.. CENTER for RETIREMENT RESEARCH at BOSTON COLLEGE WHAT ARE THE IMPLICATIONS OF RISING DEBT FOR OLDER AMERICANS? Angqi Chen, Siyan Liu, and Alicia H. Munnell CRR WP 2023-11 September 2023 Center for Retirement Research at Boston College Haley House 140 Commonwealth Avenue Chestnut Hill, MA 02467 Tel: 617-552-1762 Fax: 617-552-0191 https://crr.bc.edu Angi Chen is a senior research economist and the assistant director of savings research at the Center for Retirement Research at Boston College (CRR). Siyan Liu is a research economist at the CRR. Alicia H. Munnell is director of the CRR and the Peter F. Drucker Professor of Management Sciences at Boston College's Carroll School of Management. This research was supported by funding from the RRF Foundation for Aging. Any opinions expressed herein are those of the authors and do not necessarily represent the views of the RRF Foundation for Aging or Boston College. © 2023, Anqi Chen, Siyan Liu, and Alicia H. Munnell. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including © notice, is given to the source. About the Center for Retirement Research The mission of the Center for Retirement Research at Boston College is to produce first-class research and educational tools and forge a strong link between the academic community and decision-makers in the public and private sectors around an issue of critical importance to the nation's future. To achieve this mission, the Center conducts a wide variety of research projects, transmits new findings to a broad audience, trains new scholars, and broadens access to valuable data sources. Since its inception in 1998, the Center has established a reputation as an authoritative source of information on all major aspects of the retirement income debate. Center for Retirement Research at Boston College Haley House 140 Commonwealth Avenue Chestnut Hill, MA 02467 phone: 617-552-1762 Fax: 617-552-0191 https://crr.bc.edu Affiliated Institutions: The Brookings Institution Mathematica - Center for Studying Disability Policy Syracuse University Urban Institute Abstract The share of older Americans with debt has been on the rise in recent decades, raising concerns about their financial security in retirement. However, having debt does not always signal financial fragility, so understanding the distinctions in household debt is crucial to determining the implications of this trend. Using data from the Survey of Consumer Finances and Health and Retirement Study, this study identifies which older borrowers are at high risk and low risk of financial hardship, determines whether the growth in borrowing is driven by high- or low-risk households, and identifies different types of high-risk households. The results suggest that more than half of older borrowers are at high risk, and this group is driving the growth in debt holding. Four subgroups of high-risk borrowers stand out, each with different characteristics. Thus, no one-size-fits-all solution exists, so recognizing the diverse characteristics of high-risk borrowers is essential to developing effective policies to help them. Introduction The share of older Americans with debt has been on the rise over the last several decades. Having debt, however, does not always signal financial fragility, because debt can be used for a variety of purposes. A low-interest mortgage for buying a house is typically an investment in a growing asset, while unpaid credit card debt can snowball and lead to financial distress. Identifying these important distinctions in household debt is crucial to better understanding this growing trend. This paper addresses three key questions: 1) As more older households carry debt in retirement, what share are at "high-risk" and "low-risk" of financial hardship? 2) Is the growth in debt holding driven by the high- or low-risk households? and 3) What are the different types of high-risk households? Using data from the Survey of Consumer Finances (SCF) and Health and Retirement Study (HRS), this paper separates high-risk borrowers from low-risk ones and examines which group is behind the growth in older borrowers. It then focuses on high-risk borrowers to examine whether meaningful subgroups exist within this group. The results will help policymakers determine which types of borrowers are most vulnerable and develop tailored solutions for assisting them. The discussion proceeds as follows. The first section provides background on trends in debt holding and measures of debt burdens among older Americans. The second section describes the SCF and HRS data. The third section describes how we identify high- and low-risk households, decompose the growth in low-risk borrowers, and identify subgroups among high- risk borrowers. The fourth section presents the results, which show that more than half of older borrowers are at high risk, and that this group is driving the growth in debt holding. Four subgroups of high-risk borrowers stand out, each with different characteristics. Given the diverse situations of high-risk borrowers, the fifth section suggests some potential ways to address each group's specific needs. The final section concludes that high-risk households are driving the growth in the share of older households with debt, but a one-size-fits-all solution does not exist, so targeted interventions would be most effective. Background The share of older Americans with debt has been increasing in recent decades, growing from around 40 percent in 1989 to over 60 percent in 2019 (see Figure 1). This trend has raised concern among policymakers and researchers about financial security in retirement (Brown, Dynan, and Figinski 2019; Butrica and Karamcheva 2013, 2018; Lusardi, Mitchell, and Oggero 2018, 2020).! But the growth in older borrowers may not be all bad. On one hand, much of the growth in debt holding among older Americans is driven by rising mortgage debt (Moulton, Haurin, and Loibl 2019; Brown, Dynan, and Figinski 2019). Given the low-interest rate environment in recent decades, carrying mortgage debt into retirement may actually be financially savvy.? Indeed, prior studies have suggested that older households are refinancing their mortgage without extracting any additional equity (Bennett, Peach, and Peristiani 2001 and Spader 2021). Mortgage defaults among older households have also stayed steady (Collins, Hembre, and Urban 2020). On the other hand, unsecured forms of debt (such as credit card debt, student loan debt, and medical debt) and other secured debt (such as auto loans) have also grown (see Figure 2). These types of debt can put older households at risk of financial distress for various reasons. Credit cards - the dominant form of non-secured debt (see Figure 3) - have high interest rates, which can lead to rapid accumulation of large balances and may result in serious financial consequences, such as bankruptcy (Domowitz and Sartain 1999; and Gross and Souleles 2002).4 Older credit card borrowers may be particularly susceptible to financial hardship as many have limited understanding of interest compounding (Lusardi and Tufano 2015). As a result, older borrowers are more likely to cite credit card interest and fees as the reason for filing for bankruptcy than younger borrowers (Pottow 2012). The smaller forms of non-secured debt can also raise concerns, particularly for otherwise financially vulnerable groups.