PARENTAL OBLIGATIONS IN BANKRUPTCY
by
Belisa Pang* and Katherine Fang**
Parenthood is difficult—especially for women participating in the chapter 13 bankruptcy process. Using a novel dataset comprising 102,952 bankruptcy cases across sixty-four districts, we find that debtors with dependents are not only more likely to choose chapter 13 over the quicker and simpler chapter 7 bankruptcy but are also eight percentage points more likely to have a bankruptcy court dismiss their cases without the discharge of their debt within three years, compared to other individuals in chapter 13 bankruptcy. This effect is persistent even after we control for gender, marital status, financial information (assets, liabilities, income, expenses, etc.), and district- and year-fixed effects. The effect of raising children is particularly acute for women in chapter 13, whose cases are dismissed within three years at a rate that is 10.2 percentage points higher if they have dependents. Further, although joint filers are about 6.7 percentage points less likely to have their cases dismissed, the overall effect of marriage is not statistically significant and does not reduce the negative effect of parental obligations. Overall, this article shows that currently, bankruptcy law inadequately protects children from the impact of bankruptcy. We propose legal changes that might alleviate these disparities.
I. Introduction
Between 2010 and 2019, 9.6 million Americans filed for bankruptcy protection.[1] Two million of those debtors ultimately saw their cases terminated without the discharge of their debt.[2] In this article, we explore one explanation for differential discharge rates across otherwise similarly situated debtors.
Using a novel dataset containing detailed information for 102,952 bankruptcy cases across sixty-four districts, we find that having a dependent reduces the discharge rate of chapter 13 debtors by approximately eight percentage points.[3] This effect is particularly acute among women, whose chapter 13 cases are over ten percentage points more likely to be dismissed when they have a dependent.[4] In contrast, dependents impact male debtors’ bankruptcy outcomes much less.[5]
This gap between bankruptcy debtors with dependents and those without dependents persists despite Congress’s efforts to safeguard children and their caretakers in such proceedings. For example, some childcare-related expenses are routinely carved out from a debtor’s income in chapter 13 bankruptcy proceedings, reflecting legislators’ intent to “plac[e] women and children at the highest level of protection” in bankruptcy.[6] Arguably for similar reasons, some exemptions available to a debtor (which may be based on federal or state law under § 522 of the Bankruptcy Code) and some deductions used to calculate the bankruptcy plan payment under § 1325 of the Bankruptcy Code are based on a debtor’s household’s size. In addition, during the COVID-19 pandemic, the American Rescue Plan Act of 2021 expanded child tax credits and allowed for a 50% advance payment of the estimated child tax credit for the 2021 tax year.[7] Both benefits were excluded from the calculation of projected payment to creditors.[8] In addition, these benefits did not affect parents’ eligibility for chapter 7 bankruptcy as an alternative to chapter 13.[9] Further, for parents whose former spouses declare bankruptcy, any obligation of the debtor (the former spouse) to pay child support is usually considered a nondischargeable debt in the bankruptcy case.[10] Taken together, these legislative initiatives reduce the financial burden on individuals with dependents when they or their former spouses file for bankruptcy. In all these ways, Congress protects children when their parents experience financial distress.
But our study shows that there remain significant gaps in existing law—and opportunities for Congress to better assist families when a parent is undergoing bankruptcy. As our Table 1 shows, the baseline discharge rate across cases is 38%. The fact that debtors with dependents are eight percentage points more likely to have their chapter 13 cases dismissed suggests that currently, bankruptcy law significantly underestimates the burden of childcare expenses.
For clarification, in this article, we use the term “parental obligations” to refer to both the financial and non-financial responsibilities associated with having a dependent. We generally use this phrase interchangeably with “having a dependent.” We deliberately refer to “obligations,” rather than to “costs” or “expenses,” to emphasize that parenthood entails not only financial costs that are addressed at least partially through deductions and exemptions but also non-financial duties that the bankruptcy system often overlooks. We strive to identify the impact of such non-financial burdens through the study of differential discharge rates, which reveal potential gaps in the financial relief available in the bankruptcy system.
This article proceeds as follows. Part II reviews the existing and relevant literature regarding the impact of bankruptcy on individual consumer debtors. Part III provides background information on consumer bankruptcy. In particular, this Part first describes how a chapter 13 payment plan is set, and then explains the protections available to parents in bankruptcy cases. Part IV introduces the dataset, and Part V presents the main results. Part VI analyzes the effects of gender, marriage, and parental obligations on outcomes in chapter 13 cases. Part VII studies the effect of parental obligations on the debtor’s choice of bankruptcy chapter. The article then considers potential policy implications in Part VIII and concludes in Part IX.
