Major Treatment Findings
Table 1 presents the univariate results from testing our central hypotheses. Specifically, Panel A reports the borrower’s stated intention to engage in forbearance ranging from 1 (definitely will CONTINUE paying my mortgage) to 7 (definitely will STOP paying my mortgage) if there is no screening of financial (income) hardship involved. A cursory glance immediately reveals many borrowers simply will (1) and will not (7) participate in this opportunity. Combining categories 6 and 7, 24.22% of our participants report they would likely stop paying their mortgage under such a scenario. This number, while very much in line with the previously reported Lending Tree survey results, is quite large, particularly if it is generalizable to the broader universe of all 50 million U.S. residential mortgage loans which account for approximately $11 trillion in financial capital.Footnote 23
Panel B of Table 1 reports similar information but employs our second pool of borrowers whose forbearance participation was conditioned on a required attestation of financial need with lender recourse. That is, before they are allowed to forbear, they are required to sign a 1-page document stating they are “experiencing a COVID-19 related decline in income.” It is further explained that “after the pandemic is over, the lender will review all mortgage forbearance cases, and if you are found to have participated without experiencing a COVID-19 related decline in income, stiff penalties will be enforced.” The purpose of this attestation with recourse is to weed out those who are looking to free ride on the government’s stimulus package, and thereby mitigate the potential bailout costs to U.S. taxpayers. This minor requirement results in a statistically significant (at the 99% confidence level) reduction in the number of borrowers who plan to forbear at all points along the mortgage payment cessation portion of the scale (6 and 7), reducing the total number of borrowers who would stop paying their mortgage from 24.22% to 15.85%, an 8.37% reduction.
Since some loans are not covered under the CARES Act, we investigate in Panels C & D how borrowers might respond under private label programs. More specifically, Treatment 3 in Table 1 is segmented into two panels. Panel C reports the number of borrowers who would forbear IF approved, whereas Panel D shows the necessary condition of first applying for forbearance eligibility. Recall, at least one major private label institution’s approach with borrowers is to require them to complete a financial hardship verification package. Within seven days, the servicer then notifies them if they are eligible to stop paying their mortgage. The average score associated with those who would apply is 3.70 (where 1 = will NOT apply; 7 = will apply). If a borrower is truly in financial need, submitting an application prior to initiating forbearance seems to represent a relatively minor hurdle.Footnote 24 That said, because the time involved in collecting one’s financial documents comes with potentially non-trivial or even substantial search costs, it may well mitigate the free rider problem and only remain attractive to those with a reasonable expectation of receiving payment cessation assistance. The results from Panels C & D suggest 25.32% of borrowers would apply for forbearance, and if approved roughly 18.41% would accept it.Footnote 25
Conceptually, it is important to note that there is no recourse to borrowers in either Treatment 1 or 3. In Treatment 1, they are automatically approved for forbearance, whereas in Treatment 3, they must apply. In neither scenario can they be argued to have engaged in technical/legal wrongdoing because they are fully welcome to participate. Instead, a reduction in income related to COVID-19 is either trusted (Treatment 1) or verified (Treatment 3). It is only in Treatment 2 where a borrower can retroactively be held responsible via a post-mortem lookback provision. As such, it is Treatment 2 that results in a statistically significant reduction in the number of borrowers who would strategically forbear and stop making their mortgage payments in the absence of a significant, pandemic related income disruption.Footnote 26
Allocation of Capital upon Forbearance
Seeking additional insight, Table 2 investigates where borrowers will allocate their would-be mortgage payments if they decide to forbear. These results are further segmented by treatment number and whether borrowers indicated they would versus would not continue to make their mortgage payments. Interestingly, 7.48% of these newly available funds would reportedly be invested directly into the stock market which has experienced unprecedented volatility since the outbreak of the pandemic.Footnote 27 While this potential capital infusion may be highly welcomed by equity market participants (and provide some measure of price support and stability), taxpayers bear the risk in that if the market does well, forbearing borrowers capture all the upside gains, while if the market falls, borrowers default and lenders (or taxpayers) bear the brunt of the financial consequences.Footnote 28On the other hand, 5.24% would reportedly be invested in much safer CDs, TIPS or T-Bills, while 21.15% of funds would be held in cash, presumably reflecting uncertainty surrounding how long this pandemic will restrict the ability to earn a living.Footnote 29 Similarly, the category receiving the greatest allocation of funds (31.31%) is that used to buy necessities such as food and clothing. While not directly related to housing market outcomes, these uses of funds do enhance the social safety net of potentially impaired borrowers and may well provide a needed buffer and level of support to at-risk individuals and communities. To the extent such allocations mitigate extreme financial hardship and facilitate successful long-run mortgage repayments and/or modifications, they may not represent an actual deadweight cost to this policy intervention.
