Consumer Filings Trends and Indicators Part II
This part of the article examines the correlation of the following economic indicators to the non-business bankruptcy filing rate (per 1,000 population). I have chosen various sets of economic data that might plausibly be thought to be correlated to the rate of consumer bankruptcy filings. Where applicable, I have adjusted for inflation, using the CPI (Consumer Price Index), 1982-1984 = 100, and have adjusted for population growth, calculating pertinent data per 1,000 population. The data considered are:
1. total consumer credit outstanding (Charts 8 and 9);2
2. revolving consumer credit (Charts 10 and 11);3
3. nonrevolving consumer credit (Chart 12);4
4. revolving credit as a percentage of total consumer credit (Chart 13);
5. personal, non-farm mortgage debt (or "home" mortgages) (Charts 14 and 15);5
6. sum of total consumer credit and home mortgage debt (Charts 16 and 17);
7. ratio of total consumer credit to personal income (Chart 18);6
8. ratio of "total debt" (i.e., total consumer credit + home mortgage debt) to personal income (Chart 19);
9. credit card charge-off rates (Chart 20 and 21);7 and
10. credit card delinquency rates (Chart 22 and 23).8
Table 1 (p. 103) contains the statistical data for these 10 independent variables, with the nonbusiness bankruptcy filing rate as the dependent variable. In addition, one-year and two-year time lags from the independent to the dependent variable are calculated for several of the categories—total consumer credit, revolving consumer credit, home mortgage debt and total debt. Testing at a 95 percent confidence level, statistically significant correlations to the nonbusiness bankruptcy filing rate were found for every independent variable except for credit card delinquencies. Correlation coefficients of greater than 90 percent were found for six of the independent variables—total consumer credit, revolving consumer credit, revolving credit as a percentage of total credit, home mortgage debt, total debt and the ratio of total debt to personal income. Revolving consumer credit showed a correlation coefficient of .974. It is important to bear in mind that these data shows only that these economic data sets likely are correlated to nonbusiness bankruptcy filings; they do not mean that these factors cause consumer bankruptcy.
Table 2 (p. 104) sets out the dollar amount for consumer credit (total, revolving and nonrevolving), which is shown visually in Charts 8-13. Recall from Table 1 discussed above that each of these three independent variables correlates with statistical significance to the nonbusiness bankruptcy filing rate. For total consumer credit, the correlation is especially close when a two-year time lag (Chart 9) is used. This is partially consistent with the findings of Prof. Robert Lawless in a forthcoming paper, "The Paradox of Consumer Credit." Prof. Lawless, examining filing rates from 1933 to 2004, found that "[i]n the long run, increasing consumer credit leads to increased bankruptcy filings."9 However, he also found that in the short term, increased consumer credit actually led to a decrease in bankruptcy filings (thus the "paradox"). This paper did not find a similar negative short-term correlation for the period from 1980-2006, but instead found that increased total consumer credit correlated with increased nonbusiness bankruptcy filings with no time lag, a one-year time lag and a two-year time lag.
How does the picture look when the two components of total consumer credit (revolving and nonrevolving credit) are considered separately and compared to the non-business bankruptcy filing rate? Consider revolving credit (typically credit card debt) first.
Chart 10 reveals an extremely close relationship between the amount of revolving consumer credit and the nonbusiness bankruptcy filing rate for the past quarter century. Table 1 indicates that this visual picture is not misleading: Revolving consumer credit has a correlation coefficient with nonbusiness bankruptcy filings of more than 97 percent, and correlates with a P value of less than 0.01. If a time lag (two years) between the revolving credit and the nonbusiness filings is used, a similar very close relationship exists. Chart 11 presents that data.