II. Literature Review
Prior legal publications discuss the higher bankruptcy filing rate of women with dependents.[11] However, there is a dearth of empirical scholarship concerning exactly why these debtors might not receive a discharge of their debt in their bankruptcy cases. This article endeavors to fill that void. To our knowledge, it is the first to articulate the gendered impact of having a dependent on post-petition outcomesin bankruptcy.
In the 1980s, Teresa A. Sullivan, Elizabeth Warren, and Jay Lawrence Westbrook founded the Consumer Bankruptcy Project, which produced a series of survey datasets that enabled study of the demographics of consumer bankruptcy filers.[12] Since then, they have, individually and collectively, published numerous academic works highlighting the effects of gender, marriage, and children on bankruptcy filing rates.[13] For example, Elizabeth Warren found that in 2001, women filing individually constituted the “largest group” of debtors seeking bankruptcy—comprising 39.1% of filers.[14] She determined that this demographic’s representation in bankruptcy had grown by 800% since 1981.[15] This article builds on her work, and that of related scholars, and explores these family-related questions from a new angle.
A separate strand of scholarship analyzes bankruptcy law’s interaction with parental obligations normatively. Many of these articles were published around 2005, when the Bankruptcy Abuse Prevention and Consumer Protection Act (“BAPCPA”) transformed the consumer bankruptcy landscape.[16] These articles tend to argue that current bankruptcy law disadvantages parents and their children.[17] Our research provides empirical evidence that supports this conclusion.
This article also contributes to a growing literature on household finance that attempts to understand why people underutilize bankruptcy.[18] The reason appears partly rooted in “[e]thical qualms, stigma, the value of the option to file for bankruptcy in the future, the probability that creditors will not take action to collect delinquent debt, and lack of knowledge of bankruptcy procedures.”[19] These explanations, however, do not show why so many debtors exit the bankruptcy system after they have already incurred the considerable financial, and potential social, costs of filing in the first place. This article supplies an additional explanation for bankruptcy’s decreasing popularity for indebted households.
III. Background on Consumer Bankruptcy
In the United States, individuals typically file for bankruptcy under one of two chapters: chapter 7 or chapter 13. Chapter 7 is commonly referred to as the liquidation chapter. It requires debtors to disclose all their assets and liabilities and allows the chapter 7 trustee to liquidate the debtors’ non-exempt assets to repay creditors. Chapter 13 is often termed the rehabilitation chapter. It permits debtors to retain their assets and to repay their creditors through a court-approved repayment plan that satisfies certain statutory requirements.
Owing to the different obligations these chapters impose, as well as the duration of cases under each,[20] obtaining a discharge in chapter 13 bankruptcy is significantly more challenging than in chapter 7. Table 1 summarizes the outcomes of all consumer cases filed between 2010 and 2015, as reported in the Federal Judicial Center’s Integrated Database.[21] While the vast majority of chapter 7 cases in that timeframe resulted in successful discharge, about 62% of the chapter 13 cases were terminated without a discharge. The most common reported reason that an individual chapter 13 case was terminated without discharge is that the debtor stopped making payments according to her court-ordered payment plan (24.31% of cases).[22] Another 8.88% of reported cases were dismissed for failure either to submit proof of completion of the mandatory financial-management course that individuals undergoing bankruptcy must take or to make domestic support payments. Although converting from chapter 13 to chapter 7 bankruptcy is an alternative path to discharge for some of these filers, only 10% of cases ended in conversion.[23]
Since missed plan payments are historically the leading cause of dismissals in chapter 13 cases, the following Section examines how chapter 13 plans are structured and how they intersect with parental obligations. This discussion is intended to orient readers unfamiliar with consumer bankruptcy.
A. Chapter 13 Payment-Plan Calculation and Exemptions for Parents
All chapter 13 debtors must complete a payment plan before the discharge of their debts.[24] Two salient components of a chapter 13 plan are the monthly payment amount (the plan funding) and the plan duration (of three to five years), which is also referred to as the commitment period. For most debtors, factors described in § 1325 of the Bankruptcy Code determine the monthly payment amount and the plan duration, as set forth in Forms 122C-1 and 122C-2.[25]
Specifically, Form 122C-1 calculates the debtor’s and her spouse’s average monthly income over the previous six months.[26] That calculation includes wages, net income from the operation of a business, net income from real property, investment income, retirement or unemployment income, alimony, and child support payments, among others sources.[27] The total income for this six-month period is then used to determine the debtor’s current monthly income (“CMI”), which has significant implications for her chapter 13 case.
