Financial collapse and class status: Who goes bankrupt?

Osgoode Hall Law Journal

Volumn 41, Issue 1, 2003, Pages 115-146

  Warren, Elizabeth ^a  

^a HARVARD LAW SCHOOL   (United States)

Author keywords

Bankruptcy; Social aspects; United States

Indexed keywords

EID: 33645664834     PISSN: None     EISSN: 28175069     Source Type: Journal    
DOI: 10.60082/2817-5069.1432     Document Type: Article

References (56)

1. In 1991, there were872,438 non-business bankruptcy filings; in 2001, there were 1,452,030 non-business filings. Of course, population increased during the decade as well. The increase in filings as adjusted for population can be expressed as a filing rate per 1,000. In 1991, the filing rate per 1000 adults was 6.34 people; in 2000 the rate was 9.25, an increase of 46.1 per cent. Administrative Office of the United States Courts, Tables F2 (total filings, total business filings, total business Chapter 13 filings, total non-business filings), F2E (total joint business Chapter 13 filings) and F2F (total joint non-business filings). U.S. Census Bureau, Population Division, Population Estimates Program, online: U.S. Census Bureau (1991 to 1999 population data); (2000 population data); (2001 population data). The number of adults filing for bankruptcy was calculated by adding all individual filings to twice the number of joint filings. This number was divided by the population in thousands aged eighteen and over to get the filing rate per 1,000 adults.
2. Douglas Cecil North, Institutions, Institutional Change and Economic Performance (Cambridge: Cambridge Press, 1990).
3. Teresa A. Sullivan, Elizabeth Warren & Jay Lawrence Westbrook, As We Forgive Our Debtors: Bankruptcy and Consumer Credit in America (New York: Oxford University Press, 1989) at 7.
4. Ibid. at 102.
5. Teresa A. Sullivan, Elizabeth Warren & Jay Lawrence Westbrook, The Fragile Middle Class: Americans in Debt (London: Yale University Press, 2000) at 27 [Sullivan, “Fragile Middle Class”].
6. David A. Moss & Gibbs A. Johnson, “The Rise of Consumer Bankruptcy: Evolution, Revolution, or Both?” (1999) 73 Am. Bank. L.J. 311 at 322-23 [Moss & Johnson].
7. Ibid. at 332-35.
8. Ibid. at 335.
9. The basic analysis of credit card profitability and pricing was laid out in the seminal article by economist Dr. Lawrence M. Ausubel, “Credit Card Defaults, Credit Card Profits, and Bankruptcy” (1997) 71 Am. Bank. L.J. 249.
10. The Federal Reserve Board documented the high profitability of credit card lending, noting, for example, that in the third quarter of 1997 credit card banks showed a 3.14% return on assets, compared to a 1.81% return on assets reported by all commercial banks. “The Profitability of Credit Card Operations of Depository Institutions” (June 2001), online: Federal Reserve Board .
11. An example of the profits to be made as the spread between the wholesale and retail cost of funds widens occurred in 2001. The Federal Reserve cut interest rates nine times, but only a fraction of that savings was passed on to consumers in the form of lower interest rates. As a result, total savings from the nine interest rate cuts created a $10 billion windfall for credit card issuers, whose cost of funds declined sharply while the rates they charged their customers remained surprisingly sticky. Cecily Fraser, “A $10 billion windfall: Credit card lenders don’t pass on full interest-rate cuts” CBS MarketWatch (3 October 2001), online: CBS MarketWatch.com . The reason rates did not drop was attributed in part to the 25 per cent of cards that offer variable interest rates but that have a minimum, or so-called “floor”, to ensure that rates do not dip below a certain price.
12. Moss & Johnson, supra note 6 at 337.
13. In 1991 we administered questionnaires to debtors who filed for bankruptcy in sixteen judicial districts: The Eastern District, Middle District, and Western District of Pennsylvania; the Northern District, Central District, and Southern District of Illinois; the Western District and Middle District of Tennessee; the Northern District, Western District, Eastern District, and Southern District of Texas; and the Northern District, Central District, Eastern District, and Southern District of California. We selected a random sample of the 59,000 people who responded to questionnaires administered at the meeting that debtors are required by statute to attend so that they may be questioned by their court-appointed trustee and by any of their creditors who wish to participate. This effort created a sample of approximately 150 debtors per district, or a total of 2,452 debtors. We then collected court records for debtors in five of the districts: The Eastern District of Pennsylvania, the Northern District of Illinois, the Western District of Texas, the Middle District of Tennessee, and the Central District of California. For a more detailed description of the 1991 sample, see Sullivan, “Fragile Middle Class”, supra note 5 at 263-87.
