Board of Governors of the Federal Reserve System

March 04, 2021

Why is the Default Rate So Low? How Economic Conditions and Public Policies Have Shaped Mortgage and Auto Delinquencies During the COVID-19 Pandemic

Lisa Dettling and Lauren Lambie-Hanson

Summary

Delinquencies and defaults on household debt typically closely follow the business cycle. As economic conditions deteriorate, falling employment and incomes put a strain on family finances, leading to a rise in missed debt payments and defaults. Yet, against the backdrop of a historic rise in unemployment associated with the COVID-19 pandemic, delinquencies have fallen. This FEDS Note documents trends in delinquency on mortgages and auto loans during the COVID-19 pandemic, and unpacks how changes in economic conditions and public policies have been associated with borrowers’ debt repayment behavior.

Trends in mortgage and auto delinquencies

We begin our analysis by examining trends in delinquencies and unemployment. We measure mortgage and auto delinquencies using data derived from the Federal Reserve Bank of New York/Equifax Consumer Credit Panel (CCP), a nationally representative, anonymized 5 percent random sample of adult consumers with Social Security numbers and a credit history.¹ Figure 1 displays the data for two time periods: panel (a) shows the data encompassing the Great Recession period, while panel (b) displays data for the COVID-19 recession period.

Figure 1. Delinquencies were pro-cyclical in the Great Recession, but counter-cyclical During COVID-19

The patterns present a stark contrast: during the Great Recession, mortgage and auto delinquencies rose sharply as the unemployment rate began to rise, with mortgage delinquencies in particular following the unemployment rate in near lock-step. But during COVID-19, as the unemployment rose there was no commensurate increase in auto or mortgage delinquencies, and in fact, both fell throughout the pandemic. This FEDS Note seeks to understand the reasons for these divergent patterns.

Factors shaping delinquencies during the COVID-19 Pandemic

One key feature of the pandemic that differentiates it from previous recessions is the widespread availability of loan forbearance. In particular, the Coronavirus Aid, Relief, and Economic Recovery (CARES) Act, which was passed by Congress in March 2020, included a generous forbearance program that allowed homeowners with federally-backed mortgages to enter into (penalty-free) forbearance for up to a year if they had trouble making payments due to the pandemic. And although auto loans and other types of mortgages were not directly covered under this provision, lenders ultimately offered forbearance on many of those loans on their own volition, perhaps in part because the Act provided lenders guidance that they did not need to classify loans in forbearance as troubled debt restructurings in their accounting and disclosures.² Furthermore, the CARES Act specifies that loans that are in forbearance may not be reported as delinquent if they were current at the time the accommodation was granted.³ But even beyond freezing account status, new research shows that some loans which were in delinquency prior to the pandemic entered into forbearance and were then reported to the credit bureaus as current (Haughwout, Lee, Scally, and van der Klaauw, 2020). All told, this suggests there may have been substitution away from delinquency and into forbearance during the COVID-19 Pandemic, an option that was not widely available in past recessions. To better understand borrowers’ ability to service their debts during COVID-19, we therefore need to examine the share of borrowers with loans which are either in delinquency and/or in forbearance.

Figure 2 plots trends in unemployment and the share of borrowers who were delinquent and/or enrolled in forbearance, where forbearance is defined broadly as any type of loss mitigation program wherein the borrower is not required to make a full payment.⁴ In contrast to the pattern in figure 1b, figure 2 indicates that the share of borrowers with a mortgage or auto loan in either delinquency and/or forbearance was strongly pro-cyclical during the COVID-19 pandemic, closely following the path of unemployment. In other words, weakening economic conditions were indeed correlated with non-payment during the pandemic. But rather than being classified as delinquent, those who could not pay were often in forbearance—and labelled as current even if they were not making payments—during the pandemic.⁵

Figure 2. The share of borrowers with a mortgage or auto loan delinquent and/or in forbearance was pro-cyclical during COVID-19

In addition to a strong time series correlation, there is also a strong cross-sectional correlation between unemployment rates and rates of delinquency and/or forbearance in 2020. Figure 3 indicates that states with elevated unemployment rates in June also had higher rates of delinquency and/or forbearance for both mortgages and auto loans. This further confirms the importance of including borrowers in forbearance in an analysis of trends in, and determinants of, families’ ability to meet their debt obligations during the COVID-19 pandemic.