Indeed, the debtor’s CMI generally dictates whether the plan’s commitment period is three or five years long. For example, if the debtor’s CMI is below the median for other families of the same size in the debtor’s state, her minimum commitment period is usually three years; otherwise, it is five years.[28] This is true unless the debtor repays her unsecured debt in full in a shorter period of time.[29]
If the debtor’s CMI is above the applicable median in her state, the Bankruptcy Code also includes a formula to determine the amount of that debtor’s projected disposable income, which must be devoted to the repayment of creditors under the chapter 13 plan. The projected disposable-income formula is somewhat complicated, and relies on both standardized and case-specific expense deductions to determine the ultimate amount that the debtor must pay unsecured creditors. For simplicity, in the discussion below, we use the term “monthly disposable income” to mean the amount before subtracting the secured debt and priority debt payments—that is, the total amount available for all creditors every month.
B. Existing Protections for Parents
This Section describes at greater length bankruptcy law’s recognition of parental obligations. To illustrate legislative efforts to assist debtors with family obligations, particularly those related to children, we summarize aspects of both chapter 7 and chapter 13 that address family-related obligations. Specifically, we describe the treatment of childcare expenses, courts’ recognition of childcare-related tax credits as exempt from income calculations, and the prioritization of alimony and child support in a chapter 13 plan.
1. Childcare Expenses
In a chapter 13 case, childcare expenses are an allowed expense deducted from the debtor’s income before payments to creditors under the plan. As with many expense calculations, however, the process for determining childcare expenses can be flawed. A significant issue is that this excluded amount is determined based on the six months prior to the declaration of bankruptcy, which may not accurately predict future childcare-related expenses when a debtor has a growing child.
In addition, such expenses also can be deducted under BAPCPA’s means test for the purposes of determining eligibility for chapter 7.[30] The means test creates a presumption of abuse for debtors with relatively high income, which may prevent them from filing or maintaining a chapter 7 case. Under this test, if a debtor’s income is above the state median, she must rebut the presumption of abuse to file or maintain a chapter 7 case.[31] A debtor may deduct any childcare expenses from her income in determining whether she is an above- or below-median debtor.[32] Therefore, parents on the margins who might otherwise not qualify for chapter 7 (for example, because their income is too high to rebut the presumption of abuse) might use any childcare-expense deductions to reduce their income and qualify for chapter 7.[33]
Both chapter 13’s payment plan calculation and the chapter 7 means test only include direct, monetary expenditures. In both calculations, the formula does not account for “soft” indirect expenses that parents incur.[34] For example, parental obligations might reduce a debtor’s ability to maintain steady employment or render her more dependent on her assets.[35] As a result, the childcare-expense metric in bankruptcy will often underestimate the true cost of caring for a dependent. So, out-of-pocket expenses directly tied to raising or supporting a child are an imperfect proxy for the actual amount parents spend on dependent care.
2. Exemption of Childcare-Related Tax Credits
Ordinarily, state law determines the exemption of childcare-related tax credits from a debtor’s estate, from which a bankruptcy trustee may collect. As of this article’s writing, many states exempt the Earned Income Tax Credit—which is derived from a formula based in part on the number of qualifying children in a recipient’s household—and the related Additional Child Tax Credit from the bankruptcy estate.[36] This exemption is far from universal, however, as a significant number of states continue not to recognize such exemptions.[37]
Relatedly, we note a recent change in the treatment of certain tax credits at the federal level. Under the American Rescue Plan Act of 2021, “Chapter 7 and 13 trustees should not consider . . . child tax credits in administering estate assets or calculating disposable income in [C]hapter 13 repayment plans.”[38] So, with respect to child tax credits covered under the American Rescue Plan and the CARES Act specifically, debtors with dependents receive a benefit not available to other debtors.[39]
3. Alimony and Child Support as Priority Claims
Bankruptcy law also gives priority to alimony and child-support obligations. Before a debtor may discharge debt under chapter 13 bankruptcy, she must certify that all domestic-support obligations due have been paid.[40] Further and more importantly, childcare support obligations and alimony are excepted from discharge in both chapter 7 and chapter 13 bankruptcy.[41] Therefore, if the debtor does not fully satisfy these obligations through her bankruptcy plan, she remains liable for those amounts after the conclusion of the bankruptcy case.[42]
In sum, Congress, through the mechanisms discussed above, has sought to safeguard childcare expenses in the bankruptcy process. As we will show in Part IV, these existing protections for filers with caregiving obligations do not fully neutralize the negative effects of filing for bankruptcy with dependents.
IV. Data
For this project, we obtained access to cases from sixty-four bankruptcy district courts across the United States through the PACER system.[43] We then randomly selected 2,000 consumer cases filed between 2016 and 2018 from each of the largest offices in each district for a total initial possible sample size of 128,000. Out of these 128,000 cases, we were able to automatically locate and download the relevant PACER records for 114,200 cases.[44] Using a combination of automatic and manual processing, we extracted an individual’s marital status from Form 107 and the dependent checkboxes from Schedule J for 102,952 of those cases, which compose our final dataset.