> 1 While this paper largely focuses on the demand for credit, since the late 1990s, the supply side of the credit market has also experienced changes in lender practices such as risk-based pricing that improved access to credit and contributed to rising debt (Furletti 2003, Edelberg 2006). 2 In addition to low interest rates, prior research has also documented reductions in transaction costs that make mortgage borrowing more accessible (Bennett, Peach and Peristiani 2001). 3 See Butrica and Karamcheva (2018) on credit card debt, Federal Reserve Bank of New York (2022) on student loan debt, and Lusardi, Mitchell, and Oggero (2020) on medical debt. 4 For recent data on trends in credit card debt levels and delinquency rates, see Haughwout et al. (2023). 5 Student loan debt among older households cannot be discharged in bankruptcy and defaulting can lead to reduced Social Security benefits (U.S. Government Accountability Office 2014). Moreover, student loan debt has grown more rapidly among older households with lower wealth, resulting in greater financial fragility; see Wettstein and Liu (2023) and Brown, Dynan, and Figinski (2019). Rising medical debt may lead to greater inequality in retirement security as well, since it is also more prevalent among lower-income households (Kluender et al. 2021). Finally, unsecured debt is not only detrimental to the financial security of older households but can also negatively impact their health (Mudrazija and Butrica 2021). Another dimension of financial fragility in retirement is debt burden. For example, refinancing a mortgage in retirement may be a financially savvy move for some but, for others, mortgage payments can be a substantial portion of retirement expenses. Low-wealth borrowers, for example, experienced a disproportionate rise in household mortgage debt-to-equity ratios in recent years (Lusardi, Mitchell, and Oggero 2020; Brown, Dynan, and Figinski 2019; Collins, Hembre, and Urban 2020). But, similarly, recent research examining debt burdens also shows different stories. On the one hand, younger cohorts of retirees are now carrying much higher levels of debt into retirement (Lusardi, Mitchell, and Oggero 2020; U.S. Government Accountability Office 2021). On the other hand, average monthly debt payments peaked around the Great Recession but have come down since then (Bhutta et al. 2020). In short, it is unclear to what extent the growth in the share of older households with debt is a policy concern. The goal of this study is to distinguish households that are at higher risk of putting their retirement in jeopardy from the financially savvy ones that made the most of low- interest rates. Data The analysis draws from two nationally representative surveys - the SCF and the HRS. The SCF, administered triennially since 1983, captures rich information on household balance sheets, including different forms of debt - such as credit card, medical, and student loans.® While the SCF is often considered the best publicly available survey on household debt, its sample of households in racial minority groups is limited compared to other surveys. To supplement the SCF, the analysis uses the HRS, a biennial survey of households ages 51 and older with a large range of questions on household assets and debts, to provide more information on racial/ethnic groups." However, it is worth noting that documenting household debt using the HRS has some limitations: 1) respondents are not asked about their required debt payments, limiting our ability to evaluate the burden of debt holdings; 2) it does not distinguish credit card debt from other non-secured debt prior to 2008; and 3) it is not possible to separate out medical ¢ Since the data were not consistent in survey years prior to 1989, the analysis will begin with the 1989 SCF. Brown et al. (2015) suggest that the SCF provides accurate measures of debt holding and balances among older households compared to administrative data. 7 The HRS allows for a sample of between 336 and 631 Black households and between 173 and 416 Hispanic households ages 65 or older in each wave between 1994 and 2018. In contrast, the SCF only has a combined 156 Black and Hispanic households in their 2016 survey, which has the largest share of racial minorities (see Table 1). loans.® As aresult, we focus on households ages 65 or older and arrive at a sample of 12,184 households in the SCF between 1989 and 2019 for our baseline analysis and draw from the HRS between 1994-2018, which has a sample of 98,843 households, for a larger sample on racial/ethnic groups (see Table 1). The following analysis evaluates the extent to which rising debt among older households is concerning and examines the characteristics of borrowers at high risk of financial hardship. Methodology The analysis proceeds in three steps. The first step is to define and separate the low-risk "financially-savvy" households from the high-risk ones. Our main concern is the growth in high-risk borrowers, so the second step is to confirm that low-risk households are most likely not jeopardizing their financial security. The final step examines the characteristics of high-risk households to help inform which policy solutions may help. Identifying Low- and High-Risk Households The first step of the analysis is to separate older borrowers into those at "low-risk" and "high-risk" of financial distress in retirement due to debt. We consider three factors when grouping households into their risk category - debt-to-asset ratio (DTA), debt-payment-to- income ratio (DTI), and type of debt - following lender practices and the