Figure 3. State-level unemployment rates are positively correlated with mortgage and auto delinquency and/or forbearance rates

Another key feature of the CARES Act was generous income support. In particular, stimulus payments, unemployment insurance expansions, and forgivable loans to small businesses who retained employees helped to replace lost income. Stimulus payments were distributed by income and family size, so that a family of four with an adjusted gross income under $150,000 received a one-time payment of $3,400 in April 2020. Unemployment insurance was expanded substantially, including an expansion of eligibility to self-employed and gig workers, and an additional $600 a week in payments on top of the standard state-determined benefits through the end of July 2020. As a result, the majority of unemployed workers faced income replacement rates over 100 percent, whereas in a typical recession replacement rates are closer to 40 percent (Bhutta, Blair, Dettling, and Moore, 2020; Ganong, Noel, and Vavra, 2020). Finally, small businesses were eligible for forgivable loans if they retained employees through the paycheck protection program (PPP).

Together this income support from the CARES Act led incomes to rise during the COVID-19 pandemic. Figure 4 displays data on aggregate personal incomes, with and without government transfers, obtained from the Bureau of Economic Analysis.⁶ While non-transfer income fell during the pandemic—consistent with rising unemployment—total income including transfers increased significantly. This reflects the relative generosity of CARES.⁷ Research indicates that more generous transfer assistance for the unemployed can significantly reduce mortgage delinquencies (Hsu, Matsa, and Meltzer, 2018), suggesting CARES support for incomes likely averted delinquencies and potentially also enrollment in forbearance during the COVID-19 pandemic.

Figure 4. Aggregate personal incomes rose during the pandemic because of the expansion of transfer income under the CARES Act

Unlike in the Great Recession, when house prices were falling and many mortgage borrowers were subject to an equity shock, the residential real estate market in the pandemic has been strong. In the Great Recession, reductions in household liquidity coupled with declines in home equity combined to form a “dual trigger” of mortgage default (Foote, Gerardi, and Willen 2008; Elul, Souleles, Chomsisengphet, Glennon, and Hunt, 2010; Scharlemann and Shore, 2016). But in the pandemic, home prices rose 6.7 percent nationally between February and October 2020 (Katz, 2020), bolstered by low mortgage interest rates and limited housing supply. Between 2019 Q3 and 2020 Q3, U.S. homeowners gained an estimated $1 trillion in aggregate equity, an increase of 10.8 percent, and the share of mortgaged properties with negative equity fell from 3.7 percent to 3.0 percent (CoreLogic 2020).

Another key feature of the current economic downturn is the pandemic itself, as well as associated social distancing policies and lockdowns. Both have likely had direct effects on families’ ability to meet their debt obligations. For example, families who test positive might be required to self-quarantine, which could result in income losses. And higher local case counts could discourage economic activity and have negative spillover effects on employment and incomes. Similarly, state and local lockdowns and other social distancing policies could lead to further reductions in economic activity (above and beyond the effects of case counts).⁸

Of course, economic and health conditions are related to one another, and federal and state policies are not enacted in a vacuum. For example, increasing case counts spur more stringent social distancing policies, which could spur changes in employment, and so on. To consider the joint effects of economic conditions, fiscal policies, social distancing policies and health, we compiled monthly data on state and local health and economic conditions and the government policy response so that we can estimate multivariate regression models of the determinants of borrower nonpayment, as captured by the share of borrowers delinquent and/or in forbearance, shown in figures 2 and 3.

There is considerable variation both across time and space in the severity of the pandemic, economic conditions, and the government policy response. For example, figure 5 shows state-level data from April, June, and August 2020 on the stringency of state social distancing measures (obtained from the Oxford COVID-19 Government Response Tracker) and cumulative cases per 1,000 people (obtained from The New York Times). There is considerable variation across states and time in caseloads and social distancing policy; and states with more cumulative COVID-19 cases have not always had the strongest policy responses, and vice versa. Indeed, in June higher state-level case counts were positively correlated with the stringency of state social distancing measures, but by August the correlation had flipped, perhaps reflecting the efficacy of those policies. This variation will help us to separate the effects of cases and the government policy response on families’ ability to pay their debts.

Figure 5. State social distancing policy stringency not consistently correlated with COVID-19 case loads

There is also variation across time and place in the generosity of the economic support available through the CARES Act. For example, the extra $600 supplement to unemployment insurance replaced a larger share of income in lower wage states, and the stimulus payments went to a larger share of the population in lower income states. Figure 6 displays state-level variation in an index of the generosity of CARES Act income support (i.e., UI, stimulus checks and PPP) relative to pre-pandemic incomes, based on quarterly data obtained from the Bureau of Economic Analysis.⁹ There is considerable variation across states in the generosity of CARES Act income support.

