Has Growing Inequality Contributed to Rising Household Economic Distress?

Abstract

The personal bankruptcy rate increased more than fourfold in the last quarter century. Other measures of economic distress, particularly foreclosure and credit default rates, also increased sharply. A possible explanation for this is greater household indebtedness. Household debt relative to income, however, did not even double over the same period, suggesting that the aggregate increase in household economic distress was disproportionate to the rise in household debt. We consider if the simultaneous increase in income inequality has contributed to the rise in household economic distress, Specifically, we hypothesize that greater inequality led to a larger expansion of credit, especially in the form of credit card debt, among low and moderate income households than among higher income ones. This expansion of disproportionately more expensive credit may have contributed to the growth in household economic distress. Based on data from 1980 to 2004, we find robust evidence for a link between inequality and credit card debt and between credit card debt and economic distress.

Notes

¹We end all our time series in 2004. Data on household economic stress for 2005 and thereafter are distorted due to the effects of new bankruptcy legislation.

²See the Appendix for a list of variables, definitions and sources.

³These figures refer to the outstanding amount owed on credit cards, including balances that are carried forward and those that are paid off. Both types of balances constitute credit, although with different maturities. The arguments in this paper pertain to all credit card balances since balances that are typically paid off could become balances carried forward if actual income gains fall below expected income gains.

⁴A rising share of consumption was also financed by fewer savings. The personal saving rate declined from an average high of 9.7% in the early 1970s to 2.2% in the latest business cycle (Table 2).

⁵The arguments here focus on consumer credit, especially on credit card debt. They are less relevant for other forms of credit, such as mortgages or student loans.

⁶Credit card debt in Table 4 refers to balances carried forward since the Survey of Consumer Finances does not ask about balances regularly paid off and thus is not comparable to the aggregate figures from the Flow of Funds (Board of Governors, 2006a). The data here are provided as an approximation of the distribution of all credit card debt by income.

⁷Authors' calculations based on American Bankruptcy Institute (2005), US Census Bureau (2006b), Board of Governors (2006c), and Mortgage Bankers Association (2005).

⁸Formally, Atkinson = 1–(n/l)[E(X _i /u) ^l –e] ^l /(l–e), where X _i is the weighted income of individual i and l is the mean income of the sample population. The e parameter reflects the aversion to inequality and lies between 0 and 1. A lower parameter value of e – less aversion – means that the index becomes more sensitive to changes at the upper end of the income scale and vice versa. In effect, the index calculates something also referred to as equity sensitive average income. It is defined as that level of per capita income which, if enjoyed by everybody, would make total welfare exactly equal to total welfare generated by the actual income distribution. The interpretation is, for instance, for e = 0.2, that with an equal income distribution, it would take only 80% of actual income to achieve the same level of welfare.

⁹All series are non–stationary, integrated at the first degree, and cointegrated at the 1% level.

¹⁰All estimations are done using a Prais–Winsten regression to correct for serial correlation. This technique has the advantage over other methods that correct for serial correlation of preserving the number of observations.

¹¹Credit card debt refers to revolving consumer credit (Board of Governors, 2006e).

¹²Our previous results on debt levels are unaffected by the choice of time period.

¹³If credit card balances that are carried forward and those that are regularly paid off expand at the same rate, the estimated aggregate effect on household economic distress should be unaffected by our use of a composite measure of credit card balances.