Mr. Chairman and Distinguished Members;
In my testimony today, I will explore the extent to which household medical debts can be held responsible for consumer bankruptcy filings. This is an issue of tremendous concern not only for American families battling illnesses and injuries, but for policymakers as they attempt to reform the health care system to provide affordable and efficient care to patients. In recent times, the debate surrounding the topic has become particularly heated with studies claiming that more than 50 percent of all personal bankruptcy filings are caused by rising medical debt. This is obviously an emotional issue and anecdotal evidence of the hardship suffered by families struggling with medical bills and loss of jobs is hard to ignore. While sympathetic to the plight of these families in tough economic times, I believe that to positively inform and steer the debate, we need to disentangle the rhetoric from the facts. My own analysis of micro data from nationally representative datasets covering thousands of American families over several years has led me to conclude that the extent of the problem is being overstated and therefore misdiagnosed. A flawed understanding of the problem will inevitably lead us to the wrong solution.
The American Bankruptcy Institute provides statistics on consumer bankruptcy filings for the U.S. for several years. The data show a rise in filings from about 1.2 million in 2000 to 2.0 million in 2005. In 2006, filings dipped to 597,000 presumably due to the Bankruptcy Abuse Prevention and Consumer Protection Act of 2005 which instituted a means-test provision by which only low income filers could file for bankruptcy and discharge their (unsecured) debts. In 2008, bankruptcy filings have again crossed a million.
Data on medical debts is available from the Survey of Consumer Finances (SCF). The SCF survey samples approximately 4500 households every three years to assess families' financial situations and provides a picture of their debt and asset levels. The households are randomly selected to avoid biased results. A look at the latest SCF data (2007) shows that medical indebtedness has not changed significantly over the past decade or so. The SCF includes medical debts with other debts incurred for "goods and services", including credit card debt. These debts have risen marginally from 5.5 percent of all debt in 2001 to 5.8 percent in 2007.
The SCF shows that this change is mainly being driven by rising credit card debts where the average value has increased from $4800 to $7300 (Medical debts are excluded from the credit card debt category). Since there is no significant change in the proportion of medical debt as a fraction of all debt, it is hard to conclude that medical debts are responsible for an increasingly large fraction of bankruptcy filings. A paper by Bucks (2008) analyzing the SCF data for 1989-2004 shows, in fact, that the number of families reporting any medical debt has declined from 3.6 percent in 2001 to 2.8 percent in 2004. The same paper also shows that medical debts as a fraction of all debts have remained steady at 0.3 percent between 2001 and 2004.
II. What Explains Consumer Bankruptcy Filings?
There is a fairly large economics literature relating to the household personal bankruptcy decision. Very few of the papers however, find medical problems or medical debt as significant explanatory variables causing bankruptcies.
In the literature, there are two views about consumer bankruptcy filings. The adverse events view suggests that sudden shocks to income in a situation of high consumer indebtedness may provoke a bankruptcy filing. Sullivan et al (1989) conclude that the primary cause of bankruptcy filings in their sample was unemployment or employment interruptions. A divorce, also, may create an unexpected shock to household income or reduce the economies of scale from living in a single household.
The second is the strategic view. This emphasizes that households file for bankruptcy when the financial benefit from filing increases. Under Chapter 7 personal bankruptcy, debtors in the US can retain some or all of their property and shield it from being used to repay creditors at the time of a bankruptcy filing. The value of assets that they can protect depends upon the exemption level in the state of filing. The 2005 surge in bankruptcy filings in anticipation of a change in the personal bankruptcy law supports the strategic view of bankruptcy since it appeared that households were filing to take advantage of the higher exemptions associated with the older, more lenient system. Fay et al (2002) find support for the prediction that households are more likely to file when their financial benefit from filing is higher.
Medical problems can lead to bankruptcy either directly through the accumulation of medical debt or indirectly through loss of work days and subsequent loss in earnings. There is little evidence in the literature to support the indirect effect of medical problems on bankruptcies. In the paper by Fay et al. (2002), health problems faced by the household head or spouse, spells of unemployment, and the household head being divorced in the previous year were not significant factors in explaining bankruptcy filings. A study by Ian Domowitz and Robert Sartain, also finds little correlation of medical debt with other sources of financial distress, such as job loss or income interruption. In my own research, I find no evidence that the poor health of the household head could significantly raise the probability of a bankruptcy filing. A possible reason for this is that for the majority of cases, a substantial level of these indirect effects eventually should show up also as significant medical debts. To the extent that job losses occur as a result of a medical emergency, they can be considered to be the proximate cause of a bankruptcy filing only if the filing happens within a short period of time after the medical emergency. Job losses or lost weeks of work due to illness that occurred almost a year prior to the filing are unlikely to be significant predictors of bankruptcy filings since there may have been several other factors in the intermediate period that could be held responsible for the filing. Moreover, because we ultimately care most about the effect that rising health care costs have on families, it remains hard to justify how bankruptcies due to illnesses that lead to job losses are a direct consequence of rising health care costs alone.