In addition, some participants would re-allocate mortgage funds to various forms of consumer debt. The CARES Act was designed to ease the financial burden of citizens by softening their requirement to make mortgage payments, but we also find some of the money saved by missing mortgage payments will likely be spent to reduce debt in other areas of the consumer’s balance sheet. This leads to the question, “Is it fair to lenders/servicers that credit card companies, student loan sources, and other credit offering institutions like auto and furniture sales financiers are using the money lenders would have received in the form of mortgage payments to reduce their risk exposure?” This reallocation of funds fundamentally shifts the risk profile of these investments, and therefore the interest rates each party would have charged had they known ex-ante the way the government was going to handle this black swan event.
As a singular example, delaying mortgage payments not only increases the likelihood of eventual default (or, at a minimum, the need to modify the loan at the end of the forbearance period), but pushing the payments to a later point in time also increases the effective duration of the loan, making it more sensitive to changes in interest rates. Since interest rates have been reduced to near zero, it seems disproportionately likely they may well go up after the crisis is resolved. Taken together, these assertions suggest that when interest rates rebound, the market value of outstanding mortgages will decrease even further than what would have been observed had the CARES Act not been designed in this fashion. Conversely, with respect to other forms of consumer debt, by allowing missed mortgage payments to go toward paying down these competing balances, the CARES Act advantages credit card, auto loan, student loan, and other consumer credit issuers at the expense of mortgage lenders. In terms of economic magnitude, using the numbers from Table 2, amongst borrowers indicating they are likely to “stop paying” their mortgages, an estimated 24.82% (8.77% + 3.80% + 4.00% + 8.25%) of the forbearance proceeds will be reallocated toward reducing other debts.Footnote 30
Implications and Magnitude of Findings
Returning to the overall magnitude of our findings, suppose 5% of these reallocated cash flows ultimately end up in serious delinquency or default, and therein incur a loss rate conditional upon default of 40% (note: we view both of these estimates as conservative assumptions given that delinquency rates on all CARES Act covered products experienced both delinquency and loss rates significantly higher than these levels during the global financial crisis of 2008-09).Footnote 31 Under these assumptions, without intervention nearly $200 million of capital per month that would have been collected by mortgage lenders will never be recaptured. Our simple attestation approach could prevent approximately $50 million per month of these losses. That said, we readily acknowledge these numbers are highly speculative, as in these unprecedented times both default/delinquency rates and/or loss rates conditional upon default could skyrocket, or alternatively, future government assistance programs targeted at either distressed borrowers or mortgage lending institutions could significantly soften the economic impact.Footnote 32
In many fields, experimental results are discussed only as they relate to univariate analysis since the environment is controlled for at the design level. Nevertheless, we recognize additional exogenous variables may well meaningfully impact our results. As such, we next introduce an array of potential explanatory variables. Table 3 reports descriptive summary statistics for the variables we, and/or others across the previous literature, have argued may impact mortgage forbearance proclivities. We report both overall summary statistics as well as results segmented by treatment group. P values correspond to one-way ANOVA tests of whether the means across all three treatment groups significantly differ.
Beginning with respondent attributes regarding economic expectations and personal beliefs, unique to this study is the finding that 39.06% of borrowers in our sample find engaging in strategic forbearance – the act of stopping mortgage payments even when the borrower can afford to continue paying – is immoral. Continuing, the vast majority of those in our sample view their home as more of a consumption good rather than as an investment, while our subjects are slightly bearish on future home prices in their city. Consistent with current national trends, there is a nearly even split across experimental participants with respect to political affiliation and ideology.