It is worth noting that, because we randomly selected cases from the entire pool of consumer bankruptcy cases, approximately 35% were originally filed under chapter 13, while the remaining 65% were filed under chapter 7. This article primarily focuses on chapter 13 cases, though in Part VII, we examine the effects of parental obligations on the choice between filing under chapter 7 and chapter 13. Part V examines only individual chapter 13 filers (about 76% of all chapter 13 cases), and we add joint filers back into our dataset in Part VI. Generally, throughout this article, the number of observations used in each empirical analysis is reported in the last row of our regression tables.
We downloaded roughly the same number of cases from each bankruptcy district regardless of the size of the district. In the Appendix, we show that the results do not meaningfully change when we weigh the observations to make them proportional to the actual caseload of each district. The exact number of observations from each district is provided in Table A3.
We next matched these cases with the Federal Judicial Center’s Integrated Database (“FJC Data”).[45] The FJC Data provides quantitative information about debtors, such as their total assets, liabilities, monthly income, and monthly expenses derived from bankruptcy filings. Unlike most empirical analyses using FJC Data, we did not use the FJC’s disposition information. Rather, we used the most current information available on PACER at the time we downloaded the cases, in the period December 2021 to February 2022. For all cases in our sample, we gathered at least three years of data.[46]
Finally, bankruptcy filings do not require a debtor to indicate her gender. Thus, to impute the gender of the debtors, we matched the first names in the PACER dataset to the Social Security Administration’s list of baby names between 1932 and 2012.[47] This list provides the number of newborns of each gender associated with a given name, which allowed us to infer the likely gender of each debtor. From the Social Security Administration’s list, we used the most common birth gender of individuals with each first name.[48] For names not included in the Social Security Administration list, we referenced Harvard Dataverse’s World Gender Name Dictionary v2.0 to impute gender.[49]
Table 2 below presents summary statistics of select characteristics of individuals within the dataset, separated out by those with and without dependents.
The table illustrates that filers with dependents differ significantly from other filers across all identified attributes. Compared to debtors without dependents, those with dependents are more likely to be married couples (52.11% vs. 32.81%) and are more likely to file jointly (31.64% vs. 19.67%). Among debtors filing individually, those with dependents are more likely to be women (62.39% vs. 55.64%). Debtors with dependents also tend to have more assets and more debts ($96,642 & $139,786 vs. $77,459 & $112,080). Similarly, debtors with dependents have a higher average monthly income, with higher expenses ($3,451 & $3,308 vs. $2,558 & $2,420). These characteristics are consistent with prior scholarship on debtors with and without dependents.[50]
On the other hand, the differences illustrated in Table 2 underscore the difficulty of directly comparing debtors with dependents to those without dependents—and the importance of controlling for confounding factors. Even though the raw averages show that debtors with dependents are nearly four percentage points less likely to obtain discharge compared to other debtors (39.89% vs. 43.55%), that disparity could, without controlling for other variables, simply reflect other underlying differences. We therefore used logistic-regression analysis to address this issue.
Logistic regressions help to isolate the effect of having dependents on discharge outcomes by accounting for other factors that might otherwise influence our results—such as income, assets, geographic location, and the year of filing. This method holds constant these other factors that might have explanatory power, so that we might measure whether having dependents alone makes a difference in the outcomes we observe. In the next Part, we will walk through our methodology at greater length and explain how we used additional information to ensure the accuracy of our findings.
V. Parental Obligations and Chapter 13 Dismissals
In this Part, we examine how having a dependent affects the dismissal rates in individual debtors’ cases. Because in this Part, we focus on comparing male and female debtors, we exclude joint filers. We add joint filers back into our discussion in Part VI.
Figure 1 below is a visual representation of the difference in dismissal rates between debtors with and without dependents by quarter after the petition date, conditional on observed case characteristics.
Figure 1. Conditional Dismissal Rate by Quarters After Filing
For each quarter after filing, we ran a logistic regression where the dependent variable is a dummy variable equal to 1 if the case was terminated without a discharge before the end of the quarter (the y-axis).[51] We set the independent, treatment variable—the parent dummy—to 1 if the debtor had a dependent (the x-axis). Additionally, we controlled for the debtor’s gender, marital status, and financial characteristics,[52] as well as for district- and year-fixed effects.[53]
As a result, each line in Figure 1 represents, for the sample, the cumulative probability of dismissal of an average filer’s case, either with or without a claimed dependent, holding all observed characteristics of the filer constant. The x-axis is the number of quarters following the opening of the bankruptcy case. Figure 1 shows a smooth, widening gap between debtors with and without dependents. Using the regressions described below, we estimate that having a dependent increases the risk of dismissal within three years by about eight percentage points.
To explain the observed gap in Figure 1, we examine whether, and by how much, a debtor’s marital status, gender, and having a dependent, independently or together, impacted her likelihood of a dismissal within three years of filing. Table 3 below represents the results of our analysis of an individual’s dismissal rate within three years.
