How Risky Is It to Make a Non-QM Mortgage? And Is QM Going to Hold ...

        One of the huge questions hanging over the mortgage market today is what will happen to access to credit for credit impaired or non-traditional borrowers.  There is a real concern that the Dodd-Frank Act’s mortgage reforms will reduce the availability of mortgage credit because lenders’ fear liability for making mortgage loans that fail to qualify as “Qualified Mortgages” (QM) and are thus potentially subject to an Ability-to-Repay (ATR) defense.  I've blogged on aspect of QM before (here, here, here, here, here, here, here, and here). Based on a preliminary analysis, I think this concern is overblown, and in this very long post I attempt to work through the potential liability for lenders that make non-Qualified Mortgages.  (I note that all of this is my tentative readings of the statute; we really don’t know how courts will interpret it, and others may see better readings than I do now.) 

        Still, my back-of-the-envelope calculation suggests that it is quite low in terms of loss given default and could probably be priced in at around 10 basis points in additional cost for a portfolio with weighted average maturities (actual) of five years.  I invite those who would calculate this differently to weigh in in the comments—it’s quite possible that there are factors I have overlooked here, as this is a really preliminary analysis. 

        Because this post is REALLY long, here’s where it goes (yes, I feel like I'm doing one of those unwieldy 100+ page UFTA decisions, so I'm going to have a table of contents!):

First, it gives the legal background on the Dodd-Frank Act ATR requirement;

Second, it covers the Qualified Mortgage safe harbor to the ATR requirement;

Third, it covers the civil liability to the CFPB for failing to verify ATR;

Fourth, it covers private civil liability for ATR violations;

Fifth, it covers the private civil damages for ATR violations;

Sixth, it addresses the likelihood of liability of ATR liability;

Seventh, it attempts to calculate a reasonable loss given default rate for ATR and the affect on mortgage pricing (please note that I am not attempting to work in compounding anywhere here; this is a ballpark analysis, not a precise figure);

Eighth, it considers whether the real issue is ATR so much as the costs imposed by contested foreclosures (which ATR might make more common); and

Ninth, it suggests that ATR may not really matter in terms of access to credit, as the real limitation will be the lack of liquidity for loans that do not meet the GSEs’ underwriting guidelines independent of QM. 

Phew.  

1.  The Ability-to-Repay Requirement

        The single-most substantive regulation for consumer banking included in the Dodd-Frank Act is the Ability to Repay (ATR) requirement for mortgages (sorry Durbin Amendment, I love you, but mortgages are just bigger than debit).  Title XIV of the Dodd-Frank Act amends the Truth in Lending Act (TILA) to provide that “no creditor may make a residential mortgage loan unless the creditor makes a reasonable and good faith determination based on verified and documented information that, at the time the loan is consummated, the consumer has a reasonable ability to repay the loan, according to its terms, and all applicable taxes, insurance (including mortgage guarantee insurance), and assessments.”  

2.  The Qualified Mortgage Safe Harbor

        The Dodd-Frank Act directs the CFPB to create a safe harbor for “Qualified Mortgages” or QMs.  The CFPB has defined a QM (with some minor exceptions) as a mortgage meeting six criteria

1. regular payments that are substantially equal (ARMs and step-rate mortgages excepted) and always positively amortizing 

2. term of no more than 30 years
3. limited fees/points (caps vary with mortgage size)
4. underwritten using the maximum interest rate in the first five years to ensure repayment
5. income verified
6. backend debt-to-income ration of no more than 43% (including simultaneous loans)

        The last three requirements are satisfied prior if the loan is eligible for GSE purchase or guarantee.  If the loan is a QM it qualifies for a safe harbor from the ATR requirement.  If the loan is a regular QM, this is an absolute, irrebuttable safe harbor, while if the loan is a high-cost QM (defined as pricing at 150 basis points over prime for first liens, and 350 basis points over prime for junior liens), then the safeharbor is a rebuttable one. 

        To understand the importance of the safe harbors, it is necessary to understand what happens if a lender doesn’t comply with this ability to repay (ATR) requirement.  This has been a source of a lot of confusion.  There is definitely some risk for lenders, but I think it is much less than generally perceived by the lending industry.  What is critical to understand is that a loan can be non-QM, but still satisfy the statutory ATR requirement.  The statute only prohibits loans that are made without verified ATR. 