literature.' Households with a high DTA ratio hold large amounts of debt relative to their assets. While this situation might be typical for younger households who have just taken out a long-term loan, such as a mortgage, older households with a high DTA ratio may have less flexibility to adjust their debt payments or liquidate their assets to pay off debt when faced with income shocks (Lusardi and Mitchell 2013; Lusardi, Mitchell, and Oggero 2018, 2020). For this analysis, we follow Lusardi and Mitchell and define households with DTA > 0.5 as at-risk. 8 While the amount of credit card debt is not available until the 2008 HRS, we can observe household's "other debt," which includes credit card balances, medical debts, life insurance policy loans, and loans from relatives. We consider all debt under this category high-risk since it covers both credit card and medical debt. While this approach may overstate high-risk borrowing, the share of high-risk households in the HRS is slightly lower compared to the SCEF results (see Appendix Figure Al). A possible explanation is that the HRS omits auto loans and business loans, which may lead to underestimates of debt burdens among borrowers. ? Lusardi, Mitchell, and Oggero (2020), Brown, Dynan, and Figinski (2019), and Butrica and Karamcheva (2018) examine mortgage debt and non-mortgage debt separately. One caveat of comparing debt balance to asset values is that it does not account for the interest on the loans, which can be high. Therefore, another factor that is considered high risk is a high DTL!® A high DTI means households' required principal and interest payment on their debt takes up a large portion of their income, indicating stress when making payments. Lenders also typically require DTI to be below 36 percent to 45 percent, so households with high DTI will also have limited access to credit when faced with income or expenditure shocks.!! For this analysis, we define households with DTI > 0.4, the midpoint of what lenders typically use, as high-risk. Households with any revolving credit card debt, the dominant form of unsecured debt, are also included in the high-risk group.'? It is important to include households with any revolving credit card debt in the high-risk group since many of these borrowers could experience bad outcomes, even though they would not be captured by the other debt measures. Specifically, since minimum payments on credit cards are typically very small, borrowers would rarely exceed the debt payment-to-income threshold. And a borrower would have to carry extremely high levels of credit card debt to exceed the debt-to-asset threshold. Even so, many credit card borrowers do carry a substantial level of debt. '3 Households that satisfy any of the three factors above -a DTA >0.5, a DTI >0.4, or any credit card or other unsecured debt - are defined as high-risk.!* For a summary of how we distinguish high-risk older debt holders from low-risk ones, see Table 2. Once we have separated high-risk households from low-risk ones, we can evaluate to what extent the growth in older borrowers is driven by low-risk "financially-savvy" households versus high-risk households. 10 Dynan and Kohn (2007), Johnson and Geng (2010), and Bricker et al. (2014). 11 In 2014, the Consumer Financial Protection Bureau (CFPB) introduced the qualified mortgage (QM) rule that requires lenders to make "a reasonable, good faith determination" of a consumer's ability to repay a mortgage loan, with having a DTI lower than 43 percent as the main criterion (Kaul, Goodman, and Zhu 2019). 12 Households with other forms of unsecured debt, such as medical debt, student loans, and personal loans are also included, but they represent a small portion of the population. Over 90 percent of households with unsecured debt hold credit card debt. 13 Among older households with revolving credit card balances, the median household's credit card debt equals 70 percent of their monthly income. Even at the 25% and 10® percentile, revolving credit card balances equal 20 percent and 6 percent of income, respectively. 14 While personal loans are a form of unsecured debt, households with personal loans are not automatically considered high-risk. This is because personal loans can be for a variety of purposes and neither datasets contains enough information on personal loans to classify them. Only about 3 percent of households have personal loans but no other type of unsecured debt. Decomposing the Growth in Low-Risk Households Our goal is to understand the growth in high-risk households. But to ensure that we have appropriately identified low-risk households, we explore two reasons - low interest rates and increased diversity - that may contribute to the growth in older borrowers but do not signify more households in financial trouble. First, rapidly declining interest rates since the early 2000s may have enticed some "financially-savvy" borrowers to take on debt and it would not have jeopardized their retirement security. To decompose the share of households that were enticed by the low-interest rate environment, the analysis identifies households in the 2001-2019 SCF that obtained or refinanced a mortgage in the prior three years and have no other forms of debt. These households likely would not have carried any debt in retirement if not for the low borrowing costs and should account for the vast majority of the growth in low-risk households.' Second, increased diversity may affect the growth of low-risk borrowers in retirement. Households from racial/ethnic minority groups are more likely to hold debt in retirement. As the population of retirees becomes more diverse, a greater share of older houscholds will be debt holders. To decompose the share of the growth in low-risk households due to increased diversity, we calculate the share of debt holding among White and non-White households for each year and reweight them using the racial composition in 2001 to obtain the counterfactual debt-holding rate without increased diversity. Characteristics of High-Risk Borrowers Once we have ensured that low-risk borrowers are correctly categorized, our attention shifts to the high-risk borrowers. The growth in older households with debt that is driven by high-risk borrowers is a policy concern. But in order to develop appropriate policy solutions, we have to understand the characteristics and subgroups of high-risk households. Latent Class Analysis (LCA) is one method that can be used to determine whether some identifiable subgroups exist within a population by finding relationships between observed 15 Households may also have taken up additional borrowing against their home equity by refinancing a mortgage. We still consider households who recently refinanced a mortgage "financially savvy" because most older households refinancing their mortgage did not do a cash-out refinance (Spader 2021). categorical variables as a function of some unobserved grouping.'