In the next section, we therefore focus on the direct effect of medical bills on bankruptcy filings.
III. Medical Debts and Bankruptcy Filings
A. Evidence from Surveys
The evidence on the extent of medical debts in bankruptcy filings is fairly mixed. Most studies so far have relied on surveys of households or bankruptcy filers. While surveys provide relevant information, their results are not necessarily conclusive since they can be driven by the choice of sample ("sample selection" issues) such as whether the respondents involve bankruptcy filers or general households, whether respondents are married and their income, wealth and earnings status, as well as the sample size and the framing of questions which could bias the results. These are discussed in a separate section with reference to the Himmelstein et al. (2005 and 2009) studies. In this section, we show how these issues lead to inconclusive estimates of medical bankruptcies based on survey data.
For instance, relying on surveys of 1032 bankruptcy filers, Himmelstein et al. (2009) conclude that approximately 62 percent of all bankruptcies in 2007 were "medical." Their earlier study (Himmelstein et al. (2005)), based on a 2001 survey of 1000 filers, concluded that approximately 46 percent of all bankruptcies had medical causes. Note that in both studies, "medical" refers to all sorts of medical reasons for a bankruptcy filing, not just medical debts. These include lost weeks of work due to own illness or spouse's illness, as well as when the debtor said that a medical problem of a family member caused the bankruptcy filing. As mentioned earlier, lost weeks of work due to illness and other medical problems cannot be blamed on rising healthcare costs. Therefore, the number that we really care about is the number of bankruptcies actually caused by medical bills. The authors claim that this number is approximately 57 percent. As we discuss in the next section that critiques the methods used in the two studies, even this number is overstated. Moreover, this number is in stark contrast to findings reported by other surveys that rely on a larger sample size and a stricter definition of a medical bankruptcy.
The closest comparable survey to this is a study of bankruptcy filers by the Department of Justice's Executive Office of the United States Trustee (USTP). The USTP examined the records of 5,203 bankruptcy cases filed between 2000 and 2002, the most thorough study of the problem to date of those who actually filed bankruptcy. It reported that 54 percent of the cases in the sample listed no medical debt, meaning that the median amount of medical debt in the study was zero. Medical debt accounted for 5.5 percent of total general unsecured debt and 90.1 percent of filers reported medical debts less than $5,000. There were a few cases where extremely high medical debt likely explained the subsequent filing—one percent of cases accounted for 36.5% of medical debt and less than 10 percent of all cases represented 80% of all reported medical debt. Of the minority (46%) of cases in the sample with medical debt, the average medical debt was $4,978 per case, 78.4 percent of those cases reported medical debts below $5,000 (an average of $1,212 for this group), and medical debts accounted for 13.0 percent of the total general unsecured debt for those reporting medical debt. Thus, even among those who reported medical debt, few reported medical debt levels sufficiently high to conclude that they were a primary cause of bankruptcy.
A more nationally representative survey is the Panel Study of Income Dynamics (PSID), which is a longitudinal survey tracking households since 1968. In 1996, the PSID asked respondents whether they had ever filed for bankruptcy between 1996 and 1984, and if so, what were the primary, secondary and tertiary reasons for filing from a given a list of possible reasons, which included medical bills, job loss, injury or illness, etc. This is the most definitive survey so far in terms of determining the proximate cause of a bankruptcy filing. The largest contributor to bankruptcy filings was high credit card debt. Nearly 42 percent of respondents reported high credit card bills as the primary reason for filing, while an additional 9 percent claimed it as the secondary reason for filing. Other big reasons were job loss (13 percent) and divorce or separation from spouse (12 percent). Only 9 percent of the sample claimed medical bills as the primary reason for filing, and 7 percent claimed it as a secondary reason. Illness and Injury accounted for only 6 percent of the filings. Therefore medical bills can be ascribed as the primary cause of the filing in only 9 percent of the cases. The number rises to 16 percent if we include all filings that claim medical bills as at least one of the causes of the filing. The most important cause appears to be credit card debt accounting for more than 50 percent of all filings.
Other surveys focus on a wider sample of all households rather than only households that filed for bankruptcy. For instance, a 2008 Report by the Center for Studying Health System Change shows that about 15 percent of families in 2003 and 19 percent in 2007 had problems paying medical bills. Of these, only about 20 percent actually considered filing for bankruptcy and an even smaller fraction actually did file for bankruptcy. A Kaiser Family Foundation Health Tracking Poll conducted in June 2009 found that 26 percent of respondents claimed that they had had problems paying medical bills in the previous year.