Table 3—Regression Results Among Debtors Who Filed for Chapter 13 Bankruptcy Individually
Notes:
Sig. Levels: ∗∗∗p < .01, ∗∗p < .05, ∗p < .1.
Clustered standard deviation reported in parentheses.
The dependent variable is a dummy variable that equals 1 if the case was terminated without a discharge within 3 years after the filing date. The average marginal effect (partial effects) of the covariate, ∆ Margin, is reported for Columns (1) to (3) to facilitate the interpretation. Statistically, ∆ Margin represents the difference between the predictive margins (potential-outcome means) when the corresponding dummy variable equals 1 and when the corresponding dummy variable equals 0, which roughly estimates the effect of the dummy variable on the possibility of the termination of a bankruptcy case without a discharge within 3 years after the filing date.
Columns (1)-(3) show the results of a logistic regression with different controls. We first summarize our results in non-technical terms before turning to a more in-depth accounting of our statistical methodology. Our regressions show that having a dependent significantly increases one’s chances of dismissal without discharge (i.e., the coefficient for the Parent dummy variable is significant and positive). This is especially true for unmarried women parents, relative both to male parents and women non-parents (i.e., the coefficient for the Female × Parent × Not Married term in Column (3) is positive and significant). Meanwhile, women who are not parents, especially if they are unmarried, fare better than men, parents and non-parents alike, in the bankruptcy process—meaning that without the parenthood component, women debtors tend to have better outcomes. All told, our study shows that parenthood negatively affects one’s ability to stay in bankruptcy for at least three years. Our results also indicate that unmarried women parents are most vulnerable to this negative effect.
We now describe the statistical mechanisms behind our observations at greater length: the dependent variable in Table 3 is again a dummy that equals 1 if the debtor’s case is dismissed within three years. To facilitate interpretation, we report the average marginal effect of the covariates (∆ Margin) in addition to the logistic regression coefficients, which reflect the changes in log odds. ∆ Margin roughly estimates the effect of the corresponding independent dummy variable on the dependent dummy variable. For purposes of Table 3, then, the ∆ Margin statistic roughly estimates the amount by which the likelihood of termination without discharge within three years increases if the corresponding independent variable changes from 0 to 1. Column (4) is a simple ordinary least squares regression that shows similar estimates compared to the marginal change from the logistic regression.
As expected, controlling for a debtor’s finances increases the estimated treatment effect from Column (1) to Column (2). The average marginal effect is approximately eight percentage points in Column (2). Column (3) shows that unmarried women parents are the primary drivers of the treatment effect, given the interaction term’s significant and positive coefficient. Column (3) also indicates that unobserved characteristics unique to women, who are much more likely to be the primary caretaker of a child, do not explain the effect. That is, the negative and significant coefficients for both the “Female” and “Female × Not Married” dummies suggest that women are generally less likely to have their bankruptcy cases dismissed compared to similarly situated men after controlling for their financial characteristics. Because the coefficients for the female dummy and the parent dummy have opposite signs, the gender gap in parental obligations itself does not explain the positive coefficient for the parent dummy.
Next, we considered whether being a homeowner materially impacts dismissal rates. As prior scholarship notes, homeowners often file for bankruptcy to delay foreclosure and/or to cure mortgage arrears they are otherwise unable to cure outside of the chapter 13 process.[54] If these debtors are more likely to have their cases dismissed because their desired outcome is not actually having their debt discharged, but rather the retention of their homes and/or the curing of arrearages on mortgage debt, then whether a debtor is a homeowner can bias the results.
To explore this possibility, we divided debtors filing individually by whether they were homeowners. Much like for our gender data, the official forms do not require debtors to specifically identify whether they are “homeowners” or “renters.” So, we used a proxy. The bankruptcy schedules require a separate disclosure of real property and secured debt. Thus, we define homeowners as individuals who have over $50,000 in real property listed in schedule A/B, and non-homeowners as those who have no real property.[55] Table 4 below shows our results. While here, we focus exclusively on comparing homeowners and non-homeowners, our results for both groups are similar to that of the full sample in Column (3) of Table 3. Curiously, however, the observed difference in dismissal rates between parents and non-parents is even stronger among non-homeowners, though it does remain sizable among homeowners. This suggests that while home ownership could play a role in dismissal rates, it is not the only reason why debtors with dependents are more likely to have their cases dismissed within the first three years compared to debtors without dependents. Put differently, home ownership might contribute to the disparity, but it does not fully explain why debtors with dependents face higher dismissal rates. Other factors are also driving the outcome.
Table 4—Regression Results Among Debtors Who Filed for Chapter 13 Bankruptcy Individually, Homeowners vs. Non-Homeowners
Note:
Sig. Levels: ∗∗∗p < .01, ∗∗p < .05, ∗p < .1.
