Executive Office for U.S. Trustees; Washington, D.C.1
In an article published recently on the Internet, Prof. Robert Lawless presented data showing a tight and positive relationship between annual measurements of consumer debt level and the numbers of non-business bankruptcies filed in the same year.2 From the perspective of this column, which attempts to characterize bankruptcy questions by reference to quantitative information, Prof. Lawless's work provides a useful starting place to consider some problems facing the analysis of the causes of consumer bankruptcy. There are also lessons to be learned about the relationship between the causes of bankruptcy and decisions about how the bankruptcy system should operate.
It will come as no surprise that, as a general proposition, consumer bankruptcy filings fluctuate with debt levels. We would be surprised, in fact, if the opposite were true. But there is more to the story than that.
In the first place, different measures of consumer or household debt are published on a regular basis. Everything else being equal, we would expect that any of these measures would show the expected co-variation with filings. The truth is more complicated. For example, Prof. Lawless found that annual changes in household debt service burden measured over a 20-year period did not correlate closely with annual changes in non-business filings, even though other measures of debt did.3 Lawless concluded that the lack of a demonstrated relationship most likely arose from a combination of misestimations of household debt burden and insufficient sensitivity in the methods that he used to test the association between burden and bankruptcy filings.
The question of measurement sensitivity leads to the next point, which is that our intuition tells us that increases in debt level should be a cause of increased filings. That means that the debt levels should grow before their effects show up in the filings. We should expect the tightest relationships between the two factors to occur when debt levels reported for one year are compared to filing levels of the following year, or even the year after that. Using such offsets is often referred to as lagging one variable behind the other.
One study has shown that when bankruptcy filings were lagged one year behind debt service burden over the years 1981-99, a reasonably clear positive relationship emerged.4 The author of that study, Kim Kowalewski of the Congressional Budget Office, concluded that "[g]rowth in the filing rate tends to increase (or decrease) in the year following a greater (or smaller) increase in the debt-service burden. In particular, the slowdown in the rate's growth during the past few years [late 1990s] corresponds to smaller increases in the debt-service burden."5
We can get yet a closer view of relations between debt levels and filing levels by comparing percentage changes in both factors and lagging the bankruptcy numbers behind the debt-level numbers. The figure below demonstrates a tight relationship between changes in consumer debt levels (not debt service burdens) and filing levels lagged by two years. The relationship is shown over a 30-year period.6
In sum, in spite of problems of data availability and methods for measuring the relationship between debt levels and filing levels, the conclusion that the two are causally related is supported by objective analysis as well as common sense. But that conclusion still cannot be the end of the storyfor two reasons. First, we know that many consumers with large debts do not file for bankruptcy and that some consumers with low debts do file, so there must be more to the causes of filing, or not filing, than debt. Second, accepting that high consumer debt tends to produce bankruptcy provides an insufficient foundation for answering bankruptcy policy questions. We consider each of these reasons briefly.
The relatively tight connection between debt levels and bankruptcy rates only moves our search for answers one step back; for now, we must inquire about the causes of high debt levels. There are many candidatese.g., marital break-up, job loss or illness. Increases in the availability of credit are another factor often cited. In addition, some have argued that a decline in moral standards (phrased as a reduction in shame or stigma associated with debt and bankruptcy) has produced both debt and bankruptcy growth.7
There are also candidates for causes of variations in bankruptcy filing rates that would not necessarily affect debt levels. These can arise with the law itself, including, for example, differences between states in their exemption levels and wage garnishment laws, as well as the provisions of the 1978 Act and subsequent amendments that have been called more or less "debtor-friendly." And of course, there are provisions of new legislation pending that are intended to affect both filing rates and choice of chapter by consumers. Taken all together, this list creates an almost impenetrable network of interconnected factors that are plausible causes of (or reasons for) a consumer's decision to file.
Efforts to discern the effect of divorce on bankruptcy filings is illustrative. One study concluded that divorce rates partially predicted consumer filing rates. The authors of the study interpreted the relationship using two different theories. One theory was that divorces cause bankruptcy because the dissolution of the household produces financial distress. The second theory was that divorce and bankruptcy are both examples of breaking promises, and that an increase in promise-breaking "across the board" has produced the increases in both divorces and bankruptcy filings. Although the authors of the study chose to emphasize the second theory over the first, they acknowledged that their data, standing alone, could not support their choice.8 This interpretive problem holds for all studies that attempt to assess the effect of changes in shame or stigma on bankruptcy filings, because there are no direct measures of such changes (or of the factors themselves); inevitably, their import resides in the judgment of the analyst. The judgments are not necessarily wrong, but they are always subject to a different interpretation that is not refutable by the numbers themselves.
This is why studies about the causes of bankruptcy provide ambiguous or insufficient guidance for answering bankruptcy policy questions. The data always require interpretations that include a set of assumptions that go beyond the numbers themselves. Given different assumptions, the numbers will be interpreted differently. If we assume that there is less shame in society than there used to be, we are likely to interpret statistics regarding debt, divorce and bankruptcy differently than if we assume otherwise, but the current data don't prove the assumption either way. Our attitudes about debtors and about appropriate legal changes are, nevertheless, guided by our assumptions as well as our interpretations of the data.
1 All views expressed in this article are those of the authors, and do not necessarily represent the views of the Executive Office for U.S. Trustees or the Department of Justice. Return to article
2 Lawless, Robert M., "The Relationship Between Non-business Bankruptcy Filings and Various Basic Measures of Consumer Debt," (Version 1.1, July 18, 2001). Return to article
3 This is a statistic reported by the Federal Reserve and is, approximately, the percentage of income absorbed by household debt service. Return to article The failure to show a relationship between debt service burden and filings was over the 20-year period from 1980-00. Over the period from 1993-00, the expected positive relationship could be shown; this discrepancy is currently unexplained. The Federal Reserve states that the statistic is based on "rough approximations" gathered from several data sources. See www.federalreserve.gov/releases/housedebt/about.htm. Return to article
4 Kowalewski, Kim, "Personal Bankruptcy: A Literature Review," Congressional Budget Office, September 2000, p. 10. This paper contains a very thoughtful analysis of the causes of consumer bankruptcy and how different views about them affect policy preferences. Return to article
5 Id. at 9. Return to article
6 See Flynn, Ed, and Bermant, Gordon, "On the Evidence of These Numbers: Why Consumers File for Bankruptcy," ABI Journal, April 2000, p. 22, for an earlier discussion of the significance of this graph. Return to article
7 Buckley, F.H., and Brinig, Margaret F., "The Bankruptcy Puzzle," 27 J. Legal Stud. 187, 201-202, 205 (1998). See Kowalewski, supra n. 4, for a review of the different positions taken on this topic. Return to article
8 Buckley and Brinig, supra n. 7. Return to article