Given that revolving credit is principally credit card debt, Charts 10 and 11 indicate that the enormous increase in credit card debt in the past quarter century may have helped trigger the correlative jump in, and closely predicts the rate of, nonbusiness bankruptcy filings. This conclusion suggests that changing the bankruptcy law to make it harder for consumer debtors—as Congress did in 2005—may be a fool's errand. If revolving credit continues to grow, or even to remain stable, then these data suggest that nonbusiness bankruptcy filings are likely to follow in like measure (unless BAPCPA can somehow change history). The post-BAPCPA drop in filings is likely to end, and soon, given the amount of outstanding revolving consumer credit. Thus, if Congress really wants to slow or even reverse the increase in bankruptcy filings, the real target should be the underlying cause—credit card debt—and not the bankruptcy law itself.
If one believes that the close relationship between revolving consumer credit and nonbusiness bankruptcy filings is likely to persist, the logical question then would be: What will happen to the amount of revolving credit in the future? Looking at Table 2 and Charts 10 and 11, we can see that the dollar amount of revolving consumer credit has remained fairly constant since 1999, after more than quadrupling from 1983-98. Each year since 2002, revolving credit has gone down by very slight amounts. We may be approaching a market saturation point; by this time, there may be few untapped candidates for revolving consumer credit out in the marketplace. At the very least, it seems unlikely that we ever again will see the sort of explosion in revolving credit that occurred from 1983-98—and correlatively, also unlikely that we ever again will see the sort of exponential increase in consumer bankruptcy filings that occurred during that same time period. Recall that the rate of nonbusiness filings has hardly changed at all in the past decade: In 1997, it was 4.717 per 1,000 population, and in the year ending June 2006, it was 4.872 per 1,000. The bottom line is that bankruptcy filings will return to close to pre-BAPCPA levels, and soon, but will probably stay in that range for some years to come.
How does the data for nonrevolving credit correlate with nonbusiness bankruptcy filings? According to Chart 12 and Table 1, they correlate, but not as well. The correlation coefficient is only 72 percent, but the correlation is statistically significant. Chart 12 suggests that the mapping of non-revolving credit and nonbusiness bankruptcy filings has been closer since 2001 than it was before.
Given that revolving credit is such a good indicator of filing rates, and nonrevolving credit less so, I examine whether an even better indicator would be the percentage of revolving credit of the total amount of consumer credit. As Chart 13 and Table 1 show, this percentage is correlated with statistical significance with nonbusiness bankruptcy filings, but not quite as closely as the total amount of revolving credit. So the better measure remains total revolving credit.
The debt measures considered so far exclude real estate debt. Real estate debt burdening consumers, however, far exceeds non-real estate debt (as a comparison of Tables 2 and 3 shows, by a factor of four, in fact), and thus a full consideration of the possible relationship of debt to bankruptcy should account for home mortgage debt. Table 3 includes data for personal, non-farm mortgage debt (i.e., "home" mortgages), and then adds together the data for total consumer credit (from Table 2) and home mortgage debt to get a total debt measure. As before, all data is adjusted for inflation and for population. Data for home mortgage debt only goes through 2005. The data is then compared to the nonbusiness bankruptcy filing rate in Charts 14-17. For both data sets, a zero-time-lag chart and a two-year time-lag chart is included.
According to Table 1, both of these independent variables—home mortgage debt standing alone, and home mortgage debt combined with total consumer credit—correlate positively and in a statistically significant way with the rate of nonbusiness bankruptcy filings. The correlation coefficient for home mortgage debt is 92 percent and for total debt (consumer plus mortgage) is just over 93 percent. Prior studies likewise have found mortgage debt to be correlated to bankruptcy filings.10 According to Table 1, the correlations hold whether a zero-time-lag (Charts 14 and 16), one-year time lag or two-year time lag (Charts 15 and 17) is used.
This data suggests that nonbusiness bankruptcy filings may rise to above pre-BAPCPA levels, contrary to the predictions following from a consideration of non-real estate consumer debt alone, as discussed above. Why? Table 3 shows that home mortgage debt grew by more than 50 percent from 2000-05. It will be useful for purposes of this analysis to see what happens to consumer bankruptcy filings, as the new millennium has heralded a much different debt era than the 1980s and 1990s. Where non-real estate consumer credit (and especially revolving credit) jumped up dramatically in the 1980s and 1990s (and consumer bankruptcy filings spiked in tandem), the 2000s have been a time of tremendous growth in home mortgage debt, but of slight decreases in consumer credit. Indeed, home mortgage debt grew slightly more from 2000-05 than it did from 1980-2000! Note from Table 1, though, that the regression coefficient for home mortgage debt as an independent variable is fairly small (0.494418), so the nonbusiness bankruptcy filing rate increase may be relatively modest as well.