Clustered standard deviation reported in parentheses.
The dependent variable is a dummy variable that equals 1 if the case was terminated without a discharge within 3 years after the filing date. As in Table 3, ∆ Margin is reported to facilitate the interpretation, and it roughly estimates the effect of the dummy variable on the possibility of a bankruptcy case’s termination without a discharge within 3 years of its filing date. Homeowner is defined as possession of over $50,000 in real property. Non-homeowner is defined as owning zero real property.
In sum, we estimate that having a dependent increases the risk of dismissal of a debtor’s case within three years by about eight percentage points (i.e., the ∆ Margin statistic from Column (2) of Table 3). Considering the overall discharge rate for chapter 13 bankruptcy is 38%, as shown in Table 1, this eight percentage point difference is quite large.
VI. Gender, Marriage, and Parental Obligations
As explained above, Column (3) of Table 3 shows that parental obligations likely affect women more than men, even when conditioned on being a parent. In this Part, we further explore the interactions among gender, marriage, and childcare obligations. We examine how the effect of having a dependent on dismissal rates differs among individual male and female filers and joint filers.
In the sample, only 56.6% of married debtors filed for bankruptcy jointly.[56] We note that the Bankruptcy Code only allows spouses to file jointly.[57] Indeed, nearly 99% of the joint filers in our sample reported in Form 107 that they were married.[58] Filing jointly allows a married couple to wipe out both individuals’ debts at the cost of one filing fee and perhaps lesser attorney’s fees.[59] On the other hand, filing separately might allow couples to claim more exemptions in some states. Moreover, in many instances the bankruptcy will not affect the non-filing spouse’s non-marital properties and credit score as severely as if she elects to file jointly. Importantly, however, filing separately does not protect the non-filer’s income. A debtor’s attorney may also play a role in the choice to file either separately or jointly.
In Part V, we focused on debtors who filed individually. We now discuss the outcomes of joint debtors compared to those of individual filers.
Table 5 provides a comparison of the likelihood of dismissal within three years across individual female and male filers and joint filers. Overall, our results in Columns (1), (3), and (5) imply that having a dependent increases an individual woman’s chances of having her case dismissed in the first three years by 10.2 percentage points, compared to 3.9 percentage points for individual men and 5.5 percentage points for couples filing jointly.
To elaborate, the dependent variable in these logistic regressions is once again equal to 1 if the chapter 13 bankruptcy case was terminated without discharge within three years of the filing date. First, note that the coefficients on the parent dummy are positive in all three groups, which is consistent with the hypothesis that dependent care negatively affects parents’ bankruptcy outcomes regardless of these individuals’ gender or marital status. Second, while the coefficient is both economically and statistically significant in the female and joint debtor groups, it is barely statistically significant, and somewhat diminished in effect size, in the male individual debtor group after controlling for marital status. As we have previously observed, parenthood has the least negative impact on dismissal rates in men’s chapter 13 cases, compared to those of women and of joint filers. The lesser statistical significance of the coefficient for men means that it is more likely that the effects of parenthood we seem to observe in their bankruptcy outcomes is the result of chance, as opposed to an informative statistic—after we build out a more robust regression that accounts for more variables.
All told, our data show that women generally have a higher discharge rate than men when they do not have a dependent but experience a more drastic increase in likelihood of dismissal relative to men when they do have dependents. One potential explanation for this result might be that women are more likely than men to assume daily, non-financial custodial childcare responsibilities.[60] If women are more frequently shouldering non-financial parenting responsibilities, they are likely to experience greater bankruptcy consequences for having a dependent.[61]
We next consider the effects of marriage. Column (2) of Table 5 below suggests that, while marriage reduces the dismissal rate for all women, it does not meaningfully change the effect of having a dependent among women. Similarly, Column (6) shows filing jointly does not meaningfully change the effect of having a dependent, even though joint debtors are approximately 6.7 percentage points less likely to have their cases dismissed than married debtors filing individually. These statistics suggest that the individual and joint filing schemes are currently equally inadequate at alleviating the burdens of childcare obligations on bankrupt parents.[62]
Finally, Column (6) also suggests that couples who filed jointly differ, in a statistically significant manner, from couples who filed individually. The former are less likely to have their cases dismissed within three years. As described above, what drives this difference is unclear.
Table 5—Regression Results Among Debtors Who Filed Individually Versus Debtors Who Filed Jointly
Note:
Sig. Levels: ∗∗∗p < .01, ∗∗p < .05, ∗p < .1.
Clustered standard deviation reported in parentheses.