B. Evidence Using Regression Analysis
By their very nature, survey data are unable to account for a host of other factors that might help explain why households file for bankruptcy. For instance, factors like average household wealth and income, state-level factors such as bankruptcy exemptions and unemployment rates, and household expenditures such as rent and taxes could each play a significant role in a household's decision to file for bankruptcy. The standard methodology in the economics literature for accounting for all of these factors is multivariate regression analysis. With regression analysis, it is possible to study the effect that each factor has on the probability of filing for bankruptcy while holding the effect of all other variables constant. This is the only way that one can establish causation, rather than correlation. In other words, only when we use regression analysis to control for the effect that each of the other factors has on a bankruptcy filing can we be sure that medical debts cause bankruptcy filings.
A 1999 study by Ian Domowitz and Robert Sartain in the Journal of Finance uses exactly this approach. The authors examined 827 households who filed for bankruptcy in 1980 matched against 1,862 households not in bankruptcy. Accounting for prevalence of various sources of debt, Domowitz and Sartain found that "the largest single contribution to bankruptcy at the margin is credit card debt."
In an AEI Working paper that I wrote, I estimated a model of the household bankruptcy filing decision, using PSID data for the period 1994-1996 and a three year panel covering the years 1984, 1989 and 1994. The main aim in the paper was to test whether medical debts can be ascribed as the leading cause of bankruptcy filings. To this end, I first developed a classification of households into medical and other debtors. Then I regressed the probability of bankruptcy on medical (and other) debts using a probit model and a hazard model.
The results from my paper do not support the view that medical debts are the leading cause of bankruptcy filings. In fact, households who are most likely to file are those with primarily other forms of debt, such as credit card or car debts, who also incur medical debts. Altogether, a 10 percent increase in debts (as a fraction of income) of households with primarily credit card debts would cause bankruptcy filings to go up by 36 percent on average. A 10 percent increase in debts (as a fraction of income) of households with primarily medical debts would cause filings to go up by 27 percent on average.
What does this imply for current estimates of medical bankruptcies? The SCF shows that for all families, total debts as a fraction of income have increased by about 10 percent over the period 2001-2007. Therefore, even if all of this increase was due to medical debts (which is not likely since credit card debts have been rising faster over this period), the additional bankruptcy filings due to medical debts should be approximately 27 percent of the total.
IV. Problems with the Himmelstein et al. (2005 and 2009) Studies
(1) Sample Selection Issues
A major shortcoming with both the Himmelstein et al. (2005 and 2009) studies is what economists dub the "sample selection issue". Himmelstein et al. (2005, 2009) conducted a survey of bankruptcy filers from public court records for the year 2001 and 2007. Based on a sample of 1000 debtors, they concluded that more than 50 percent of these had filed for bankruptcy due to a medical reason. By limiting the sample to those who had already filed for bankruptcy, the study overstated the incidence of medical debt. To account for causation, the study sample should have, at the very least, included a "control" group of medical debtors who did not file for bankruptcy. In other words, if the authors were trying to establish whether medical debts cause bankruptcy filings, the appropriate sample should have included households with and without medical debt, and households who filed or did not file for bankruptcy. In short, what the authors have established is some correlation, but not causation.
The sample also seems skewed towards debtors with high medical debt. The USTP report of bankruptcy filers, which included a much larger sample of 5203 filers, found that 90 percent of filers had medical debts less than $5000. The Himmelstein et al.(2009) study reports nearly 35 percent of filers with more than $5000 in medical debt. The authors make no attempt to reconcile or explain their findings or reveal the distribution of medical debts across filers in their sample.
(2) Regression Analysis
The study also should have allowed for the possibility that other household characteristics, such as the filer's work status, marital status, income, and other kinds of debts could have influenced the filing. As explained earlier, this could be done through the use of appropriate regression techniques applied on a suitably large, random sample of filers and non-filers. Mainstream economics literature discussing the relationship between debts and bankruptcy amply outlines these standard considerations. The study does claim to have done multivariate analysis, but the analysis is done on an even more restricted sample than the original 1032 in 2007. The sample only includes people who reported having any medical bills. Therefore, it simply assumes that medical debts are important for bankruptcy filing, rather than testing for that hypothesis in the entire sample of bankruptcy filers.