® The categorical variables included to identify subgroups are: e household wealth terciles;!? e homeownership status; e DTA ratio >0.5; e DTI ratio > 0.4 (SCF only); e revolving credit card or other unsecured debt; e struggling to pay for essentials; and e recent shocks that could increase the need for borrowing. Not being able to afford essentials is defined as receiving SNAP, TANF, SSI, or Medicaid benefits.!® Three types of shocks - housing, medical, and marital - are included. A housing shock occurs when the house's value falls below its initial purchase price or when a decline in the house price pushes the household's DTA > 0.5, reducing its access to assets.!® Medical shocks are defined as taking out a medical loan in the last two years or having unusually low income due to an illness.?® And marital shock is a recent divorce or widowhood, which can substantially reduce a household's income. 2! We focus on recent borrowers from the 2016-2019 SCF and the 2016-2018 HRS to provide an up-to-date description of older borrowers at high risk of experiencing financial hardships. 16 Conditional on an assumed number of classes, LCA produces two sets of estimates: 1) the share of the population within each class; and 2) the conditional probabilities of having a given value for each observed variable within each class. These parameters are estimated by maximum likelihood estimation, where the inputs are the observed prob- abilities. The conditional probabilities have a special interpretation within LCA since they represent a measure of association between the class and the observed characteristic. 17 Wealth terciles are based on wealth adjusted for household size, dividing wealth by two for married households. Homeowner status includes ownership of the primary residence and second homes (if any). 18 We also include falling behind on mortgage payments or struggling to pay for food as a sign of not being able to afford essentials in the HRS. 19 The HRS also defines a foreclosure as a housing shock. 20 We define out-of-pocket medical spending that is larger than 10 percent of household income as a medical shock in the HRS, since this survey does not contain information on medical loans. 21 The model also accounts for whether a household has non-collateralized debt and whether it is overleveraged (DTA>0.5). A medical shock using the SCF is defined as the household head or spouse having unusually low income due to poor health, taking payday loans to pay medical pills, or taking personal loans for medical expenses in the last three years. Results and Discussion Distinguishing older households that are at low- and high-risk of jeopardizing their retirement is the first step in understanding how much of the growth in older households with debt is a concern. Using our definitions outlined in Table 1 reveals that while both groups have grown over time, overall growth is driven by the high-risk households (see Figure 4). The share of low-risk households grew 7 percentage points, from 13 percent to 20 percent between 1989 and 2019. In comparison, high-risk households grew by 18 percentage points, from 25 percent to 43 percent over the same period. % A comparison of high-risk and low-risk households is shown in Table 3. High-risk households are disadvantaged in a number of ways. Compared to low-risk households, high-risk borrowers are more likely to be Black or Hispanic and less likely to be college-educated and married. Despite having similar ages and propensity to work at older ages, high-risk households have lower income, wealth, and homeownership levels than low-risk households. As a result, high-risk borrowers carry more burdensome debt with higher DTA and DTI, even though debt balances are lower than that of low-risk borrowers. Should We Be Concerned About Low-Risk Borrowers? While the main concern is the growth in high-risk households that is driving the share of retirees with debt, it is worth determining whether the growth in low-risk households is also troubling. The theory is that the growth in low-risk households is mainly driven by two factors - low interest rates and increasing diversity - that may contribute to the growth in low-risk households but do not signify financial trouble among these households. Our analysis finds that about half of the growth in low-risk households is due to financially savvy borrowers who took advantage of low-interest rates to obtain or refinance a mortgage loan (see Figure 5).2 This estimate is a lower bound because we only count households that refinanced their mortgage in the prior three years and do not have any other forms of debt as financially savvy. For example, 22 The share of older borrowers is also lower in the HRS. Reassuringly, the share of high-risk borrowers among households with debt in the HRS (60-69 percent) is similar to that in the SCF (65-69 percent) across waves. Results using the HRS show similar trends of high-risk borrowing, albeit at slightly lower levels (see Appendix Figure Al). 23 Low interest rates enticed some households to make investments in housing, which may not necessarily harm their retirement security. These financially savvy borrowers may have been taking advantage of the rapidly declining borrowing costs and may not have carried debt in retirement had interest rates stayed at early 1990s levels. To identify the impact of low interest rates on growth in low-risk households, the analysis classifies households who obtained or refinanced a mortgage in the prior three years and have no other forms of debt as "financially savvy." other older borrowers who refinanced more than three years ago or have a small car loan would not be included but could also be considered "financially savvy." Not surprisingly, virtually none of the growth in high-risk households involves financially savvy households, confirming our definitions.?