3.  Civil Liability to CFPB for ATR Violations

        If a loan is made without verified ATR, there is the possibility of a CFPB enforcement action.  TILA is a “Federal consumer financial law,” by virtue of being an “ennumerated consumer law” and the Dodd-Frank Act provides that “It shall be unlawful for…any covered person…to offer or provide to a consumer any financial product or service not in conformity with Federal consumer financial law, or otherwise commit any act or omission in violation of a Federal consumer financial law”.  The CFPB can seek a civil monetary penalty and/or an injunction for violations of Federal consumer financial laws.  The CFPB, of course, has limited enforcement resources; as long as a lender is making a good faith effort to verify ATR, CFPB enforcement seems unlikely.  The mere fact that loans have high default rates should not by itself be a problem for ATR liability, although it does suggest a possible problem with the lender’s ATR analysis and might raise some eyebrows. 

4.  Private Civil Liability for ATR Violations

        The real risk, it would seem, for failing to verify ATR is that that it creates a defense to foreclosure for the consumer.  A violation of the ATR requirement is a TILA violation.  TILA specifies that damages for such a violation is a “defense by recoupment or set off”, and that the “amount of recoupment or set-off…shall equal the amount to which the consumer would be entitled [to receive in] damages for a valid claim brought in an original action against the creditor, plus the costs to the consumer of the action, including a reasonable attorney’s fee.” 

        This means is that the consumer does not have an affirmative cause of action under TILA.  Instead, TILA creates a counterclaim or defense.  This is significant because it means that the TILA remedy is only triggered if and when the creditor tries to collect on the loan and only reduces the collection on the loan.  It does not prevent the foreclosure (unless, presumably, the TILA setoff is greater than the unpaid balance on the loan). 

        How this applies in a nonjudicial foreclosure is not clear.  A defense or counterclaim simply has no application to a nonjudicial foreclosure.  Perhaps this means that any borrower that can raise an ATR claim can convert a nonjudicial foreclosure into a judicial one, but there’s nothing making clear how this works. 

5.  Calculation of ATR Private Civil Liability

        Irrespective of how the TILA defense works procedurally, it does not prevent a foreclosure.  Instead, if successful, it simply reduces the foreclosing lender’s claim on the foreclosure sale proceeds.  The reduction is by the sum of:

1. actual damages resulting from the failure to verify ATR.
2. statutory damages of between $400 and $4000. 
3. the consumers’ costs of the litigation plus “reasonable” attorneys’ fees
4. the sum of all finance charges and fees paid by the consumer on the loan, unless the failure to verify ATR “is not material.”

It’s worth unpacking all of these.  What are “actual damages” from failure to verify ATR?  From other similar “actual damages” provisions, such as those in the FDCPA, we know that “actual damages” can include emotional harm to the borrower.  It might also include consequential damages, such as harm to the consumer’s credit score, which can have effects on employment, insurance, and other loan pricing.  It might also include the relocation costs from a foreclosure (although generally those costs would not have been incurred at the time the defense is raised).  But what about the loan itself?  Is that actual damages?  Actual damages requires a loss causation requirement, meaning that it must be shown that but for the failure to verify ATR, the borrower would not have incurred the damages.  But I could see a reading in which there is an assumption of loss causation, in that the loan was made only because the lender failed to verify ATR.  If so, then actual damages would be all amounts paid on the loan plus the outstanding balance (principal, interest, fees, etc.).  That said, I don’t think the loan itself would likely be included in actual damages as it is too speculative to say that the borrower wouldn’t have gotten a loan at all based on after-the-fact underwriting.  Perhaps the borrower could have gotten a loan, just at a higher rate.  Still, there is some question as to what, if anything, “actual damages” are. 

        The statutory damages are the easy part and are capped at $4,000, with judicial discretion within the allowed range. Costs and attorneys’ fees can be quite expensive.  If a case gets litigated all the way, it could easily be $50,000 (remember, consumer attorneys are generally not billing $1,000/hr) and possibly much more, especially if the consumer retains expert witnesses.    