The dependent variable is a dummy variable that equals 1 if the case was terminated without a discharge within 3 years after the filing date. As in Table 3, ∆ Margin is reported for Columns (1), (3), and (5) to facilitate the interpretation. This statistic roughly estimates the effect of the dummy variable on the possibility of a bankruptcy case’s termination without a discharge within 3 years after the filing date.
VII. Parental Obligations’ Effect on Chapter Choice
Parental obligations not only affect a debtor’s chances of succeeding in the chapter 13 process, but also her decision to choose chapter 13 in the first place. Table 6 below shows that debtors with a dependent are about 2.5 percentage points more likely to file under chapter 13, rather than under chapter 7, as compared to other debtors in the dataset. In particular, Column (3) shows us that this difference is, again, primarily driven by unmarried female parents.
Table 6—Logistic Regressions on Chapter Choice
Note:
Sig. Levels: ∗∗∗p < .01, ∗∗p < .05, ∗p < .1.
Clustered standard deviation reported in parentheses.
The dependent variable is a dummy variable that equals 1 if the case was filed under chapter 13. As in Table 3, ∆ Margin is reported to facilitate the interpretation. ∆ Margin represents the difference between the predictive margins when the corresponding dummy variable equals 1 and when the corresponding dummy variable equals zero, which roughly estimates the effect of the dummy variable on the possibility of filing for chapter 13 bankruptcy instead of chapter 7.
Stata was not able to achieve convergence for the regression in column (2) but the log pseudo-likelihood was stable after 5 iterations. We used the results from the 10th iteration. All other regressions achieved convergence.
The true effect of having a dependent on chapter choice is likely higher than 2.5 percentage points, because household size directly impacts the means test thresholds for chapter 7. In other words, in comparing two debtors with identical incomes, the debtor with dependents is more likely to qualify for chapter 7 than the debtor without, because the median income applicable to the former is higher. We would likely observe an even stronger effect of having a dependent on whether a filer chooses chapter 7 or chapter 13 with this selection bias removed.
There are many reasons why filers, and particularly women, with dependents are more likely to choose chapter 13 over chapter 7. For example, the ability to retain assets can be more important to parents than to non-parents. Unlike chapter 7, chapter 13 allows parents to retain their houses and cars even when these assets are not fully exempted from the debtor’s bankruptcy estate. Chapter 13 also permits filers with dependents to attempt to delay any foreclosure or repossession for the duration of the chapter 13 plan if her creditors are hostile. These advantages can prove crucial to parents striving to provide a stable environment for their children.
VIII. Policy Recommendations
As earlier Parts of this article indicate, the current Bankruptcy Code does not fully achieve the articulated goals of Congress to protect children from the effects of their parents’ bankruptcy. As our data show, parental obligations likely make it harder for parents to discharge their debts compared to similarly situated non-parents. Parents are more likely to see their chapter 13 cases dismissed in the first three years. We next discuss bankruptcy-related policy options that might aid in closing this gap and attendant, potential roadblocks.[63]
Congress might consider allowing additional deductions for parents when calculating their disposable income. This solution would help decrease parents’ monthly payment-plan amount so that they are better situated to afford childcare while undergoing bankruptcy. However, these additional deductions would support these debtors if and only ifthey retain substantial surplus income after paying off their creditors who are entitled to full payment under a given plan. That is, additional deductions would only allow parents to keep what would otherwise be disposable income to be paid to unsecured creditors. If their income is not high enough to generate such surplus, however, additional deductions may not be of use.
Moreover, additional deductions could help some parents pass the means test. Parents who wish to file for chapter 7 but are currently unable to rebut the presumption of abuse because they have too much disposable income could benefit from deductions that lower their calculated disposable income. However, for parents who prefer to file under chapter 13 rather than chapter 7, this particular benefit would not apply.
Alternatively, a longer commitment period could reduce debtors’ minimum monthly payment amount and thus alleviate financial strain. Suppose, for example, that a debtor owes $1,500 in priority debt. This debtor’s must have at least $500 in disposable income a year to qualify for a three-year plan. If she only makes $300 a year, she can still file for bankruptcy under chapter 13 with a five-year plan. If her disposable income drops to less than $300 a year while she is undergoing bankruptcy, her chapter 13 case will be dismissed—because the Bankruptcy Code does not allow plans to last more than five years and she would not be able to pay back all her priority debt by the end of the original plan.[64] However, if the debtor might extend her plan, her minimum payment would decrease. A longer commitment period might enable her to stay in the bankruptcy process and ultimately obtain a successful discharge.
In fact, in other contexts, Congress has already facilitated the use of extended plans to provide relief. Under the CARES Act, a chapter 13 debtor could seek plan modifications to extend her payment plan to up to seven years, if she was experiencing COVID-related hardships and if the plan was confirmed prior to a statutory cutoff date.[65] A similar policy might aid parents in coping with childcare-related financial challenges while undergoing bankruptcy.