* Interestingly, increased racial/ethnic diversity among retirees played a trivial role in the growth of both low-risk and high-risk borrowers. Who Are the High-Risk Borrowers? Having confirmed that the growth in low-risk borrowers is largely benign, the next step is to take a closer look at the characteristics of high-risk borrowers. Two characteristics that tend to be associated with the likelihood of being financially vulnerable - race/ethnicity and wealth - show the expected patterns. Non-White households are more likely to fall into the high-risk borrower group, though the gap with White households has been narrowing over time (see Figure 6.). Similarly, households with less wealth are also more likely to be high-risk borrowers; here, though, the gap with high-wealth households has been widening (see Figure 7). But high- risk households have also grown in other socioeconomic groups too, suggesting that several subgroups of older borrowers exist, each with different reasons for accumulating debt. To help identify the different subgroups, as noted above, the analysis turns to an LCA model that groups households based on wealth, medical and financial shocks, homeownership, DTA, DTI (in the SCF), credit card and other unsecured debt, and whether they might struggle with essential expenses. A 6-class model in the SCF and a 4-class model in the HRS provide the best balance between fit statistics and interpretability.2> While the SCF produced two additional classes, they represented a further breakdown of two of the subgroups in the HRS, likely because the SCF has more information on household debt and assets and contains information on DTL. For expositional purposes, we combined the six subgroups from the SCF into four.26 The characteristics of the subgroups are largely similar across the two datasets, as shown in Tables 4 and 5. 24 High-risk households may have also been enticed by the cheap borrowing costs during this period and taken on additional borrowing, but the vast majority of this group already had other forms of debt (see Figures 2 and 3). 25 For AIC, BIC, and Adjusted BIC statistics, see Tables Al and A2 in the Appendix. 26 Four groups of high-risk borrowers stand out after we combine two "financially constrained" groups that largely vary by marital status and two groups with "too much mortgage" that vary by levels of wealth. The largest group, representing 33 percent of high-risk borrowers in the SCF and 40 percent of borrowers in the HRS, is "financially constrained" households. This group is less likely to be married and have a college degree, and more likely to be Black or Hispanic. They have low levels of income and wealth, with the median household having less than $45,000 in wealth. The vast majority of this group has credit card debt, and one out of ten has medical debt. About a third of these financially constrained households are also struggling with essentials. Their debt holding, combined with demographic and other financial characteristics, suggests that this group may be borrowing just to get by. Another notable subgroup is the "credit card borrowers," representing 26 percent of high- risk households in the SCF and 32 percent in the HRS. This group includes middle-wealth and income households with no obvious need to borrow. Median household wealth is around $300,000, almost all are homeowners, and their DTA and DTI ratios are relatively low. But the vast majority of this group has credit card debt. The third subgroup, the smallest of the four groups, has "too much house." This subgroup consists of low/middle-wealth households whose house has become a large liability and constraint on their income in retirement. The median required debt payment equals almost half of their monthly income. While they have relatively little other debt, their housing debt equals 60 to 80 percent of their total assets.?'" One explanation is that, relative to other high-risk households, this group is more likely to have experienced a housing shock. In fact, about one- tenth have homes that are worth less than their initial purchase price or have gone through foreclosures. Similar to financially constrained households, the homeowners burdened by "too much house" are also disproportionately Black and Hispanic, and are less likely to be college- educated. Lastly, about one-fifth of the high-risk borrowers are "wealthy spenders." Households in this group have high wealth with more than a million in assets and about $80,000 in income. About 40 percent also own a second home. Despite having ample resources in retirement, many have revolving credit card debt and their monthly debt payments represent about a quarter of their income. Similar to the credit card borrowers, it is not clear why the wealthy spenders need to revolve credit card debt in retirement. 27 Credit card debt holding for the "too much house" group differs across the HRS and SCF. In the HRS, this group does not hold any credit card debt, but it does in the SCF. 10 How Can We Help High-Risk Borrowers? What can be done to reduce the financial vulnerability of high-risk borrowers? Clearly, given their diverse characteristics, no one-size-fits-all solution exists. Thus, the best approach would be for financial counselors, consumer advocates, and/or policymakers to develop tailored solutions for the specific needs of the four groups and to target assistance to those most vulnerable to financial hardship. While a detailed strategy is beyond the scope of this study, a few thoughts may help. The first and largest group of high-risk borrowers is characterized by their "financial constraints." For these households, debt counseling and consolidation may help, but many struggle to meet basic needs, so they need more resources. One approach is strengthening means-tested programs, like Supplemental Security Income (SSI). Since SSI currently has extremely low earnings and asset thresholds, many low-income seniors are not eligible, and the SSI benefit amount is less than the poverty level. Improving a program like SSI could provide a lifeline to reduce their burden. Another approach is to enhance Social Security's Special Minimum Benefit, which is meant to provide an adequate level of benefits for lifetime low- earners. However, the value of this benefit has eroded, and the number receiving the minimum benefit declined from around 200,000 in the 1990s to about 25,000 in 2022.28 The second group - "credit card borrowers" - look very different from the "financially constrained" households and have no obvious need to accumulate debt, particularly costly credit card debt. Households in this group may not understand the implications of revolving credit card debt, how the high interest rates affect unpaid balances, and what the minimum payment means.?' Some in this group, who do not have an emergency fund, may be using their credit card to help smooth expense shocks. These households could benefit from traditional financial counseling programs to curtail the use of high-interest-rate debt and to encourage precautionary saving for unexpected expenses. Legislation requiring credit card issuers to provide better information to consumers could also help. 3° For example, when consumers navigate to a credit 28 Fewer new beneficiaries are receiving the price-indexed special minimum benefit because wage growth typically exceeds price growth; thus, their wage-indexed regular benefit is usually higher. 