14. The 2001 sample of debtors was drawn prior to the events of September 11, 2001, and the subsequent economic fallout. For this research, we refocused our resources somewhat. We visited the same five states as we did in 1991, but instead of collecting less data from each of the districts within the state, we collected more data from only one district. For the 2001 study, we collected data from the Eastern District of Pennsylvania (Philadelphia), the Northern District of Illinois (Chicago), the Middle District of Tennessee (Nashville), the Northern District of Texas (Dallas), and the Central District of California (Los Angeles). Four of the districts are the same as those for which we also had both court records and questionnaire data from 1991, with the Northern District of Texas substituted in 2001 for the Western District of Texas in 1991. We substituted the Northern District of Texas for the Western District so that we could include Dallas within the sample rather than San Antonio and West Texas, in part so that we could study the Dallas homeowners in more detail in order to compare them with homeowners in Chicago, Los Angeles, Nashville, and Philadelphia. In 2001, we collected longer questionnaires from a larger group»—250 debtors per district—as well as court records for each debtor in the sample and information from one to two hour telephone surveys for each debtor we could reach. In order to explore particular research questions, we also collected data from an additional 521 homeowners in bankruptcy in those districts and an additional 48 rural debtors, for a total of 1,819 debtors in the five districts. Because these additional debtors were chosen specifically because they were homeowners or lived in a rural area, their selection may bias the core sample, so they are not used for analysis in this article. For more details on the 2001 study and the expanded database, see Elizabeth Warren, “Bankrupt Children” (2002) 86 Minn. L. Rev. 1003 at 1026-32 (reproducing a copy of the questionnaire used to collect data) [Warren, “Bankrupt Children”].
15. We collected the court records for every debtor in the sample, copying the petition and schedules and coding about 150 pieces of information from each one.
16. Hach debtor is required to attend a meeting with the assigned bankruptcy trustee; creditors may attend as well to ask the debtor questions under oath. At these meetings, dubbed “341 meetings” in honour of the section of the bankruptcy code that requires them (11 U.S.C. 341), each debtor was handed a four-page questionnaire that captured basic demographic information (age, sex, marital status, race, number and age of dependents, occupation, etc.) as well as specific financial information (was the debtor currently employed, had the debtor lost a home for financial reasons, etc.). Debtors who returned the questionnaires were part of the sample.
17. In the questionnaire, we asked each debtor for permission to call their home for an interview about the bankruptcy filing. Thanks to generous funding from the Ford Foundation and the Robert Wood Johnson Foundation, we were able to offer each family $50 for completing a telephone interview. The response rate was high; we eventually completed interviews with 67.7 per cent of the debtors in the core sample. Interviews were conducted by trained interviewers under the direct supervision of Dr. Deborah Thorne, now of Ohio University Department of Sociology.
18. More details about the study are available in Warren, “Bankrupt Children”, supra note 14 at 1026-32; Raisa Bahchieva et al., “Home Ownership and Financial Distress: The Interplay of Tax, Real Estate and Bankruptcy Laws: A Report to the Ford Foundation” (2002) archived with Elizabeth Warren.
19. The General Social Survey reports that 91.6 per cent of Americans identified themselves as “working class” or “middle class,” a number that has held steady over the past two decades. General Social Survey, 1976-1996, Variable 185 A CLASS . The problem of ubiquitous self-identification as “middle class” causes sociologists to use alternative classification criteria, such as education or occupational prestige. See Thomas J. Gorman, “Cross-Class Perceptions of Social Class” (2000) 20 Soc. Spectrum 93 at 100; David Halle, America’s Working Man: Work, Home, and Politics among Blue-Collar Property Owners (Chicago: University of Chicago Press, 1984); Annette Lareau, Home Advantage: Social Class and Parental Intervention in Elementary Education (London: Falmer Press, 1989).
20. Thorstein Veblen, Theory of the Leisure Class (New York: Penguin, 1967).
21. See generally Gorman, supra note 19; Halle, supra note 19; Lareau, supra note 19.
22. Indeed, some researchers make a study of contrasting self-identification of class standing with objective criteria of class standing. See generally Eric Plutzer & John F. Zipp, “Class, Gender, and the Family Unit; A Dynamic Model of Stratification and Class Politics” (2001) 30 Soc. Sci. Res. 426; Janeen Baxter, “Is Husband’s Class Enough? Class Location and Gass Identity in the United States, Sweden, Norway, and Australia” (1994) 59 Am. Soc. Rev. 220; Ye Luo & April Brayfield, “Women’s and Men’s Subjective Class Identification from the 1970s to the 1990s” (1996) 16 Soc. Spectrum 83.