Figure 6. CARES Act income support varied in its generosity

To understand the importance of government policy-mandated forbearance—that is, forbearance on federally backed mortgages through the CARES Act, as opposed to voluntary forbearance on other mortgages and loans—we also compiled data on the share of mortgages in each county held in GSE securities or federally backed through the FHA, VA, or Rural Housing programs, using Black Knight McDash data.¹⁰ CARES-covered mortgages made up an estimated 90 percent of active first-lien mortgages in Alaska, as compared to 69 percent in New York and California (where jumbo loans—those with loan amounts too large to be federally backed—are more common).

Figure 7. Estimated Share of Active Mortgages in June Eligible for CARES Act Protection

With the county and state level data in hand, we then estimate a simple multivariate regression model of the association between rates of delinquency and/or forbearance and economic conditions, cases, and the government policy response. We focus our analysis on the period March through September 2020, and the main independent variables of interest are county unemployment rates ($${urate}_{ct}$$), state-level CARES income support generosity ($${CARES\ Inc}_{st}$$), the share of mortgages in each county eligible for federal mortgage forbearance under CARES ($${CARES\ Mtg}_{ct}$$), cumulative county case counts per thousand people ($${cases}_{ct}$$), and the stringency of state-level social distancing policies ($${stringency}_{st}$$). Because many of these variables are indices without a specific economic meaning, and because we wish to compare the effects of each, we standardize all of the independent variables of interest, setting the mean to zero and standard deviation to one. We then estimated the following model using ordinary least squares:

$$$$ y_{cst} = \beta_1{urate}_{ct} + \beta_2{CARES\ Inc}_{st} + \beta_3{CARES\ Mtg}_{ct} + \beta_4{cases}_{ct} + \beta_5{stringency}_{st} + \gamma_s + \delta_t + \varepsilon_{ct} $$$$

Where $$y_{ct}$$ represents the share of either mortgage or auto loans that are in delinquency and/or forbearance in county $$c$$ in month $$t$$ in state $$s$$, and $$\gamma_s$$, $$\delta_t$$ are state and month fixed effects, respectively. Because we incorporate state and month fixed effects, the coefficients can be interpreted as the correlation between within-state changes in economic, health and policy variables and within-state changes in delinquency and/or forbearance. In all specifications, standard errors are adjusted for clustering at the county-level, and all regressions are weighted by the county population. Figure 8 displays the point estimates and associated 95 percent confidence intervals for separate estimates of the conditional correlation between each of the independent variables of interest and the proportion of mortgage and auto loans in delinquency and/or forbearance.

Figure 8. Economic, Health, and Policy Drivers of County-Level Rates of Delinquency and/or Forbearance on Mortgages and Auto Loans

Figure 6 indicates that, all else equal, higher unemployment rates are correlated with a greater percentage of borrowers with loans in delinquency and/or forbearance. For example, a one-standard-deviation increase in the unemployment rate (roughly 4.5 percentage points) is associated with a 1.7-percentage-point greater share of mortgage borrowers in delinquency and/or forbearance, which amounts to about a 20 percent increase over the mean.

Support from the government through the CARES Act may have reduced the share of borrowers in delinquency and/or forbearance. A one-standard-deviation increase in the index of CARES income support generosity is associated with delinquency and/or forbearance rates on mortgages and auto loans that is about 2 percentage points lower, or roughly a 25 percent reduction. A higher share of federally-backed mortgages covered by the CARES mortgage forbearance provision has a statistically insignificant relationship with mortgage delinquency and/or forbearance. This is consistent with the notion that lenders with loans not directly covered by CARES have also voluntarily offered forbearance (although this may have also been an indirect result of CARES, as noted earlier).

A higher cumulative COVID-19 case count is positively correlated with the share of borrowers in delinquency and/or forbearance. A one-standard-deviation increase in case counts per capita is associated with a 1.5 percentage point increase in mortgage delinquency and/or forbearance rates, almost as much as the effects of unemployment. And conditional on case counts, stricter social distancing and other containment measures are actually negatively correlated with delinquency and/or forbearance rates, although the coefficients are not statistically significant at conventional levels.

