Measuring the gap: Refinancing trends and disparities during the COVID-19 pandemic

Background

Amidst widespread economic instability brought on by the COVID-19 pandemic, the Federal Reserve moved to quell market fears by swiftly dispatching its toolkit. It cut its headline policy interest rate to zero and implemented quantitative easing (QE)—large-scale asset purchases that feed liquidity into financial markets. These measures, aimed at mitigating steep declines in economic activity, work in part through household balance sheets to encourage consumer spending by making borrowing more affordable. As a result of the Federal Reserve’s actions, mortgage rates dropped to record lows, leading to a surge in demand for mortgage credit in 2020. However, the pandemic also triggered broader economic changes that affected different household groups in varying ways and raised critical questions about the distributional effects of the Federal Reserve’s COVID-19 monetary policy interventions.

Homeowner financial stability is closely connected to macroeconomic developments (e.g., high unemployment) and resulting monetary policy responses, which affect household balance sheets to varying degrees. With monetary policy easing, many homeowners, aided by historically low interest rates, opted to refinance their mortgages. Refinancing has the potential to improve the financial stability of liquidity-constrained households by providing supplemental liquidity from reduced mortgage payments to increase consumption or savings. On average, borrowers using cash-out refinancing consumed 33 percent of the equity received within a year, and consumption rates were higher for younger households and those with liquidity constraints (Farrell et al., 2020). Lower interest rates also make it easier for borrowers to keep up with payments and substantially reduce the likelihood of default. In fact, a 10 percent reduction in a homeowner’s mortgage payment could reduce their likelihood of default by 27.5 percent (Ehrlich and Perry, 2015). Despite extraordinarily favorable interest rates during the COVID-19 pandemic, mortgage refinancing was not accessible to all households. White, Asian, and higher-income borrowers benefitted disproportionately compared to Black, Hispanic, and low-income borrowers (Agarwal et al., 2021; Gerardi et al., 2020; Gerardi et al., 2021).  

This brief expands on previous analyses of COVID-19 refinancing trends by examining the variation in mortgage affordability and refinancing cost burden across race, ethnicity, and income. We utilized property-level deed records from a proprietary source in conjunction with publicly available HMDA data to evaluate the downstream effects of refinancing and present five key findings:

1. Refinancing rates for Black, Hispanic, and lower-income borrowers were the lowest compared to Asian, White, and higher-income borrowers. Refinancing disparities were evident across all income quartiles, even in the context of a favorable interest rate environment.
2. Roughly 1.2 million eligible mortgage holders were missing refinancers who should have taken advantage of the profound opportunity to refinance but did not.
3. Roughly one in five refinancers opted for a shorter mortgage term, accelerating the pace at which they could pay off mortgage balances and accumulate equity.
4. Liquidity Seeking refinancers saw notable reductions in their payment burden. Additionally, most refinancing households reaped significant wealth gains from refinancing; these benefits were widely distributed across all demographic groups.
5. Black and Hispanic borrowers experienced disproportionately higher refinancing costs. Higher fixed refinancing costs make refinancing at lower loan amounts less advantageous.

The return to zero-bound interest rates provided homeowners an exceptional opportunity to capitalize on historically low interest rates and gain access to wealth-building through mortgage refinancing. Nevertheless, our findings suggest that these gains were unevenly distributed, with significant implications for the widening racial wealth divide and homeowner financial stability. Our findings underscore the importance of lowering barriers to mortgage refinancing and implementing policies that encourage low-income, Black, and Hispanic homeowners to refinance.

Data

We focused on analyzing general refinancing patterns in the U.S. housing market. To that end, we utilized a mix of proprietary and publicly available data sources including information on the financial structure of mortgage loans and borrowers’ attributes. Our analysis is based on a sample of matched CoreLogic Property and Tax Deed data and Home Mortgage Disclosure Act (HMDA) records. CoreLogic compiles real estate transaction data from tax forms and deed records, covering nearly the entire universe of U.S. mortgages. CoreLogic’s microdata provide a comprehensive and detailed view of mortgages, including transaction dates, lien types, loan types, interest rates, and property characteristics, among other things. To gain insights into household characteristics, such as race and income reported at the time of application, we supplemented CoreLogic with HMDA data. HMDA requires financial institutions originating closed-end mortgages or open-end lines of credit to report annually on their mortgage activity.¹ We utilized the linkage between HMDA and CoreLogic to create a panel of mortgage financings by chaining observed refinances to previous financing activity over time.

We matched HMDA and CoreLogic records using a similar process as that used by Goodman et al. (2018). Our harmonized sample includes mortgage records matched on several criteria, including origination year, loan type, census tract, and a string comparator score that maximizes similarity in lender names. Following this matching routine, we obtained a dataset containing possible matches at the property level, assuming that a sequence of loans made on a particular property belongs to the same homeowner if certain conditions are met. We found potential matches for 60 percent of the CoreLogic property-level records.  However, after reconciling instances of many-to-one matches and implementing conservative matching rules to reduce false positives, our match rate was reduced to 33 percent. Further information on our matching procedure can be found in the appendix.²

To understand refinancing patterns, it is essential to identify a sequence of mortgage financing transactions that are likely held by the same homeowner. This process allows us to assess changes in mortgage affordability by linking recent refinances to interest rate conditions that existed before refinancing. To achieve this goal, we created a base sample that includes 6.5 million refinances with linked mortgage sequences, where the most recent first-lien mortgage loan is credibly linked to previous financing events of either a purchase or other refinance. As this report focuses on homeowners’ liquidity preferences and responsiveness to monetary policy rather than their home equity extraction, we focused our analysis on the 1.1 million rate-term refinances present in Sample 2 and excluded cash-out transactions. Additionally, we used this secondary panel to determine the extent of refinancing inertia across racial groups. Using these different sample configurations, we calculated changes in monthly payments, estimated interest expense savings, and analyzed differentials in closing cost burdens. Additional details on the samples used can be found in Table 1.

Table 1: Data Asset Overview