Turning to the depth of personal experience with, and/or exposure to, the pandemic, nearly 11% of participants have a family member who has been diagnosed with, or has good reason to believe they have contracted, the coronavirus, while 19% know of at least one person inside their close circle of friends who has been infected. In terms of financial sophistication, our pool of sample participants is deemed above average in terms of financial literacy and more highly educated (76.60% have completed at least a 4-year college degree) relative to both prior studies of mortgage market outcomes and society at large. Nearly 1 in 7 (or 14% of) borrowers in the sample have previously defaulted on a mortgage, with more than 1 in 6 of those defaults being strategic in nature.
From a behavioral perspective, 20.94% of our sample participants are deemed overconfident. Additionally, they lean towards being more conscientious and agreeable. Finally, a cursory review of their demographic profiles reveals our homeowners are broadly similar to those surveyed in previous studies in terms of income, ethnicity, and gender, with the exception that our participants are slightly younger (38.37 years old). Thus, taken together, we view our pool of experimental participants as being broadly reflective of the universe of U.S. residential mortgage borrowers.
Table 4 reports the results from two different regression specifications (OLS and Ordered Probit) where the dependent variable is a Likert scale measure of borrower self-reported likelihood of participating in a CARES Act related mortgage forbearance. This variable ranges from 1 if the borrower plans to definitely continue paying their mortgage to 7 if they definitely plan to stop paying their mortgage and exercise their strategic forbearance option. While OLS estimates are provided for consistency with the previous literature, we also estimate an Ordered Probit since the dependent variables are both ordinal and constrained. Not surprisingly, both models provide qualitatively similar results. Notably, the observed statistical significance on the Treatment 2 dummy variable confirms our univariate finding that the simple requirement of attestation with recourse results in a reduction in the incidence of strategic forbearance. In terms of economic magnitude, classification tests applying these estimated (Ordered Probit) coefficients to our actual sample respondent characteristics suggest our focal “Borrower Attestation with Recourse” approach fundamentally alters the predicted forbearance outcome for 6.23% (66 of 1060) of our sample borrowers. More specifically, 16 (of 347) borrowers exposed to Treatment 2 would likely have pursued forbearance had attestation not been required, while 50 (of 713) not exposed to the attestation mandate would likely have been dissuaded from forbearance had such a requirement been in place.Footnote 33 Turning to our control variables designed to capture differences in economic incentives and beliefs, strategic forbearance morality is quite robust and consistent with expectations in that those who find it morally objectionable are significantly less likely to forbear. Somewhat surprisingly, none of our remaining controls along this dimension, including Political Affiliation, exhibit statistically significant explanatory power. Thus, we conclude strategic borrower behavior during the crisis cuts across party lines, at least in terms of forbearance participation.
Turning to our second set of controls relating to financial sophistication and personal experience, our results suggest those borrowers with an immediate family member who has contracted the virus are significantly more likely to strategically forbear. However, if COVID-19 has only infected the borrower’s close circle of friends, it is not enough to significantly alter mortgage payment cessation behavior. Additionally, with respect to borrower specific behavioral attributes, three of the Big 5 personality traits are significant. Specifically, extraversion, agreeableness, and neuroticism are positively associated with mortgage payment cessation. With respect to demographic attributes and controls, higher income borrowers, Caucasians, and women are significantly more likely to continue making their mortgage payments, possibly because they are simply more able to do so. On the other hand, neither age nor familial status appear to significantly influence forbearance probabilities.Footnote 34
Estimating the magnitude of these potential effects, if we were to apply these results to all mortgages, this could result in a reduction of 3.12 million (50 million * 6.23%) fewer mortgages going into forbearance, impacting almost $685.3 billion ($11 trillion * 6.23%) in loans. Given an average mortgage payment of approximately $1250 per month, this results in a reduction of $3.89 billion (50 million * $1250 * 6.23%) per month in lost payment revenues to lenders. If only applied to those mortgages explicitly covered under the CARES Act, the numbers would still result in 1.93 million (50 million * 6.23% * 62%) fewer mortgages going into forbearance, impacting roughly $424.89 billion ($11 trillion * 6.23% * 62%) in loans, and impacting capital flows to servicers by approximately $2.41 billion (50 million * $1250 * 6.23% * 62%) per month.
