Who really defines what constitutes a “Qualified Mortgage”? The answer, though codified in federal regulation, is not as simple as it may appear.
Updates to Regulation Z in response to the Dodd-Frank Act seem straightforward enough. Under the regulation, “Qualified Mortgage” status is presupposed so long as: 1) the loan lacks certain characteristics, including negative amortization, interest-only periods, balloon payments, terms exceeding 30 years, and excessive fees and points, and 2) the borrower is demonstrably able to repay it
The regulation also sets forth how lenders must go about assessing ability to repay. Interestingly, for all the complex factors that contribute to credit and pricing decisions made by lenders, a borrower’s actual “ability to repay” is really captured in just one—the debt-to-income (DTI) ratio.
Additional considerations are important when predicting probability of default and the likelihood of loss given default. Credit scores matter, of course, but this reflects potential borrowers’ propensity
to repay more than their ability to. Loan-to-value (LTV) ratios are important, too. But LTV merely reflects borrowers’ incentive
to repay (and, perhaps more critically, the lender’s potential exposure in the event borrowers fail to repay).
But the actual ability
to repay ultimately comes down to the DTI ratio. This reality is borne out in 12 CFR 1026.43(c), which lays out the basis for determining a borrower’s ability to repay. Of the eight
components of this basis, the first seven
relate to the DTI ratio:
‘s current or reasonably expected income or assets, other than the value of the dwelling
, including any real property attached to the dwelling
, that secures the loan;
If the creditor relies on income from the consumer
‘s employment in determining repayment ability
, the consumer
‘s current employment status;
‘s monthly payment on the covered transaction;
‘s monthly payment on any simultaneous loan that the creditor knows or has reason to know will be made;
‘s monthly payment for mortgage-related obligations;
‘s current debt
obligations, alimony, and child support;
‘s monthly debt
-to-income ratio or residual income; and
‘s credit history.
The eighth consideration (credit history) has more to do with the borrower’s propensity to repay than ability to repay, as noted above. After all, a borrower with an 800 FICO score who takes on a $4,000 monthly obligation against $5,000 in gross monthly income is highly unlikely to pay it back. But the first seven all have to do with the DTI ratio, either directly (item vii) or with the numerator (items i and ii) or denominator (items iii, iv, v, and vi). Note that the LTV ratio (appropriately) does not factor into the ability-to-repay calculation at all.
The regulation establishes the DTI standard for Qualified Mortgages at 43 percent, but that’s not the end of it. Under “special rules” codified in 1026.43(e)(4)(ii), any loan
can be a Qualified Mortgage so long as it is eligible for purchase by Fannie Mae or Freddie Mac
(the GSEs). In other words, an acceptable debt-to-income ratio (and, by extension, ability to repay in particular and Qualified Mortgages in general) are whatever the GSEs say they are.
Last year, Fannie Mae announced that it was increasing its DTI threshold to 50 percent in an effort to bring homeownership within reach of more people. Our analysis of GSE originations suggests that this change has had its intended effect. The following chart illustrates that the share of Fannie Mae mortgages with DTIs over the 43 percent regulatory guideline grew substantially, from approximately 18 percent in March of last year to just under 30 percent in April of this year. The share of high-DTI Freddie Mac loans grew from approximately 22 percent to approximately 28 percent over the same period.
Higher DTI ratios are sometimes justified by other compensating risk factors, such as higher credit scores or lower LTVs (even though, as noted above, those factors don’t really contribute to a borrower’s ability
to repay). Our analysis of the data showed however, that high-DTI cohorts exhibited high-risk characteristics in other dimensions as well. The following table illustrates that, generally speaking, loans with DTI ratios of 43 percent and higher were more likely to go to single borrowers (i.e., with no co-borrower) with lower
credit scores and investment properties.
The following two charts illustrate both the more favorable distribution of FICO scores among borrowers with DTIs below 43 percent relative to that of borrowers with higher DTIs and that low-credit-score cohorts account for a higher share of high-DTI borrowers than of lower-DTI borrowers.
In other words, in relaxing the DTI requirement, the GSEs are taking on loans that are riskier not only in terms of their borrowers’ ability to repay but in other dimensions as well. The impact of this “risk layering” is quantified in RiskSpan’s monthly Vintage Quality Index
(VQI), which measures the relative tightness/looseness of underwriting standards and recently crept above 100 for the first time since the financial crisis. (Higher VQI is associated with looser underwriting standards.)
To some extent, this additional risk could be mitigated by lower LTV ratios, but the average LTV ratio has not changed appreciably in the past year, and the difference in average LTV between high-DTI borrowers and other borrowers is not significant, as illustrated below.
What to make of this?
This is not
intended as a “2008–here we go again” alarm post. Underwriting standards naturally ebb and flow in response to risk appetites and policy objectives. In reaching 100, the VQI now stands approximately where it stood in January 2003. It maxed out at 139 on the eve of the financial crisis, and the prevalence of low-doc/no-doc loans leading up to the crisis suggests that the VQI may have been understated then. (After all, who knows what the DTI on all those stated-income loans actually
was?) We are still a long way from 2008.
But it does raise some interesting questions. For example:
- Do 50% DTI ratios represent the ceiling, or are the GSEs willing to go higher? At what point does the policy objective of bringing homeownership into reach of more people become counterproductive?
- The current administration might be charitably described as “unpredictable.” To date, it has not exhibited a great deal of interest in the GSEs, but the president’s signing of the Economic Growth, Regulatory Relief, and Consumer Protection Act suggests that he is a friend to banks. Could this augur future pressure on GSEs to tighten their credit boxes and shrink their footprint?
- What could this mean for the private-label securitization market? Ten years have passed since the crisis, and subprime lending is increasing. What does it mean if Fannie and Freddie are willing to move deeper into this space that was once reserved for banks? What does it mean for non-QM lending as more and more mortgages become QM simply because of relaxed standards?
- What does this mean for mortgage insurers? To what extent is this additional credit risk being passed along to them? What impact might it have on premiums if MI companies seek to price in the additional risk of high-DTI borrowers?
Deferring to the GSEs to determine standards for whether potential borrowers are able to repay (and, by extension, whether a loan should be classified as a QM) makes a certain amount of sense. The GSEs are smart and have been trading in credit risk for decades. Arguably no one has a more robust dataset for determining the extent to which higher DTIs drive default. If Fannie believes that mortgage credit can be responsibly extended to borrowers for whom debt service accounts for half of pre-tax income, then there’s probably something to be said for running that experiment. The ultimate performance of these loans, however, is something that certainly bears monitoring.
VQI is a function the percentage of loans with the following risk characteristics: high DTI ratio (above 45%), low credit score (FICO below 660), high LTV ratio (above 80%), presence of subordinate liens, single-borrower loans, cash-out refinancings, loans secured by multi-unit properties, loans secured by investment properties, adjustable rates, and thin documentation.
Note: The analysis in this blog post was developed using RiskSpan’s Edge Platform. The RiskSpan Edge Platform is a module-based data management, modeling, and predictive analytics software platform for loans and fixed-income securities. Click here to learn more.