Charts 16 and 17 present the data for total debt, viz., total consumer credit plus home mortgage debt, as compared with nonbusiness bankruptcy filings. Very similar relationships are found as pertained for home mortgage debt standing alone.
Some have posited that nonbusiness bankruptcy filings might be correlated with the debt-to-income ratio. Intuitively, this idea makes sense; as the percentage of a debtor's income that must be devoted to debt service increases, that debtor is more likely to encounter financial distress. Table 4 contains data for the percentage of total consumer credit as compared to personal income and for the same comparison using instead the total of consumer plus mortgage debt. A statistical study of filing rates from 1947 through 1990 found that "the debt-to-income ratio is significant and positive."11 Table 1 shows that both of the debt-to-income independent variables considered in this paper—total consumer credit alone, and total consumer credit plus home mortgage debt—are positively correlated with statistical significance to the nonbusiness bankruptcy filing rate. The correlation coefficient for the sum of total consumer credit and home mortgage debt is slightly higher than for total consumer credit alone (about 90 percent versus 84 percent). The data are presented visually in Charts 18 and 19.
The next data sets to consider concern credit cards. The data is the rate of credit card charge-offs and the rate of credit card delinquencies. These data are published by the Federal Reserve. They begin in 1985 for charge-off rates and in 1991 for delinquencies, and are published on a quarterly basis. I have selected the second quarter data, thus allowing consideration of the most recent 2006 data. The data is found in Table 5 and is presented visually in Charts 20-23. An earlier study by economist Lawrence M. Ausubel found a positive correlation between personal bankruptcy filings and credit card delinquencies from 1990-96.12
According to Table 1, the credit card chargeoff rate is positively correlated to the rate of nonbusiness filings, with statistical significance, but the data for the credit card delinquency rate does not demonstrate any statistical correlation. Note from Table 1 that the P value for credit card delinquencies is 0.789, which in simple terms means that any supposed connection between the independent and dependent variable is, in a word, garbage. If one did a scatter plot for delinquency rates, it would just be a big blob. Charts 20 and 21 present the data for credit card charge-off rates, first with zero time-lag (Chart 20) and then with a one-year time lag (Chart 21). The same approach is taken for the credit card delinquency rate (Chart 22—zero time lag, Chart 23—one-year time lag). Even for credit card chargeoff rates, the correlation coefficient is only 80 percent.
According to Table 5, credit card chargeoff rates have declined dramatically—from 6.0 percent in the second quarter of 2002 to just 3.48 percent in the second quarter of 2006. If the historical correlation with nonbusiness bankruptcy filing rates holds, one would expect a drop in the near term in those bankruptcy filings. However, the regression coefficient is only 0.1189406 (for a one-year time lag, the coefficient is 0.2127374). So any changes in bankruptcy filing rates would be modest, but according to these data, the rates would decline.
Charts 22 and 23 present the data on credit card delinquency rates and nonbusiness bankruptcy filing rates. Recall that Table 1 found no statistical correlation between these variables.
This article series has examined in detail the filing data and trends for nonbusiness bankruptcies in the United States, and also has analyzed numerous sets of economic data looking for correlations with the nonbusiness bankruptcy filing rate. Several conclusions may be drawn from these data. First, the huge decline in nonbusiness bankruptcy filings following the effective date of BAPCPA is likely about to end. Filing rates probably will soon return to the range of pre-BAPCPA levels. However, those filing rates are likely to remain fairly stable over the next few years, as the amount of outstanding revolving consumer credit—historically an excellent predictor of consumer filing rates—appears to have leveled off. Having said that, two bits of data point in opposite directions: the recent increases in the amount of home mortgage debt may portend possible increases in nonbusiness bankruptcy filings, but the dramatic drop in credit card chargeoff rates points the other way.