Conclusion

Our analysis provides new evidence that economic conditions have exerted significant pressure on households’ ability to pay their debts during the pandemic, but that the availability of forbearance programs and fiscal support from the government have thus far kept many families from entering into delinquency. When we look at the joint effects of economic conditions, health, social distancing policies, and fiscal support, we find evidence that both economic and health conditions have increased combined rates of delinquency and/or forbearance, although the effects of economic conditions have been moderated by income support from the CARES Act. Still, we caution that these effects represent conditional correlations, and it is possible there are unobserved variables not captured in our analyses.

Though delinquency or forbearance both reflect underlying financial insecurity stemming from the economic and health effects of COVID-19 on families ability to pay their debts, delinquency and forbearance could have different implications for the macroeconomy and financial stability. For example, new research suggests the availability of forbearance has supported house prices (Annenberg and Scharlemann, forthcoming), which could suggest that widespread forbearance availability may have prevented a negative feedback loop, since falling house prices could increase mortgage delinquency, and so on.

References

Akana, Tom. 2020a. CFI COVID-19 Survey of Consumers — Wave 4 Tracks How the Vulnerable Are Affected More by Job Interruptions and Income Disruptions, Consumer Finance Institute Special Report, Federal Reserve Bank of Philadelphia, retrieved from https://www.philadelphiafed.org/-/media/frbp/assets/consumer-finance/reports/cfi-covid-19-survey-of-consumers-wave-4-updates.pdf (PDF).

Annenberg, Elliot and Tess Scharlemann. Forthcoming. The Effect of Mortgage Forbearance on House Prices During COVID-19, FEDS Note.

Bhutta, Neil, Jacqueline Blair, Lisa Dettling, and Kevin Moore. 2020. COVID-19, the CARES Act, and Families Financial Security. National Tax Journal 73(3): 645-672.

Black Knight Financial Services. 2020a. Mortgage Monitor: April 2020 Report, retrieved from https://cdn.blackknightinc.com/wp-content/uploads/2020/06/BKI_MM_Apr2020_Report.pdf (PDF).

Black Knight Financial Services. 2020b. Mortgage Monitor: September 2020 Report, retrieved from https://cdn.blackknightinc.com/wp-content/uploads/2020/10/BKI_MM_Sept2020_Report.pdf (PDF).

CoreLogic. 2020. Homeowner Equity Insights: Data Through Q3 2020, retrieved from https://www.corelogic.com/insights-download/homeowner-equity-report.aspx.

Elul, Ronel, Nicholas S. Souleles, Souphala Chomsisengphet, Dennis Glennon, and Robert Hunt. 2010. What “Triggers” Mortgage Default? American Economic Review: Papers & Proceedings 100 (May): 490–494

Farrell, Diana, Peter Ganong, Fiona Greig, Max Liebeskind, Pascal Noel, Daniel Sullivan, and Joseph Vavra. 2020. The Unemployment Benefit Boost: Trends in Spending and Saving When the $600 Supplement Ended. JP Morgan Chase Institute Policy Brief, retrieved from https://www.jpmorganchase.com/content/dam/jpmc/jpmorgan-chase-and-co/institute/pdf/Institute-UI-Benefits-Boost-Policy-Brief_ADA.pdf (PDF).

Federal Reserve Board. 2020. November Financial Stability Report (PDF).

Foote, Christopher L., Kristopher Gerardi, and Paul S. Willen. 2008. Negative Equity and Foreclosure: Theory and Evidence. Journal of Urban Economics 64(2): 234-245.

Ganong, Peter, Pascal J. Noel, and Joseph S. Vavra. 2020. US Unemployment Insurance Replacement Rates During the Pandemic. NBER Working Paper 27216, retrieved from https://www.nber.org/papers/w27216.

Haughwout, Andrew, Donghoon Lee, Joelle Scally, and Wilbert van der Klaauw. 2020. Following Borrowers through Forbearance. Liberty Street Economics Blog, retrieved from https://libertystreeteconomics.newyorkfed.org/2020/11/following-borrowers-through-forbearance.html.

Katz, Lily. 2020. Americans Have Gained $2 Trillion in Home Value During the Coronavirus Pandemic, retrieved from https://www.redfin.com/news/real-estate-home-value-gain-coronavirus-pandemic/

Scharlemann, Therese C. and Stephen H. Shore. 2016. The Effect of Negative Equity on Mortgage Default: Evidence from HAMP’s Principal Reduction Alternative. The Review of Financial Studies 29(10): 2850–2883.

The New York Times. 2020. Coronavirus (Covid-19) Data in the United States. Retrieved November 3, 2020, from https://github.com/nytimes/covid-19-data.