23. See U.S. Census Bureau, Statistical Abstract of the United States: 2001 (Washington: Department of Commerce, 2001), online: U.S. Census Bureau .
24. See U.S. Census Bureau, Median Household Income 2001, online: U.S. Census Bureau .
25. We use household income as the appropriate government standard because bankruptcy filers include one-person households and family income calculations for the Census Bureau eliminate all one-person households. If we compared the bankrupt debtors with U.S. data on family incomes, the numbers would be even more extreme. Fully 49.7 per cent of the bankrupt families had incomes in the lowest income quintile in the U.S. (under $24,000). More than eight out of ten (83.5 per cent) had incomes in the bottom two quintiles (under $41,127). U.S. Census Bureau, Current Population Survey, March Supplement, online: U.S. Census Bureau .
26. Median household income by quintiles comes from the U.S. Census Bureau, Current Population Survey, March Supplement, and are reported at U.S. Census Bureau, ibid. The five income quintiles in the U.S. in 2001 were:
27. Ibid. and data from Consumer Bankruptcy Project III, supra note 14.
28. See U.S. Census Bureau, Statistical Abstract of the United States: 1999 (Washington: Department of Commerce, 1999), online: U.S. Census Bureau [Statistical Abstract of the U.S. 1999].
29. Tim Heaton, “Objective Status and Class Consciousness” (1987) 68 Soc. Sci. Quarterly 611, explains that the “objective” determinants of class status are education, occupation, and income. Home ownership appears less often in the sociological literature, but it nonetheless is reported from time to time in discussions of class status.
30. Gorman, supra note 19 at 93, reports data on all three criteria, but classifies respondents based on occupation; Luo & Brayfield, supra note 22, use occupation and education to classify respondents; Heaton, supra note 29, identifies education, occupation, and income as objective measures, Plutzer & Zipp, supra note 22, use occupation and self-identification to determine class status.
31. Gorman, supra note 19 at 120, uses occupational prestige to differentiate working-class and middle-class respondents, but is quick to note both the education and home ownership status of both groups.
32. Difference significant at p =.001.
33. Gorman, supra note 19, relied exclusively on his respondents’ occupations in order to segregate them by class standing. By comparison, Baxter, supra note 22, discusses class differences between husbands and wives, using self-identification, occupation, and education as independent measures of class.
34. The occupational prestige score for clergymen is 69, for dental hygienists is 61, for heavy equipment mechanics is 33, and for longshoremen is 24. The National Opinion Resèarch Center, 1980 Census Occupational Category (Chicago: University of Chicago, 1998), online: Inter-university Consortium for Political and Social Research .
35. For a more detailed description of occupational prestige scores and their use, see Robert W. Hodge, Paul M. Siegel & Peter M. Rossi, “Occupational Prestige in the United States, 1925-1963” (1964) 70 American Journal of Sociology 286. For those unfamiliar with occupational prestige scores, it is possible to glean some understanding by scanning through the long lists of codes developed by the National Organization for Research at the University of Chicago. Food-counter workers are near the bottom at 15, along with construction workers and peddlers at 17, maids at 18, and produce packers at 19. In the middle range, a retail sales clerk is 29, a cashier is 31, an air-traffic controller is 43, and a billing clerk is 45. Near the top of the ladder, architects and aeronautical engineers are up at 71, outstripped by lawyers at 76, college professors at 78, and physicians at 84.
36. The debtors’ answers were entered verbatim into the database, and then coded numerically using the 1970 U.S. Census codes and the corresponding prestige scores developed by the National Opinion Research Council. Thus, when debtors provided their occupations, they were assigned a three-digit occupation code and a two-digit prestige code. The following additional occupational codes were developed: 990 = housewife, homemaker, at-home mom, and similar responses; 991 =retired; 992=disabled or too ill to work; 993=unemployed; 994=not in the labor force, no further information; 995=student; 998=occupation specified but not otherwise classifiable. These codes permit the identification and analysis of specific groups of petitioners, such as the disabled or the retired. Each of these codes was assigned a prestige score of 0. When two occupations were listed, the first was coded. Thus, the person who listed himself as “Self-employed ice cream retailer, riverboat captain,” was coded 245 for “managers and administrators not elsewhere