29 Prior research has shown that about 30 percent of credit card owners pay roughly the minimum amount each month. Interestingly, many continue to pay around the minimum even as their minimum required payment increases, suggesting that anchoring - rather than liquidity constraints - may be driving repayment behavior (Keys and Wang 2019). 30 The CARD Act of 2009 tried to address this problem by requiring credit card statements to include a table that outlines the monthly amount required to pay off the current credit card balance in three years. However, this 11 card payment screen, the first option listed could be the amount to pay off the full balance. Another option is for a caution to appear if consumers opt to pay an amount that is less than the balance they owe. The third group has "too much house" and would best be served by programs that help these overstretched homeowners reduce their housing burden. Options could include refinancing their mortgage to reduce their monthly payments or downsizing; albeit this step could be challenging if their housing equity has declined, for some below their purchase price. Policymakers and financial counselors could also encourage near-retirees to prioritize paying off their mortgage before retirement to avoid having mortgage payments overwhelm expenses in retirement. Finally, the fourth group of "wealthy spenders" needs to get a handle on their discretionary spending to help rein in their borrowing. Many in this group have a second home, so selling it is one way to manage their debt burden. While these households may be a lower priority for policymakers as they are the least vulnerable, general financial counseling could also help them change their habits so that they do not consume beyond their means. Conclusion A rapidly growing share of U.S. households carry debt in retirement, raising concerns about their financial security. While carrying debt in retirement may be a financially savvy move for some, the rise in debt holding at older ages is driven by households who are at "high- risk" of financial trouble. But the characteristics of these high-risk borrowers vary a lot. Some groups seem to have the resources to manage their debt, so financial counseling on the risks of excessive credit card or mortgage debt or improved disclosure requirements for lenders may help them get a handle on their borrowing. Other groups have very few resources to work with and struggle with essential expenses. For these borrowers, financial education can only provide limited help, and they need more resources, perhaps through broader access to means-tested programs like SSI. The key takeaway is the recognition that no one-size-fits-all solution exists, so information is not required for online or mobile payments, and roughly 80 percent of accountholders use the online portal and over 60 percent use mobile apps. The information presented in the web and mobile payment methods can vary by card issuer. Some will present the minimum payment first, while others will present the statement of current balance first. 12 understanding the diverse characteristics of high-risk borrowers is essential to developing effective policies to help older households struggling with debt. 13 References Bennett, Paul, Richard Peach, and Stavros Peristiani. 2001. "Structural Change in the Mortgage Market and the Propensity to Refinance." Journal of Money, Credit and Banking 33: 955- 75. Bhutta, Neil, Jesse Bricker, Andrew C. Chang, Lisa J. Dettling, Sarena Goodman, Joanne W. Hsu, Kevin B. Moore, Sarah Reber, Alice Henriques Volz, and Richard Windle. 2020. "Changes in US Family Finances from 2016 to 2019: Evidence from the Survey of Consumer Finances." Federal Reserve Bulletin 106(5). Bricker, Jesse, Lisa J. Dettling, Alice Henriques, Joanne W. Hsu, Kevin B. Moore, John Sabelhaus, Jeffrey Thompson, and Richard A. Windle. 2014. "Changes in U.S. Family Finances from 2010 to 2013: Evidence from the Survey of Consumer Finances." Federal Reserve Bulletin 100: 1-41. Brown, Jason, Karen Dynan, and Theodore Figinski. 2019. "The Risk of Financial Hardship in Retirement: A Cohort Analysis." WP2019-10. Philadelphia, PA: Pension Research Council. Brown, Meta, Andrew Haughwout, Donghoon Lee, and Wilbert van der Klaauw. 2015. "Do We Know What We Owe? Consumer Debt as Reported by Borrowers and Lenders." FRBNY Economic Policy Review 21(1): 19-44. Butrica, Barbara A. and Nadia S. Karamcheva. 2013. "Does Household Debt Influence the Labor Supply and Benefit Claiming Decisions of Older Americans?" Working Paper 2013-22. Chestnut Hill, MA: Center for Retirement Research at Boston College. Butrica, Barbara A. and Nadia S. Karamcheva. 2018. "In Debt and Approaching Retirement: Claim Social Security or Work Longer?"" AEA Papers and Proceedings 108: 401-06. Collins, J. Michael, Erik Hembre, and Carly Urban. 2020. "Exploring the Rise of Mortgage Borrowing Among Older Americans." Regional Science and Urban Economics 83(103524). Domowitz, Ian and Robert L. Sartain. 1999. "Determinants of the Consumer Bankruptcy Decision." The Journal of Finance 54(1): 403-420. Dynan, Karen E. and Donald L. Kohn. 2007. "The Rise in U.S. Household Indebtedness: Causes and Consequences." Finance and Economics Discussion Series Working Paper 37. Washington, DC: Federal Reserve Board, Division of Research & Statistics and Monetary Affairs. Edelberg, Wendy. 2006. "Risk-Based Pricing of Interest Rates for Consumer Loans." Journal of Monetary Economics 53: 2283-98. 14 Federal Reserve Bank of New York. 2022. "Quarterly Report on Household Debt and Credit: 2021: Q4." New York, NY. Furletti, Mark J. 2003. "Credit Card Pricing Developments and Their Disclosure." Federal Reserve Bank of Philadelphia Payment Cards Center Discussion Paper No. 03-02. Philadelphia, PA. Gross, David B. and Nicholas S. Souleles. 