We next explore where forbearing borrowers will allocate their money if they stop paying their mortgage. Table 5 presents the results from the first stage of this analysis where the dependent variable in Column 1 is set equal to 1 (35.8% of subject respondents) if the borrower would invest any positive percentage of their forborne proceeds into the stock market, and 0 (64.2% of subject respondents) if the borrower would invest nothing into the stock market. Examining the determinants of this allocation decision, all Economics and Beliefs control variables except Number of Months are statistically significant. Specifically, those without a moral objection to strategic forbearance are more likely to invest in the stock market, as are both those who have been in their home the longest and those who view their home as more of an investment than a consumption good. Consistent with a bullish economic outlook, if the borrower believes home prices will increase over the next 12 months, they are also more likely to invest in the market. Interestingly, Republicans are significantly more likely to invest their forgone mortgage payments in the stock market than are Democrats. In light of the aforementioned emerging evidence on disparate behavioral responses to COVID-19 policy innovations across party lines, we do not find this result overly surprising.
With respect to our behavioral attributes, not surprisingly, more conscientious borrowers appear reluctant to divert forgone mortgage payments away from their expressly authorized purpose into speculative activities, while consistent with the notion that more neurotic borrowers prefer to avoid highly stressful situations, such borrowers are more reluctant to invest their newfound proceeds in the relatively volatile equity markets. Lastly, examining the financial sophistication and demographic controls, younger people, those who earn more and/or have completed college, and ethnic minorities are all also more likely to invest in the stock market.
Turning to alternative uses of the forborne cash flows, Column 2 explores what factors and attributes lead mortgage borrowers to divert these sums into paying off various other forms of debt. Following our previous analyses, we set the dependent variable equal to 1 (65.8% of subject respondents) if the borrower would divert any positive percentage of their forborne proceeds to the repayment of outstanding debts (i.e., pay down credit cards, student loans, auto loans, and/or other debts), and 0 (34.2% of subject respondents) otherwise. From a purely economic perspective, those same factors which lead wealth maximizing borrowers to divert forborne payments into the equity market provide similar incentives for borrowers to consolidate their debt away from relatively high cost products into lower cost alternatives. As such, it is entirely unsurprising that the directionality of all Economics and Beliefs attributes remain consistent, though admittedly exhibit reduced explanatory power and statistical significance, as we move from Column 1 to Column 2. Additionally, direct personal exposure to the consequences of this pandemic appears to motivate debt consolidation proclivities in an economically meaningful fashion. With respect to borrower attributes, only Financial Literacy, Conscientiousness, and Child Dummy exhibit robust statistical significance, with each of these relations potentially attributable to past choices, behaviors, and risk-aversion.
Finally, as the continuing global pandemic increases the economic uncertainty confronting millions of American families, many mortgage borrowers may rationally respond by using forborne mortgage proceeds as a source of precautionary savings. To explore this possibility, we define and identify borrowers with precautionary savings motives as those who would simply retain the forborne proceeds in cash (i.e., “Hold onto the money – don’t invest anywhere”) or cash equivalents (i.e., “Invest in a low risk instrument (like a CD, TIPS or T-Bills”). As before, in Column 3 we set the dependent variable equal to 1 (70.4% of subject respondents) if the borrower would allocate any positive percentage of their forborne proceeds along either of these two precautionary savings dimensions, and 0 otherwise. Consistent with our previous arguments, each of our Economics and Beliefs attributes continue to retain their previously observed sign patterns, though these relations now generally fail to exhibit statistical significance at conventionally accepted levels. Both our Financial Sophistication & Experience and Behavioral Characteristics similarly fail to provide statistically significant new insights, while from a Demographics perspective male and minority borrowers appear to be more likely to embrace precautionary savings motivations with respect to forbearance proceeds.