        The finance charges and fees could add up to quite a lot, if the default is late in the life of the loan.  Figure that on a $200,000 loan at 6%, this is $12,000 per year (I’m doing simple annual interest).   If it’s a $1.2 million loan, that’s $72,000/year.  That turns into real money fast.  Of course, all of this is subject to the TILA materiality requirement.  It’s not quite clear what the materiality requirement means in this context—the TILA provision regarding materiality was drafted for other types of TILA violations, and the set-off defense is written to piggyback on the pre-existing TILA provision.  The materiality requirement means that the failure to verify ATR was material…but to what?  Here are three possible readings:

1. Material to the harm to the borrower? 
2. Material to the default on the loan? 
3. Material to the foreclosure? 

I think it should be #2 the default on the loan, but the statute isn’t explicit about this, and it strikes me as a potential problem for a borrower trying to raise an ATR defense.  Let’s assume that reading #2 is correct.  I think the burden of proof on materiality is on the defendant creditor, but this strikes me as an increasingly low burden with the age of the loan.  It’s hard to believe that failure to verify ATR could be material to a default in most cases after the first couple of years of the loan.  Indeed, the QM safeharbor only requires underwriting out to five years.  If a borrower defaults in year 5, the fact that the loan wasn’t really underwritten in year 0 doesn’t seem particularly relevant.  Underwriting is, at best, a prediction of future risks.  But it is really hard for even the best underwriting to predict a borrower’s financial condition out beyond the immediate future.  Measures like credit scores and income and DTI are inherently retrospective.  They cannot capture the prospective changes in employment, family status, health, or property values that drive many defaults.   Thus, I would think that the risk of losing finance charges is really only an issue for a couple of years post-origination.  Again, however, the burden of proof here is on the creditor, which means that the creditor will have to litigate this point, which might cost more than the liability avoided. 

        So what does all this liability mean in the worst case scenario?  Let’s assume a $200,000 loan at 6% annual interest, which defaults at the end of year 2.  Let’s assume that the LTV on the house is 95%, so the house is worth $210,500.  Let’s also assume $50,000 in costs and attorneys’ fees.  If there are no actual damages, then the lender is looking at $78,000 in set-off liability:  $4,000 in statutory damages, plus $50,000 in costs and attorneys’ fees, plus $24,000 in finance charges and fees paid.  In such a case, the lender will recover $132,500.  That’s a 33% (0.33) loss severity on the loan, which is quite serious. And if the LTV is higher, the loss severity will be worse (although it will not be paid in real dollars, as discussed below).  Moreover, ATR liability does not stop with TILA itself, however.  A TILA violation could be a predicate for a state law UDAP claim (or potentially also an FDCPA violation).  That might itself carry further statutory damages, although it is unlikely in most cases to increase the lender’s total liability significantly.  But this adds some uncertainty to the question of loss severities.  In any case, we need to apply the loss severity to the likelihood of loss. 

6.  Likelihood of Liability for ATR Violation

        In order for there to be ATR liability, several things need to happen.  First, there has to be a default on the loan.  Second, that default needs to result in a foreclosure.  Third, that foreclosure has to be contested. Fourth, the borrower needs to show that the lender did not consider his/her ability to repay when making the loan.   That’s a lot of different gates to run through. 

        If a loan is in fact well-underwritten, but simply doesn’t qualify for QM, the default rate should be fairly low.  Let’s assume 10%, which I think is really quite high.  Of course, if the loan is well-underwritten, then an ATR defense should not be successful at all.  If a lender is simply making loans based on collateral values, there is an ATR problem, but even then, let’s say 25% of loans default. 

        Even then, not all defaults result in foreclosures.  A lot depends on servicing.   Let’s assume that the foreclosure rate on defaulted loans is 90%.  Most foreclosures are not contested.  I don’t know any hard statistics on this, but I’d guess than no more than 5% are contested.  Let’s double that to 10% for our purposes here.  This means that we’re talking about .25 * .9 * .1 =2.25% (0.0225) of all non-QM loans resulting in contested foreclosures. 

7.  Loss Given Default for ATR Violations

        If we apply this 2.25% loss probability figure to the 33% loss severity estimate, we’re talking about 74.25 basis point increase (0.007425) in loss given default.  So on a portfolio with an average original loan balance of $200,000, that translates to an average loss of $1,485 per loan.  This risk is covered by increasing the interest rate by 10 basis points for five years.  That’s not a huge affect from QM itself. 

        Moreover, to the extent that a property is underwater, the setoff isn’t being paid in real dollars, but either in uncollectible deficiencies, or in deficiency judgment dollars.  If the loan is non-recourse, the lender hasn’t lost anything to the extent that the setoff is applied to the underwater part of the balance.  (The setoff reduces the amount owed and hence the amount underwater.)  If the loan is recourse, the lender’s loss is of the deficiency judgment, but given that these tend to sell for 3-4 pennies on the dollar, we’re talking about a loss of 1 basis point.  If the property is not underwater, the lender is paying in real dollars, but this means that ATR is really a concern for lenders only when there is a collection possible. 