The data indicate that BAPCPA was based on a canard. It does not appear that consumers have made a quantum shift to chapter 13 from chapter 7, as Congress had hoped would happen under BAPCPA. More importantly, the data suggest that BAPCPA was predicated on (to be generous) a false hope—that making the law "tougher" would discourage consumer debtors from filing bankruptcy. The evidence shows that debtors file bankruptcy in very predictable numbers, depending not on what the bankruptcy law provides, but on how burdened they are with debt. This should not have been surprising. If Congress wants to drive down bankruptcy filings, it needs to take steps to drive down consumer debt burdens. Otherwise, bankruptcy filings will remain high. We can only hope that a decade from now lawmakers do not repeat the BAPCPA mistake when they see that it did not deliver the results intended.
1 This article would not have been possible but for the immense help and contributions of two people: Prof. Robert M. Lawless of the University of Illinois College of Law, and Rebecca M. Tabb, M.Sc. Economics candidate 2006, University College London, J.D. candidate 2009, Stanford Law School. The complete version of this article is included in the written materials for ABI's program "A Year after BAPCPA," held Oct. 16, 2006, in Washington, D.C.
2 The Federal Reserve publishes consumer credit data as Statistical Series G. 19, at www.federalreserve.gov/releases/g19/hist/ (last visited Sept. 8, 2006). The Fed in Series G. 19 states that "Consumer Credit Outstanding" "covers most short- and intermediate-term credit extended to individuals, excluding loans secured by real estate." See www.federalreserve.gov/releases/g19/current/ at note 1 (last visited Sept. 8, 2006). It is the sum of revolving and nonrevolving consumer credit.
3 See note 2, supra, for the G. 19 statistical reference. "Revolving" credit is treated by the Fed as synonymous with "open-end credit," which it defines as: "A line of credit that may be used repeatedly up to a certain limit (also called a charge account or revolving credit)." See www.frbsf.org/tools/glossary/glossReg.html#R. Credit cards are the main type of revolving credit used by consumers.
4 See note 2, supra, for the G. 19 statistical reference. According to the Fed, nonrevolving credit "includes automobile loans and all other loans not included in revolving credit, such as loans for mobile homes, education, boats, trailers or vacations. These loans may be secured or unsecured." See www.federalreserve.gov/releases/g19/current/ at note 5 (last visited Sept. 8, 2006).
5 Mortgage debt data are found in the Federal Reserve's Flow of Funds Accounts of the United States: Annual Flows and Outstandings; the most recent data are on page 85 of the Report found at www.federalreserve.gov/Releases/z1/Current/annuals/a1995-2005.pdf (last visited Sept. 8, 2006).
6 Consumer debt = total consumer credit outstanding (see n. 2, supra). The data for personal income is published by the Bureau of Economic Analysis at www.bea.gov/bea/dn/home/personalincome.htm (last visited Sept. 8, 2006).
7 Credit card charge-off rates are published by the Federal Reserve (starting with 1985) and can be found at www.federalreserve.gov/ releases/chargeoff/default.htm (last visited Sept. 8, 2006).
8 See id. for credit card delinquency rates. The Fed's statistics begin in 1991.
9 Lawless, Robert M., "The Paradox of Consumer Credit," 2006 U. Ill. L. Rev. at 24 (forthcoming).
10 See Lawless, Robert M., "The Paradox of Consumer Credit" at 2, 5-6, 20-21, 2006 U. Ill. L. Rev. (forthcoming), and studies cited therein.
11 Bhandari, Jagdeep S. and Weiss, Lawrence A., "The Increasing Bankruptcy Filing Rate: An Historical Analysis," 67 Am. Bankr. L.J. 1, 7 (1993).
12 Ausubel, Lawrence M., "Credit Card Defaults, Credit Card Profits, and Bankruptcy," 71 Am. Bankr. L.J. 249, 253-54 and Figure 2 p. 253 (1997).