2002. "An Empirical Analysis of Personal Bankruptcy and Delinquency." The Review of Financial Studies 15(1): 319-347. Haughwout, Andrew, Donghoon Lee, Daniel Mangrum, Joelle Scally, and Wilbert van der Klaauw. 2023. "Credit Card Markets Head Back to Normal After Pandemic Pause." Liberty Street Economics Blog Post. New York, NY: Federal Reserve Bank of New York. Johnson, Kathleen W. and Geng Li. 2010. "The Debt-Payment-to-Income Ratio as an Indicator of Borrowing Constraints: Evidence from Two Household Surveys." Journal of Money, Credit and Banking 42(7): 1373-1390. Kaul, Karan, Laurie Goodman, and Jun Zhu. 2019. "Comment Letter to the Consumer Financial Protection Bureau on the Qualified Mortgage Rule." Washington, DC: Urban Institute. Keys, Benjamin J. and Jialan Wang. 2019. "Minimum Payments and Debt Paydown in Consumer Credit Cards." Journal of Financial Economics 131(3): 528-548. Kluender, Raymond, Neale Mahoney, Francis Wong, and Wesley Yin. 2021. "Medical Debt in the US, 2009-2020." JAMA 326(3): 250-256. Lusardi, Annamaria and Peter Tufano. 2015. "Debt Literacy, Financial Experiences, and Overindebtedness." Journal of Pension Economics & Finance 14(4): 332-368. Lusardi, Annamaria, Olivia S. Mitchell, and Noemi Oggero. 2018. "The Changing Face of Debt and Financial Fragility at Older Ages." AEA Papers and Proceedings 108: 407-11. Lusardi, Annamaria, Olivia S. Mitchell, and Noemi Oggero. 2020. "Debt and Financial Vulnerability on The Verge of Retirement." Journal of Money, Credit and Banking 52(5): 1005-1034. Moulton, Stephanie, Donald Haurin, and Caezilia Loibl. 2019. "Debt Stress and Mortgage Borrowing in Older Age: Implications for Economic Security in Retirement." Economic Inquiry 54(1): 201-214. Mudrazija, Stipica and Barbara A. Butrica, 2021. "How Does Debt Shape Health Outcomes for Older Americans?" Working Paper 2021-17. Chestnut Hill, MA: Center for Retirement Research at Boston College. 15 Pottow, John A. E. 2012. "The Rise in Elder Bankruptcy Filings and Failure of U.S. Bankruptcy Law." Elder Law Journal 19: 119-57. Spader, Jonathan. 2021. "Can Changing Demographics or Refinancing Behaviors Explain the Rising Levels of Housing Debt Among Older Americans?" Housing Policy Debate 31 (2): 290-305. University of Michigan. Health and Retirement Study, 1994-2018. Ann Arbor, MI. U.S. Board of Governors of the Federal Reserve System. Survey of Consumer Finances, 1989- 2019. Washington, DC. U.S. Government Accountability Office. 2014. "Older Americans' Inability to Repay Student Loans May Affect Financial Security of a Small Percentage of Retirees." Washington, DC. U.S. Government Accountability Office. 2021. "Debt Increased for Older Americans over Time, but the Implications Vary by Debt Type." Washington, DC. Wettstein, Gal and Siyan Liu. 2023. "How Do Unpaid Student Loans Impact Social Security Benefits?" Issue in Brief 23-1. Chestnut Hill, MA: Center for Retirement Research at Boston College. 16 Table 1. Number of Households Ages 65+ in the SCF and HRS, by Race, 1989-2019 Count of households ages 65+ with debt Count of households Total Black Hispanic ages 65+ Panel A. SCF sample sizes by wave 1989 313 40 7 747 1992 422 41 9 917 1995 402 28 5 949 1998 390 30 9 907 2001 410 29 8 922 2004 441 31 14 891 2007 497 29 14 980 2010 684 63 31 1,306 2013 748 79 29 1,386 2016 927 112 44 1,586 2019 938 93 35 1,593 Panel B. HRS sample sizes by wave 1994 1,736 352 174 6,750 1996 1,670 336 173 6,105 1998 2,388 440 234 7,812 2000 2,561 468 272 7,833 2002 2,629 503 256 7,955 2004 2,950 554 294 8,088 2006 3,264 591 341 8,357 2008 3,310 613 328 8,241 2010 3,292 580 331 7,929 2012 3,208 563 344 7,761 2014 3,171 574 376 7,604 2016 3,186 631 401 7,517 2018 2,919 617 416 6,891 Sources: Authors' calculations from the University of Michigan, Health and Retirement Study (HRS) (1994-2018) and U.S. Board of Governors of the Federal Reserve System, Survey of Consumer Finances (SCF) (1989-2019). 17 Table 2. Definitions of Low- and High-Risk Borrowers Low-risk borrowers High-risk borrowers Have credit card or other non-secured debt (credit card, medical, and student loans) Only have secured debt (mortgage, auto loans, and other residential) AND OR Debt payment-to-income < 0.4 Debt payment-to-income >= 0.4 AND OR Debt-to-assets < 0.5 Debt -to-assets >= 0.5 (overleveraged) Source: Authors' definition. Table 3. Characteristics of High-Risk Borrowers Ages 65+, 1989-2019 High-risk borrowers Low-risk borrowers N 3,372 2,800 Panel A. Demographics Age of household head 72.1 72.5 Married 53% 66% Black 16 6 Hispanic 5 2 College degree or above 23 38 Panel B. Household financials Working for pay 18% 18% Median income $38,849 $57,759 Median asset values 207,899 426,408 Homeowner 81% 92% Panel C. Debt holding Median debt balance $21,200 $27,849 Median debt-to-asset ratio 16% 7% Debt-to-asset ratio >=50% 24 0 Median debt payments-to-income ratio 19 12 Debt payments-to-income ratio >=40% 24 0 Has unsecured debt 88 0 Median unsecured debt balance $2,000 - Has credit card debt 83% 0 Median credit card balance 1,850 - Source: Authors' calculations from the SCF (1989-2019). 18 Table 4. Four Types of High-Risk Households in the SCF, 2016-2018 @ 2 3) @ Gr Financially Creditcard Too much Wealthy oup . constrained borrowers house spenders Share 32.7% 25.9% 18.5% 22.9% Panel A. Demographics Age 72.2 72.9 72.7 71.9 Married 43% 58% 56% 51% College or more 16 30 26 58 Black 24 16 24 8 Hispanic 9 4 5 2 Panel B. Wealth and income Lowest wealth tercile 100% 0% 37% 0% Middle wealth tercile 0 100 63 0 Highest wealth tercile 0 0 0 100 Household wealth $44,758 $302,712 $205,660 $1,010,332 Household income 31,742 53,200 38,772 80,000 Homeowner 43% 96% 99% 97% Has second home 3 18 8 36 Panel C. Debt holding Debt-to-asset ratio >= 50% 29% 0% 80% 10% Debt payments-to-income ratio >= 40% 13 16 45 26 Housing debt-to-asset ratio 7 17 60 25 Has credit card debt 86 96 67 83 Has medical debt 11 3 3 2 Has student loans 7 4 8 10 Panel D. Spending needs Struggling w/ essentials 30% 10% 13% 6% Housing shock 3 4 9 6 Medical shock 8 2 1 2 Marital shock 3 0 2 2 N 319 237 160 366 Notes: Results are based on a six-class model - the six classes are the low-wealth financially constrained; the low- wealth financially constrained who are overwhelmingly single female and medical debt holders; credit card borrowers; low-wealth households with too much house; middle-wealth households with too much house; and wealthy spenders. For simplicity, we show summary statistics after combining the two financially constrained groups into one and the two groups with too much house into one. We report in this table the median values of household wealth, house value, non-housing net wealth, income, debt payments-to-income ratio, and debt balance relative to income in each group. We report the average housing DTA for each group after limiting the sample to the bottom 99 percent of the housing DTA distribution. Source: Authors' calculations from the SCF (2016-2019). 