The GSE program that enabled borrowers to take on additional debt helped fuel price appreciation for lower-priced homes

By Edward J. Pinto and Tobias Peter.

"In July, the Consumer Financial Protection Bureau (CFPB) announced that it intends to allow the so-called qualified mortgage (QM) Patch to sunset according to its terms in 2021. Originally announced in 2013, the Patch exempted the government sponsored enterprises (GSEs), Fannie Mae and Freddie Mac, from adhering to the 43% debt-to-income (DTI) limit rule, which bestows safe harbor legal status to any lender making such loans.

A recent CoreLogic blog post concluded that, “Loan application data illustrates that the current GSE Patch disproportionately benefits younger millennials and retirees, Non-W-2 borrowers, low-income borrowers, and borrowers purchasing low-to-mid priced homes (emphasis added). Thus, these groups are disproportionately represented within the GSE Patch and will switch to Non-QM, absent additional policymaking by the CFPB.” In a related blog post, the author notes that minority borrowers and those purchasing homes in underserved neighborhoods similarly depend on the GSE Patch. Observing this dependence, the author’s implicit conclusion is that the GSE Patch helps low-income groups achieve homeownership, and thus these groups would be harmed should the GSE Patch expire in accordance with the CFPB’s plan.

In 1850, French economist Frédéric Bastiat observed:¹

In the economic sphere a law produces not only one effect, but a series of effects. Of these effects, the first alone is immediate; it appears simultaneously with its cause; it is seen. The other effects emerge only subsequently; they are not seen; we are fortunate if we foresee them.

The CoreLogic analysis focuses on the seen—borrowers who took advantage of the Patch. We shall focus on two other effects highlighted by Bastiat: the foreseeable (higher home prices resulting from the Patch) and the unseen (borrowers who would not have needed the Patch, had the Patch-induced home price inflation not occurred).²

This analysis starts with the effect that the GSE Patch has had on home price appreciation (HPA), especially for the entry-level price segments (low and low-medium). The chart below shows the cumulative home price appreciation by price tier. HPA for the entry-level segment has been much faster than for the move-up segment (medium-high and high).

The GSE Patch enabled borrowers to take on additional debt, or leverage (and a similar effect occurred for FHA borrowers under FHA’s own exemption from the QM 43% rule). The GSEs and FHA were in a favored position to acquire/insure loans that exceeded the DTI limit of 43% imposed on private lenders. This “seen” of additional leverage helped fuel price appreciation for lower-priced homes at a much higher pace than for higher-priced homes, a trend that was both foreseeable and may now be estimated.


There has been a correlation between faster home price appreciation and the prevalence of loans with high DTIs.³ As one can see in the chart on the left below, from 2012-2018, census tracts that had above average DTIs (those above 37%) experienced HPA that is faster, and in many cases much faster, than the county average. The same relationship can be seen in the chart to the right, which classifies census tracts by the share of loans with a DTI > 43% rather than by the tract-level average DTI. Thus, a policy that assists buyers in taking on high DTI levels simultaneously undermines home affordability, driving up home prices faster than they would have, absent the Patch-provided stimulus.⁴


Given that the market has experienced this unintended, but foreseeable outcome, it is worth considering a counter-factual of what home prices might have been like had the Patch not been enacted.

As a counter-factual, we will assume that in the absence of government provided leverage to help fuel the home price boom, census tracts with more rapid HPA would have only appreciated at the same rate as the county average. In effect, this leaves the rate of HPA for census tracts with below county level HPA unchanged, but it reduces the rate of HPA for tracts with above county level HPA. Tracts with the greatest appreciation, therefore, experience the greatest slowdown in HPA in the counter-factual analysis.⁵ This slower HPA in the counter-factual example would have resulted in lower home prices over time. Lower home prices would have resulted in a corresponding reduction in DTIs for borrowers over time.⁶

The chart below shows the share of borrowers with a DTI > 43% by income bin. The red line represents the actual results by income for GSE purchase loans with DTIs greater than 43% in 2017, and the yellow line represents the same in 2012 before the Patch and the boom in home prices began. Two things stand out: 1) In 2012, about 17% of borrowers with incomes below $80,000 had DTIs above 43%. By 2017, this had risen to about 26%, and 2) For higher income borrowers (those with annual incomes of at least $80,000), the reliance on the Patch had risen less, from about 11% in 2012 to 17% in 2017. The sharper increase in reliance on DTIs above 43% for borrowers with incomes below $80,000 is due to the more rapid home price appreciation for entry-level homes.

The third line in the chart, the dark blue one, represents the counter-factual, assuming that census tracts with more rapid HPA had only appreciated at the county average. The DTIs for these counter-factual borrowers have been reduced by the same percentage as the reduction in home price growth. The 2017 counter-factual line is marginally above the 2012 actual line, which implies that across all income levels, buyers would have had little additional need for DTIs above 43% in 2017. Especially for low-income borrowers with income below $80,000, the reliance on DTIs greater than 43% would have been, on average, unchanged from 2012.


While the counter-factual illustrates a hypothetical world without the Patch, reality is that the Patch was enacted, and we therefore need to think about how its sunset will affect borrowers. Since its inception, the number of GSE borrowers reliant on the Patch has increased. Due to misguided policies, there has been an explosion in the number of loans with DTIs greater than 45% since mid-2017, which the GSEs have seemingly started to rectify.

As the chart shows, the share of GSE loans with a DTI of 46-50% increased from on average around 6% before July 2017 to 20% in December 2018, but has since then fallen back to 16% in May 2019, whereas the trend of loans with a 44% or 45% DTI is largely unchanged at about 7-8%. Based on this trend, the Patch’s planned sunset in 2021 will provide the GSEs plenty of time to shrink that share further. Once that has been achieved, borrowers with DTIs of 44% and 45% can more easily adjust their DTIs downward to 43%, which will allow them to stay within the QM domain. As this adjustment occurs, the rate of home price appreciation will slow to more sustainable levels.


While it is true that today lower income and minority borrowers have had an increased reliance on DTIs greater than 43% (facilitated by the Patch), this has been the result of government provided leverage that has helped fuel rapid HPA. Without this leverage, the need for DTIs in excess of 43% would have been considerably lower. Pro-cyclical policies such as the Patch guarantee their own necessity.

1. Bastiat, Frédéric. “What Is Seen and What Is Not Seen.” (1850)
2. Pinto, “CFPB’s new ‘qualified mortgage’ rule: The devil is in the details”, January 2013, http://www.aei.org/publication/cfpbs-new-qualified-mortgage-rule-the-devil-is-in-the-details/
3. This analysis has been replicated when controlling for CLTV buckets. For loans in the CLTV >= 90% bucket, the same trend holds: Higher average DTIs in census tracts (and higher shares of loans with DTI > 43%) are correlated with higher ratios of tract to county house price appreciation. For loans in the CLTV < 90% bucket, the trend also holds, except for census tracts with the lowest average DTIs and the lowest share of loans with DTI > 43%.
4. The data used for the counter-factual analysis combine public records data from First American via Data Tree, HMDA, CoreLogic LLMA, Black Knight McDash, Fannie Mae and Freddie Mac Loan Performance data, and FHA Snapshot data. The data are weighted quarterly by loan type at the county level to make them representative. The final dataset consists of 11.5m loans.
5. Note that by using the county average, this approach likely still overstates the HPA for tracts with higher DTIs because the county average itself has been elevated by tracts with access to higher leverage.
6. There is a linear relationship between home prices growth and DTI growth. A 10% increase in a home’s price will result in a 10% increase in borrower DTI holding all else equal."