(3) Definition of Medical Bankruptcy
The 2005 study used an overly broad definition of "medical filers," which included people with any sort of addiction or uncontrolled gambling problems. The 2009 study removed these clauses but still came up with a 62 percent number i.e nearly 62 percent of bankruptcy filings are due to medical reasons. The reason for the high number is puzzling, though as mentioned earlier, it is partly driven by the fact that the authors ascribe any remotely medical factor as causing the bankruptcy filing, not just medical debts. The survey results shown in Table 2 (Page 3) of the study clearly state that only 29 percent of the respondents believed that their bankruptcy was actually caused by medical bills. However, the authors chose to add to this number the percent of people who lost weeks of work due to illness, the percent of people with more than $5000 in medical bills, and the percent of people reporting any medical problems. This is clearly an overstatement of the problem. Since the respondents themselves do not believe that these other factors caused the bankruptcy filing, it is wrong to ascribe the additional bankruptcy filings to their medical costs. A related point is that the survey fails to provide information on other causes of the bankruptcy filing or how the respondents would rank different factors, as in the PSID. Therefore, it is unclear whether medical bills were the most important cause or just another cause.
This criticism was also raised by Dranove and Millenson in reference to the 2005 paper. Exhibit 2 of that paper identified people who stated that illness or injury was a cause of bankruptcy (although not necessarily the most important cause). According to Himmelstein and colleagues, 28.3 percent of respondents stated that illness or injury was a cause of bankruptcy. They also reported that medical bills contributed to the bankruptcy of 60 percent of this group. Multiplying the two figures together, Dranove and Millenson conclude that 17 percent of their sample had medical expenditure bankruptcies. Even for that 17 percent, it cannot be stated with any degree of certainty whether medical spending was the most important cause of bankruptcy.
To summarize, data from surveys, including the Himmelstein et al. studies, would suggest that by the respondents' own estimates, the fraction of bankruptcies caused by medical debts ranges from around 16 to 29 percent. The upper bound may be an overestimate since the respondents in the Himmelstein et al. survey also do not specify whether medical bills were the immediate cause or the most important cause of the filing. The only survey that asks the right questions is the PSID, which estimates that between 1984 and 1996, an average of about 16 percent of filings were due to medical bills. Given that "goods and services" debt, which includes medical debt, as a fraction of all debts has actually declined between 1998 and 2007 from 6 percent to 5.8 percent of all debt (SCF, 2007), it is hard to imagine that medical bankruptcies have increased tremendously over this period.
The economics literature using standard regression analysis to account for household and macroeconomic conditions has generally found that medical debts are not the most important cause of bankruptcy filings. Other factors, such as the financial benefit from a bankruptcy filing or other types of debt, such as large credit card debts, are more important predictors of bankruptcy filings.
Having said that, however, I do not wish to underestimate the serious effects of medical problems on particular families. Rising health care costs are clearly an area of growing concern, and there is an urgent need to tackle the issue. At the same time, we should recognize that families are being pushed to the brink of bankruptcy for a multitude of reasons. While some recent reports, based on thinner evidence and less robust methods, would have us believe that rising health care costs are in fact the main factor responsible for household bankruptcies, I hope my testimony has provided a more substantial basis for concluding that this is not the case.
I would like to end by stating that in order to find a solution we first need a correct understanding of the problem. The most effective solution to the problem of rising bankruptcies in these tough economic times is to help families keep their jobs, retain their earning power, stay in their homes, and live within their means. If economic problems nevertheless become unmanageable, the bankruptcy system is designed precisely to give families a fresh start by discharging some of their debt. If we mistakenly focus too narrowly and simply on medical indebtedness, believing it to be a bigger problem than it is, we will be even further away from the solution we need.
Aparna Mathur is a research fellow at AEI.
 Fay, Scott, Hurst, Erik and White, Michelle (2002), "The Household Bankruptcy Decision", American Economic Review, Vol. 92 No.3, June 2002, 706-718
 Domowitz, Ian and Sartain, Robert (1999), "Determinants of the Consumer Bankruptcy Decision", Journal of Finance, Vol. 54, No.1, Feb 1999, 403-420
 Himmelstein, David, Warren, Elizabeth, Thorne, Deborah and Woolhandler, Steffie (2009), "Medical Bankruptcy in the United States, 2007: Results of a National Study", The American Journal of Medicine, available at: http://pnhp.org/new_bankruptcy_study/Bankruptcy-2009.pdf
 Himmelstein, David, Warren, Elizabeth, Thorne, Deborah and Woolhandler, Steffie (2005), "Illness and Injury as Contributors to Bankruptcy", Health Affairs (Web Exclusive), 2 February
 Mathur, Aparna (2006), "Medical Bills and Bankruptcy Filings," AEI Working Paper http://www.aei.org/paper/24680
 Dranove, David and Millenson, Michael, L. (2006), "Medical Bankruptcy: Myth vs Fact" Health Affairs 74 (2006)