        On top of this, I think I’m likely overestimating both probability of loss and loss severities due to the ATR defense.  Still, I recognize that there’s some uncertainty here, which makes pricing more difficult. 

8.  The Cost of Delay:  Direct ATR Liability May Not Be the Real Issue

        Perhaps the biggest risk for lenders created by ATR is the issue of delay.  If a borrower raises an ATR defense against a non-QM loan, the borrower should be able to get past a motion to dismiss, as there is a question of fact outstanding.  It’s not clear to me that a lender, much less a lender’s assignee, can prove ATR verification based simply on affidavits and pleadings.  (How to prove ATR easily seems to me to be a major issue lenders’ counsel should think about.)  It strikes me as requiring a more detailed investigation that might preclude an early summary judgment motion.  This delay gives the consumer significant leverage for extracting a settlement, even if ATR was in fact verified. 

        Indeed, ATR gives a borrower one more ground on which to contest a foreclosure.  The mere fact that a foreclosure is contested will impose costs on the lender.  A contested foreclosure will be a slower foreclosure, which means lost time value.  The ATR defense also likely means that any non-judicial foreclosure will effectively become a judicial foreclosure.  Beyond this, litigating a foreclosure is expensive, even if successful.  The lender may be able to recover its costs from the foreclosure sale proceeds, if there is sufficient equity in the property, or from a deficiency judgment, possibly with interest, but those are delayed recoveries that impose an up-front liquidity burden on lenders who have to fund the foreclosure litigation.  And if the loan is accruing interest at a below-market rate, delay is also imposing an opportunity cost. To be sure, none of these costs are specific to ATR.  They are generic to contested foreclosures, but the possibility of an ATR defense (and particularly the recovery of attorneys’ fees, which enables the possible funding of foreclosure litigation) might increase the frequency of contested foreclosures. 

9.  Does ATR Matter for Non-QM Loans?  Or Is the Real Problem the Lack of a Secondary Market?

        Even putting aside the TILA defense risk on non-QM mortgages, there is still the chicken-and-egg problem of liquidity in non-QM mortgages, and I think that's where the real risk of constriction of access to credit lies.  Depositories aren’t going to want to make non-QM mortgages in any volume if there isn’t a secondary market for the loans that provides them with potential liquidity.  A secondary market in non-QM loans is going to depend on the GSE credit box and/or the revival of the private label securitization market.  There's little reason to be optimistic about either.

        Last May the FHFA directed the GSEs to purchase only QM loans or loans exempt from the ATR requirement.  This means that after Jan. 10, 2014, the GSEs, will only purchase loans subject to the ATR requirement if they are fully amortizing, with terms no longer than 30 years, and points and fees of no more than 3% of the total loan amount (or other applicable limits).

        These aren't huge limitations on the mortgage market.  Eliminating IOs and Payment Option loans, 30/40s and plain 40s, and high point/fee mortgages isn't where the real credit constriction will come.  There weren't a lot of these loans happening post-bubble in the first place.  Instead, the real issue is DTI requirements.  But the GSEs are exempt from the QM requirements of a maximum DTI, income verification, and underwriting to the maximum interest rate in the first five years. Instead, their own underwriting standards (e.g., Desktop Underwriter) suffice.  That's where the real action is.  The contraction of credit is really going to be a function of how tight the GSE credit box is, not QM.  It's possible that FHFA will allow the GSEs to buy non-QM loans, but I don't see that happening any time soon, and even if it does, again the real issue is the GSE credit box, not the CFPB's definition of QM.  

        It's possible for there to a private-label secondary market in non-QM loans, but for the foreseeable future, the GSEs are the only secondary market to speak of. The Redwood and other private-label deals have financed some 16,000 loans since 2008--that's not a meaningful secondary market.  I don't see that changing in any meaningful way until (1) the various servicing and trustee problems (including reps and warrants) are solved, and (2) there’s greater certainty about where housing financing reform is going.  No one wants to invest a lot in creating a private-label securitization platform now that might not work after potential reforms.

        All of this is to say, it might not matter how much liability exists for an ATR violation.  The real issue in terms of availability of credit is going to be the depth of the secondary market, and these days that is a function of what the GSEs will buy.  That's not a legal issue, but a question of their risk appetite (and consider how a monopolist will accept lower volumes in exchange for higher margins) and public relations, in which any change to the credit box is perilous.