classified.” “Service Customer Clown Company” was assigned 394, for “miscellaneous clerical workers.” If the petitioner listed a status first, such as “disabled ironworker,” the first-mentioned term (922=disabled) was coded. Further information concerning occupational coding is available from Dr. Teresa Sullivan. A number of answers were provided in Spanish. Dr. Teresa Sullivan, who coded the occupations, sought assistance in translating them from Dr. Marta Tienda, Professor of Sociology at Princeton University. Several of these were best translated as “maid” or “housekeeper” – e.g., limpieza or ama de casa (both coded as 982, private household housekeepers). Others indicated status, such as desempleado (unemployed and coded as 993). “En una Zapateria Manager de una tienda” was coded as 245 (managers and administrators not elsewhere classified). “Operado maquina” was coded as 692 (machine operatives, not specified).
37. The 1970 occupational codes have been used in all of the preceding Consumer Bankruptcy Projects. They are also still widely in use as the last “pure” occupational codes used by the Census Bureau. They have been used in major studies such as the General Social Survey. Since the 1970 Census, the Bureau has moved to sets of codes that incorporate industry as well as occupation, but there are several “walkovers” available that permit correspondence from one set of codes to another. The baseline codes used for this set of questionnaires are available at The National Opinion Research Center, 1970 Census Occupational Category (Chicago: University of Chicago 1998), online: . Project Director Dr. Deborah Thorne re-checked the 2001 codes for consistency and accuracy.
38. The actual numbers for 1991 are:
39. Three different analyses of 1991-2001 occupational prestige scores resulted in different findings. A comparison of the occupational prestige scores of first petitioners in 1991 with first petitioners in 2001 shows a drop in prestige from 1991 to 2001 that is statistically significant (p =.001). But that analysis omits the second petitioner (nearly always the wife) in joint petitions from the calculation. When both petitioners are represented in the calculation, with their prestige scores averaged to create a single prestige score for the household, there is no statistically significant difference between the occupational prestige scores of the 1991 debtors and the 2001 debtors. When both petitioners are represented individually in the calculation with no averaging per household, a statistically significant difference between the occupational prestige score of the 1991 debtors and the 2001 debtors re-emerges. For all three calculations, people who are not working for pay outside the home are not included.
40. Melvin L. Oliver & Thomas M. Shapiro, Black Wealth/White Wealth: A New Perspective on Racial Inequality (New York: Routledge, 1995).
41. Primary residences account for 55 per cent of all non-financial assets held by families in the U.S.. Arthur B. Kennickell, Marth Starr-McCluer & Brian J. Surette, “Recent Changes in U.S. Family Finances: Results from the 1998 Survey of Consumer Finances” (2000) 86:1 Federal Reserve Bulletin 1 at 14 (reporting on data from 1998-99, the most recent available).
42. Ibid. at 17, Table 8, Family holdings of non-financial assets, by selected characteristics of families and type of asset, and of any asset, by family characteristic, 1995 and 1998 surveys.
43. U.S. Census Bureau, Statistical Abstract of the United States: 1993 (Washington: Department of Commerce, 1993) at Table 1247, Recent Home Buyers—General Characteristics: 1976 to 1992.
44. U.S. Census Bureau, Statistical Abstract of the United States: 1995 (Washington: Department of Commerce, 1995), online: U.S. Census Bureau at Table 1229, Home ownership Rates, by State: 1984 to 1994 (1995).
45. Calculated from data reported in Sullivan, “Fragile Middle Class”, supra note 5 at 347, n. 23.
46. U.S. Census Bureau, Statistical Abstract of the U.S. 1999, supra note 28.
47. The 1991 data regarding education and occupation came from the debtors’ confidential reports on questionnaires which they filled out when they met with their creditors and trustee at the courthouse, but which were never part of the debtors’ court records. By contrast, the data on home ownership were taken from the debtors’ court records. The home ownership data were extracted from Schedule A (Real Property), Schedule C (Property Claimed as Exempt), and Schedule D (Creditors Holding Secured Claims), which require information about the debtors’ real estate holdings, claimed exemptions, and outstanding mortgages. We had used the same sources when we collected data from debtors who had filed in 1981. But we presented our 1981 data on home ownership, noting that even when we could tell that the debtor seemed to own a home, there was no mortgage listed. We offered this caveat in interpreting the data: Debtors are required to list all their debts so that the court can notify all creditors of a debtor’s bankruptcy. There