found a market participant who speaks the language: “Jumbo borrowers [will] only get the benefit of guarantee in market, while the prepayment hickey and higher GSE guarantee fees are tacked onto their rate.” I assume this source is a Wall Streeter; on Main Street mortgage desks I think it's called a "prepayment ding." (Brokers, who are complete amateurs about pricing, will uniformly call it a "bump.") Bottom line: it's going to take a while for everybody involved to put a price on this stuff, and some folks may not be very happy when they find out what that price is.

5. Somebody has to name these things responsibly. I can't keep typing LFKAJ. I warn everyone involved, however, that if you decide on "super duper conforming" I will ridicule you to my dying day.

6. By the time all these ducks get into a row, it will be mid-year at best and a fair amount of money will have been spent invested. It will be considered a shame to have done all of that to originate LFKAJ for maybe two quarters. The suggestion that we just go ahead and make this permanent will be made approximately ten minutes after the first trade settles (at the outside).

7. Making it permanent will involve the same problem that the now old-fashioned conforming limit calculation had: what do you do if the average home priced used in the calculation drops? The $417,000 limit is already a touch overdue for a reduction, and heretofore everyone's expectation was that it would be lowered for 2009. (That issue will arise when the third-quarter national home price data is available in November.) If HUD doesn't choose a home price data source for the new "temporary" limits that is the same as the data source used for the "permanent" limits (which is the FHFB MIRS October price), we could be looking forward to some real fun with duelling spreadsheets later this year.

8. In five or six years, if CR and I haven't yet retired on the fabulous earnings we get from blogging, we will be fielding questions from frustrated readers that go like this: "Why were these bizarro variable area-level conforming limits designed like this? Why are these formulas so complicated?" Fortunately, we'll just be able to repost oldies out of the archives to explain that whatever conforming regime we're living under in the future wasn't, actually, "designed." It just "happened" because that was how the programmers solved the system validation constraints in mid-2008 in the middle of a credit crisis. We will be called "cynical" by people who refuse to believe that things actually happen that way, and will be sent links to academic papers from whip-smart young economic theorists who will demonstrate how the conforming loan limits constitute optimal market efficiency.


*Loans Formerly Known as Jumbo

**Those quaint old mortgage insurers still use the term "super jumbo" to refer to loans over $650,000. Now that there are super jumbos that just became "conforming" overnight, many elves will have to be busy revising many documents.

***Conforming Conforming, Not Jumbo Conforming