CRO Outlook

Mortgage Risk: Why We Should Eliminate the Debt-to-Income Ratio Limit

The current DTI rule for qualified mortgages is too restrictive and should be abandoned in favor of more reasonable underwriting standards.

Thursday, April 18, 2019

By Clifford Rossi


The Qualified Mortgage (QM) rule is a good example of well‐intended but poorly designed policy. QM came about as part of the Dodd‐Frank Act to address serious issues during the mortgage boom regarding borrowers' ability‐to‐repay their mortgages.

One of the better‐known examples of this practice was qualifying borrowers at “teaser” rates on option adjustable‐rate mortgages (ARMs) well below the actual note rate. When combined with poor income documentation, such underwriting practices significantly elevated credit risk. As egregious as such practices were at the time, the QM rule was not the answer to this problem.

Clifford Rossi Headshot
Clifford Rossi

QM provides lenders with a “safe harbor” in part from future litigation that could be brought by a borrower due to shoddy underwriting practices. Beyond imposing limits on loan attributes, fees and points charged to borrowers, it also imposed a bright line maximum of 43% debt‐to‐income (DTI) ratio on qualified mortgages.

As with any major policy, debate continues over the effect of QM on the mortgage industry and borrowers. Some argue that, since it went into effect in 2014, QM has restrained mortgage lending activity - while others maintain that temporary exemptions afforded Fannie Mae and Freddie Mac artificially boosted home prices. At the center of this debate is the so‐called “QM patch” that allowed loans that met Fannie Mae and Freddie Mac underwriting standards to be QM‐eligible even if their DTIs exceeded 43%.

In concept, the idea of ensuring lenders incorporate robust ability‐to‐repay practices in their underwriting process is sensible, but the 43% DTI restriction is unsound policy.

QM's 43% Limit

Indeed, a key problem with QM is the 43% DTI limit. This requirement for QM‐eligibility is simply too naÏve and rigid from a mortgage underwriting standpoint.

Sound underwriting practices are based on the principle of the “three C's” of underwriting; creditworthiness (i.e., willingness‐to‐repay), capacity (ability‐to‐repay) and collateral (skin‐in‐the‐game). With fully‐documented loans, DTI is an important factor in most underwriting assessments; however, it has some limitations to its effectiveness.

DTI misses the mark, at times, by not providing a true representation of the borrower's capacity to repay their obligation. Take, for example, two borrowers with a 50% DTI: one with a $75,000 income, the other $500,000.

The first borrower would have a much tougher time handling their payments based on their residual income than the second borrower. Moreover, what if that second borrower had a 720 FICO and 80% loan‐to‐value (LTV) ratio? Could this not be a low‐risk loan? These compensating factors would offset the incremental credit risk associated with the 50% DTI. Herein lies the fallacy of QM's 43% DTI limit, as well as the hue and cry over the QM patch.

Overexaggerated Risk

The risk of greater than 43% DTI mortgages is overplayed.

To make this point, I selected a random sample of 300,000 fixed‐rate mortgages, originated between 2014‐2016 and sold to Fannie Mae and Freddie Mac. Subsequently, I used this data to estimate a statistically‐based underwriting model similar to what both GSEs use in making their credit assessments today.

The question I sought to answer in this analysis is whether loans with DTIs greater than 43% are riskier than other loans, taking into account all risk factors attributed to the loan.

Loans were classified into one of two categories: performing and nonperforming. A nonperforming loan was defined as ever having attained 90 days past due or worse in its history. In addition to DTI, several industry standard underwriting risk factors were included in the model, such as FICO score, original LTV, origination channel (e.g., retail, correspondent or broker), occupancy status, loan purpose, property type, number of borrowers, number of units and a first‐time homebuyer indicator.

Several different combinations of these variables were tested in the logistic regression, including a version with both DTI and a variable for whether the loan was above 43% DTI. This specification sought to identify any incremental risk in loans with DTIs over 43%, beyond the normal relationship of DTI and delinquency risk.

The coefficient on the DTI variable was statistically significant and indicated that every 1% increase in DTI would lead to a 2.7% increase in the 90‐day‐past‐due rate. However, the variable for greater than 43% DTIs was not significant, indicating that the risk of these loans is not inherently riskier than loans with DTIs less than or equal to 43%, considering the standard effect of DTI on delinquency (along with the other risk factors).

This result corroborates the compensating risk factors argument that GSE credit policy incorporates risk factor offsets in such a manner as to neutralize effectively the marginal risk of higher DTIs. More importantly, it underscores the point that acceptable quality mortgages are not limited to 43% DTI.

Parting Thoughts

The 43% limit should be abandoned in QM and replaced with provisions that define QM‐eligibility according to GSE, FHA and VA underwriting standards. To set their credit standards, GSEs and the FHA rely on their own automated underwriting systems (AUS), which include multivariate statistical scorecards predicting serious mortgage delinquency.

As determinants of credit quality, such processes align with longstanding industry underwriting practices of taking all aspects of the borrower, property, and loan into account. For loans not sold to the GSEs or FHA or VA (they are permanently exempted from the 43% DTI limit), regulators should review the credit decisions of the largest portfolio lenders to make sure their credit risk assessments are aligned with their companies' risk appetites and that their underwriting systems are comparable to the comprehensive AUS approaches employed by the GSEs.

Abandoning the 43% DTI limit would remove artificial underwriting barriers that are simply not needed today, without posing significant credit risk to the market.

Clifford Rossi (PhD) is Professor‐of‐the‐Practice and Executive‐in‐Residence at the Robert H. Smith School of Business, University of Maryland, and a Principal of Chesapeake Risk Advisors, LLC. He has nearly 25 years of experience in financial risk management, having held a number of C‐level positions at major banking institutions. Prior to his current posts, he was the chief risk officer for Citigroup's North America Consumer Lending Division.


BylawsCode of ConductPrivacy NoticeTerms of Use © 2023 Global Association of Risk Professionals