Finally, Congress might consider reducing the downsides of the dismissal of a parent’s chapter 13 case. More empirical research is needed to demonstrate the benefits of converting an unsuccessful chapter 13 case to a chapter 7 case, but we observe a low conversion rate in Table 1. If further research indicates that obtaining a chapter 7 discharge after an unsuccessful chapter 13 case is beneficial, Congress might consider making such conversion a default rule, without depriving debtors of the choice to opt out of such conversion.
IX. Conclusion
This article studies the effects of parental obligations on consumer bankruptcy outcomes. In particular, we show that having a dependent while undergoing chapter 13 bankruptcy increases, by eight percentage points, the likelihood of dismissal of a debtor’s case within three years of the filing date. We then demonstrate that differences in marital status or home ownership between parents and non-parents cannot explain this effect.
Moreover, we find that this effect is most pronounced for women with dependents who filed for bankruptcy individually. These women are 10.2 percentage points more likely to have their cases dismissed within three years when they have a dependent. In contrast, men with dependents filing individually are only 3.9 percentage points less likely to succeed, whereas joint filers with dependents are 5.5 percentage points less likely to succeed. Our data also suggest that marriage does not seem to reduce this gender gap for debtors who filed individually.
Our results suggest that under current bankruptcy law, children with parents undergoing the bankruptcy process sometimes fall through the cracks. Changes in law to further assist parents undergoing bankruptcy might level the disparities we observe.
Appendix
This Appendix focuses primarily on robustness checks.
A1. Heckman Selection Model
Because parents and non-parents have different tendencies to choose chapter 13 as opposed to chapter 7, selection bias might drive some of our observed effect. Tables A1 and A2 apply Probit Heckman Selection Models to the full sample to account for this selection bias. An observation is “selected” if it is a chapter 13 case. Compared to Tables 3 and 5, the significance and magnitude of some coefficients are reduced but our main conclusions remain unchanged.
A2. Weighting by District Size
Table A3 shows the number of observations that we used in the main analysis, broken out by district. In the main analysis, we gave equal weight to all observations. Table A4 weights the samples by district so that the effect of cases from each district on the regression results is proportional to the actual number of cases filed under chapter 13 between 2016 and 2018. This method yields a better representation of the national average treatment effect. The results do not meaningfully differ from Table 3.
A3. Results Using 2016 Data
One potential criticism of this article is that its sample period includes the COVID-19 pandemic. Because we used the three-year dismissal rate as the dependent variable for most of our regressions, cases filed in 2017 and 2018 will include three-year windows with outcomes from 2020 and 2021.
However, we only observe outcomes through 2019 for cases filed in 2016. Therefore, we repeated our analysis using only 2016 data. As Table A5 suggests, the main results do not meaningfully change when we use only 2016 data. The economic significance of the parent dummies increases, and the signs of the coefficients are similar to Table 3, although the statistical significance is greatly reduced because of the smaller sample size.
Figure A1 is exactly the same as Figure 1 but it uses only cases from 2016. The pandemic did not seem to affect the overall trend for these cases. Although there is a slight uptick in the last three quarters, presumably due to the pandemic, the widening fork does not look much different from Figure 1.
Appendix Tables and Figures
Table A1—Probit Heckman Selection Model (Corresponding to Table 3)
Note:
Sig. Levels: ∗∗∗p < .01, ∗∗p < .05, ∗p < .1.
Clustered standard deviation reported in parentheses.
The dependent variable is a dummy variable that equals 1 if the case was terminated without a discharge within 3 years after the filing date. As in Table 3, ∆ Margin is reported in Column (3) to facilitate interpretation. The controls do not include average monthly expenses because, owing to its high correlation with average monthly income, the model cannot achieve convergence when both variables are included.
Table A2—Probit Heckman Selection Model (Corresponding to Table 5)
Note:
Sig. Levels: ∗∗∗p < .01, ∗∗p < .05, ∗p < .1.
Clustered standard deviation reported in parentheses.
The dependent variable is a dummy variable that equals 1 if the case was terminated without a discharge within 3 years after the filing date. As in Table 5, ∆ Margin is reported in Column (3) to facilitate the interpretation. The financial controls do not include average monthly expenses because, owing to its high correlation with average monthly income, the model cannot achieve convergence when both variables are included.
Table A3—Number of Observations per District
Table A3—Continued
Table A3—Continued
Table A4—Main Regressions Using Weighted Data
Note:
Sig. Levels: ***p < .01, **p < .05, *p < .1.
Clustered standard deviation reported in parentheses.
The dependent variable is a dummy variable that equals 1 if the case was terminated without a discharge within 3 years after the filing date.
Table A5—Main Regressions Using 2016 Data
Figure A1. Conditional Dismissal Rate by Quarters After Filing Using 2016 Data
***