are no exceptions, and a debtor who fails to comply risks possible perjury charges and denial of discharge…. So why is there no mortgage listed [in many cases] ? Attorneys and judges explained that some debtors do not want their mortgage companies to hear about the bankruptcy. These debtors are generally current on their payments, and they do not want any trouble. Although few mortgage lenders would foreclose a mortgage held by someone now in bankruptcy if all payments were current, debtors still worry. Both the attorneys and the courts are indulgent enough on the point that some debtors passed through bankruptcy with their files uncorrected. And even when the homeowners correctly report their home mortgages, sometimes the mortgage company’s name and address are deleted from the files or from the mailing list for the notifications to creditors. After all, if the debtor will keep paying this particular debt, why stir up trouble? Sullivan, Warren & Westbrook, As We Forgive Our Debtors, supra note 3 at 134-35 [footnotes omitted]. We continue to wrestle with the under-reporting hypothesis. But in 2001, the multiple data sources make it possible to explore home ownership among the families in bankruptcy in greater detail. The 2001 questionnaire included questions about whether the debtor owned a home at the time of the bankruptcy filing. The court records alone showed a 50 per cent home ownership rate, but another 0.9 per cent of the debtors admitted they were homeowners at the time of filing, even though their court records listed no home and no home mortgage. Because these questionnaire and phone survey data are available for the first time in 2001, it is not possible to compare under-reporting in 2001 with possible under-reporting in earlier years.
48. In Sullivan, “Fragile Middle Class”, supra note 5, my co-authors and I speculated that the reported proportion of homeowners in bankruptcy in 1991 was understated in part because the home ownership rate in one of our sample districts, the Central District of Los Angeles, was quite low, perhaps because high mortgages made the debtors ineligible for Chapter 13 to try to work out a payment plan with their lenders. Such families, we speculated, might be in Chapter 11 or out of the bankruptcy system altogether. For a more detailed explanation of the 1991 differences among districts, see Sullivan, “Fragile Middle Class”, supra note 5 at 337-38, nn. 24-28. We estimated that actual home ownership rates in 1991 might have been closer to 50 per cent. Ibid. at 204-05.
49. It is possible that the home ownership rates have remained fairly stable over time. Using the court record data only, we reported an estimated home ownership rate of about 52 per cent for the families filing for bankruptcy in 1981. Sullivan, Warren & Westbrook, As We Forgive Our Debtors, supra note 3 at 129. This report is remarkably close to the 2001 home ownership rate of 52.5. These two points in time offer some additional suggestion that the 1991 data reports may be unusually low.
50. P =.001.
51. “Education tends to be a good predictor of earning ability over the long term, and also of net worth.” Kennickell, Starr-McCluer & Surette, supra note 42 at 6.
52. “Ownership [of homes from 1989-1999] grew strongly for families with incomes of $100,000 or more…” Ibid. at 18.
53. The cutoff for an occupational prestige score for the 20th percentile of the general population was 31; this means that only debtors with an occupational prestige score of 31 or higher qualified on occupational grounds for the combined “indicia of middle class.”
54. If the 90th percentile in the U.S. population were used instead of the 80th percentile, the proportion of debtors qualifying would increase. An estimated 71.4 per cent of the debtors have jobs with an occupational prestige score of 31 or higher, placing them in the upper 80 per cent of occupational prestige scores among Americans generally. An estimated 77.2 per cent of the debtors have jobs with an occupational prestige score of 27 or higher, placing them in the upper 90 per cent of occupational prestige scores of Americans generally. Using the 90th percentile on occupational prestige, the proportion of debtors having one or more of the three characteristics that suggest participation in the middle class would increase from 91.8 per cent to 93.3 per cent.
55. The calculation is done by case, not individual by individual. For single filing individuals, of course, the case and the individual are the same unit of analysis. But for joint petitioners, the two are combined into a single household of unified status. So, for example, if the husband is a bank vice-president and the wife is employed as a day care assistant, this analysis treats the family unit as within the constructed variable for middle class based on the high prestige of his occupation even though her occupation would not qualify.
56. These are other causes of bankruptcy we identified and discussed at length in Sullivan, “Fragile Middle Class”, supra note 5.