19 Table 5. Four Types of High-Risk Households in the HRS, 2016-2018 1) (2) 3) ") Gr Financially Creditcard Too much Wealthy oup . constrained borrowers house spenders Share 40.2% 32.1% 7.6% 20.1% Panel A. Demographics Age 75.2 74.8 73.7 74.6 Married 44% 60% 44% 58% College or more 13 27 18 53 Black 14 11 18 7 Hispanic 12 9 16 7 Panel B. Wealth and income Lowest wealth tercile 100% 0% 36% 0% Middle wealth tercile 0 100 59 0 Highest wealth tercile 0 0 6 100 Household wealth $41,742 $295,250 $169,363 $1,069,010 Household income 28,594 50,992 41,618 77,310 Homeowner 49% 96% 100% 97% Has second home 3 16 5 37 Panel C. Debt holding Debt-to-asset ratio >= 50% 43% 18% 100% 3% Housing debt-to-asset ratio 12 27 79 18 Has credit card debt 75 78 0 55 Other non-secured debt A . 3 2 - 1 balance relative to income Panel D. Spending needs Struggling w/ essentials 35% 13% 24% 6% Housing shock 3 1 9 0 Medical shock 16 13 10 12 Marital shock 4 3 6 4 N 1,187 884 222 493 Notes: We report in this table the median values of household wealth, house value, non-housing net wealth, income, and debt balance relative to income in each group. We report the average housing DTA for each group after limiting the sample to the bottom 99 percent of the housing DTA distribution. Source: Authors' calculations from the HRS (2016-2018). 20 Figure 1. Percentage of All Households Ages 65+with Debt, 1989-2019 100% 75% 50% / -~ 25% O% T T T T T T T T T T 1 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016 2019 Source: Authors' calculations from the SCF (1989-2019). Figure 2. Percentage of All Households Ages 65+ with Debt by Type, 1989-2019 100% Mortgage Other-secured debt = = Non-secured debt 75% 50% g - /~§\ ,----_--- - 0% [ [ 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016 2019 Notes: Households with more than one type of debt are included in each relevant category of debt. The "other secured debt" category includes car loans and other types of non-mortgage debt that involve collateral for the loan. Source: Authors' calculations from the SCF (1989-2019). 21 Figure 3. Percentage of Households Ages 65+ with Non-Secured Debt by Type, 2019 100% 85% 75% 50% 25% 9% 2% 4% 77 0% | -0 ' ' / Any credit card Medical debt Student loans Other non- debt only only secured debt only Note: About 1 percent of households have multiple types of non-credit-card, non-secured debt. Source: Authors' calculations from the SCF (1989-2019). Figure 4. Low-Risk and High-Risk Households Ages 65+, 1989-2019 100% W Low-risk m High-risk 75% - 50% - 25% 0% 1989 1995 2001 2007 2013 2019 Source: Authors' calculations from the SCF (1989-2019). 22 Figure 5. Reasons for Growth in High- and Low-Risk Households Ages 65+, 1989 and 2019 30% Growth in high-risk households: 18.2% 20% 10% - 0% - 0.8% 0.9% Growth in low-risk households: 6.7% -0.6% B Other OFinancially Savvy EIncreased Diversity -10% Source: Authors' calculations from the SCF (1989-2019). 23 Figure 6. Percentage of Borrowers Ages 65+ Who Are High Risk, by Race, 1994-2018 50% Non-White = = White 25% 0% 1994 1998 2002 2006 2010 2014 2018 Source: Authors' calculations from the HRS (1994-2018). Figure 7. Percentage of Borrowers Ages 65+ Who Are High Risk, by Wealth Tercile, 1994-2018 50% Lowest = = Middle = Highest 25% / 7 0% 1994 1998 2002 2006 2010 2014 2018 Source: Authors' calculations from the HRS (1994-2018). 24 Appendix Figure Al. High-Risk Borrowers Ages 65+ in the SCF and HRS, 1989-2019 60% 40% ) 20% o~ -HRS --SCF O% T T T T T T T T T T 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016 2019 Sources: Authors' calculations from the HRS (1994-2018) and SCF (1989-2019). Table Al. Latent Class Fit Statistics from the SCF 3 Classes 4 Classes 5 Classes 6 Classes AIC 843.7 797.6 805.5 727.7 BIC 1048.2 1071.8 1149.5 1141.6 Adjusted BIC 917.9 897.1 930.4 878.0 Source: Authors' calculations from the SCF (2016-2019). Table A2. Latent Class Fit Statistics from the HRS 3 Classes 4 Classes 5 Classes AIC 1923.8 1923.8 1514.9 BIC 2226.4 2226.4 1894.5 Adjusted BIC 2064.3 2064.3 1691.2 Source: Authors' calculations from the HRS (2016-2018). 25 RECENT WORKING PAPERS FROM THE CENTER FOR RETIREMENT RESEARCH AT BOSTON COLLEGE Wills, Wealth, and Race Jean-Pierre Aubry, Alicia H. Munnell, and Gal Wettstein, August 2023 How Can Social Security Children's Benefits Help Grandparents Raise Grandchildren? Siyan Liu and Laura D. Quinby, June 2023 Forward-Looking Labor Supply Responses to Changes in Pension Wealth: Evidence from Germany Elisabeth Artmann, Nicola Fuchs-Schiindeln, and Giulia Giupponi, June 2023 What Factors Explain the Drop in Disability Insurance Rolls from 2015 to 2019? Siyan Liu and Laura D. Quinby, May 2023 Why Do Late Boomers Have So Little Wealth and How Will Early Gen-Xers Fare? Angi Chen, Alicia H. Munnell, and Laura D. Quinby, May 2023 The Importance of Schools in Driving Children's Applications for Disability Benefits Michael Levere, Jeffrey Hemmeter, and David Wittenburg, March 2023 Does the Drop in Child SSI Applications and Awards During COVID Vary by Locality? Michael Levere, Jeffrey Hemmeter, and David Wittenburg, March 2023 What Factors Drive Racial Disparities in Housing Wealth Accumulation? Siyan Liu and Laura D. Quinby, March 2023 What Matters for Annuity Demand: Objective Life Expectancy or Subjective Survival Pessimism? Karolos Arapakis and Gal Wettstein, January 2023 How Will COVID Affect the Completed Fertility Rate? Angi Chen, Nilufer Gok, and Alicia H. Munnell, January 2023 County-Level Drivers of Disability Benefit Claims in Times of COVID-19 R. Vincent Pohl and David R. Mann, December 2022 Health, Disability, and the Evolving Nature of Work Barbara A. Butrica and Stipica Mudrazija, December 2022 All working papers are available on the Center for Retirement Research website (https://crr.be.edu) and can be requested by e-mail (crr@bc.edu) or phone (617-552-1762). 26