Sunday, July 15, 2007

Conforming Loan Limits: The Subprime Excuse

by Tanta on 7/15/2007 12:37:00 PM

As some of you may remember, there was a fair amount of uproar late last year when the GSEs announced that the conforming loan limit--the maximum loan amount for mortgages purchased by the GSEs--would remain at $417,000. The limit remained unchanged even though the national average price calculated by the Federal Housing Finance Board, the number the GSEs are required by law to use, decreased by 0.16% (which would, applied directly, have resulted in a new conforming loan limit of $416,300).

The legal and regulatory issues surrounding the calculation of the conforming loan limit are so involved and confusing that even the more heroic UberNerds tend to give up in frustration. What frequently gets lost in arguments over interpretations of 12 U.S.C. 1717(b)(2) and methodological changes in FHFB's MIRS (go here for OFHEO's recap, if you dare) is the whole point of a conforming limit. Of course it's a safety and soundess concern, but it's also a question of mission or mandate for Fannie, Freddie, and FHA: these government-sponsored enterprises and agencies have always been mandated to provide liquidity to the low-to-moderate (moderate meaning "average") housing market, not its high end.

In any case, it has long been a source of amusement to industry-watchers to hear the endless whining of the mortgage lobby over the question of increasing conforming loan limits (which by statute increase the FHA limit accordingly). The same market participants who regularly have a cow over Fannie and Freddie's retained portfolios and capital adequacy do have a tendency to throw that "safety and soundness" religion out the window when it comes to the loan limits.

To whit, here's the joint response of MBA, NAHB, and NAR to OFHEO's (modest) proposal to adjust the regulatory guidance for conforming loan limits in such a way that would allow for potential decreases to the limit after two years of declining home prices:

Our three organizations represent major components of the housing and mortgage markets – builders, realtors and lenders. It is our concerted view that the Proposed Guidance would be detrimental to the national economy, home buyers, current home owners, the industries that serve homeownership, and to the success of the housing missions of the Enterprises, the Federal Housing Administration (FHA), and the Veterans Administration (VA). Given the importance of the CLL to the housing industry, changes such as those OFHEO proposes should be widely circulated for public comment through the Federal Register. Not only is the proposal bad public policy, it does not appear to be authorized under current law, which only permits increases in the loan limit.

The current statutes state that the conforming loan limit is adjusted annually by “adding to each such [previous] amount . . . a percentage thereof equal to the percentage increase during the twelve-month period ending with the previous October in the national average one-family house price in the monthly survey of all major lenders conducted by the Federal Housing Finance Board” (emphasis added).1 The Enterprises may not purchase loans above the conforming loan limit. The conforming loan limit also impacts limits for FHA and VA loans, as FHA and VA loan limits are tied to the CLL. In fact, a decrease in the CLL could have an adverse budget impact if the result is that borrower access to the FHA and VA loan programs is limited and the two agencies produce lower guarantee and insurance fee income.

As you are aware, the housing sector is currently undergoing a correction, and there is concern about the availability of funds for the refinancing of loans and for new loans. Reductions in the conforming loan limit could impair the ability of some borrowers to refinance out of subprime mortgages, which is of particular concern for families with problematic mortgages, as well as prevent some first-time home buyers from obtaining lower cost financing on conforming, FHA or VA loans.

It's hard to match the hilarity of the straight-faced claim that a 1.00% decrease in the FHA loan limit would have an "adverse budget impact" that we should worry about. (It is a fact that the FHA's mortgage insurance fund is currently a "negative subsidy," meaning that the excess of premiums over losses does show as a gain to the federal balance sheet. That doesn't exactly make it a source of revenue.)

But I'm more interested in how a decrease in the conforming limit "could impair the ability of some borrowers to refinance out of subprime mortgages." How many, do you suppose, is "some"? Strangely enough, our friends at MBA, NAHB and NAR don't quantify that.

Courtesy of S&P:

According to S&P, the average subprime loan balance in Q107 was $190,832. These balance figures come from rated RMBS subprime securitizations, which by S&P's estimate include about 10% second liens (most second liens are securitized in the ABS category, not the RMBS category). So that average will be lower than the actual average indebtedness of subprime borrowers in recent vintages. Credit Suisse estimates the average loan size for subprime purchase-money mortgages at $181,300 in 2004, $197,900 in 2005, and $199,800 in 2006.

You can do a lot of upward adjustments to those numbers to account for subordinate financing and still wonder just how many subprime borrowers' refinance problems have anything to do with loan amount (rather than LTV and CLTV, for instance). Of course the whole exercise is a bit pointless if you simply stop to ask why conforming loan limit calculations should have anything to do with bailing out near-jumbo subprime loans in the first place. Nonetheless, I'd like to hear the MBA cough up some data to back up their claim that some non-trivial number of subprime borrowers could lose a refinance opportunity in the absence of an increase to the conforming limits. If you're going to ask the GSEs and FHA to take on the increased marginal risk of higher-balance loans of worse quality, you might want to actually quantify the risk of their failure to do so.