skip navigation

Monday, September 15, 2014

[ – ] Text Size [ + ]  |  Print Page

Update Newsletter: Winter 2007

Where Is the Missing Credit Card Debt? Clues and Implications

The Payment Cards Center makes information on relevant industry statistics available to researchers and individuals interested in the payment card industry, including the Tools for Researchers section of the Center’s website. Tools for Researchers includes links to a Payments Bibliography, a payment cards Data Dictionary, and Articles on Consumer Credit and Payment Statistics. As the several papers included in the articles section emphasize, interpreting many of the commonly cited public reports on payment card data can be difficult because there are often multiple sources for seemingly similar statistics. Moreover, definitions and calculations of statistics are often difficult to find and interpret. In an effort to confirm details about definitions and calculations related to apparent consumer underreporting of credit card borrowing in the Federal Reserve Board’s Survey of Consumer Finances (SCF), External Link Payment Cards Center Visiting Scholar Jonathan Zinman1 worked with the Center's research assistant Christopher Ody to examine the leading measurements of household credit card borrowing in his recent paper “Where Is the Missing Credit Card Debt? Clues and Implications.” 2

Analyzing the Federal Reserve Board’s Statistical Release on Consumer Credit External Link (known as the G.19) and the SCF, Zinman reasons that underreporting of household debt in the SCF may be less severe than previously believed. Making an initial comparison between the SCF’s measure of revolving credit card debt and the G.19’s measure of bank-reported revolving credit outstanding, Zinman notes that SCF households appear to underreport credit card debt by a factor of three. Looking more closely at what each data source measures, Zinman observes that since the G.19 data represent balance-sheet reports of outstanding debt owed to issuers, it is generally assumed to be an accurate measure. On the other hand, the SCF data are based on individuals’ responses to a wide range of questions dealing with household finances. Consequently, it is not surprising that there would be differences between the two data sources. Zinman goes on to determine how much of the gap can be explained by methodological differences and then evaluates the usefulness of SCF data on credit card borrowing for research purposes.

Jonathan Zinman, Assistant Professor of Economics, Dartmouth CollegeJonathan Zinman, Assistant Professor of Economics, Dartmouth College

Zinman finds that approximately one-third of the difference in the figures on outstanding debt is due to definitional distinctions. He notes that while the G.19 includes float, business use of personal cards, and noncredit-card lines of credit in its measure of outstanding debt, the SCF excludes these categories in its survey questions. When these definitional differences are accounted for, Zinman estimates that the underreporting of credit card balances in the SCF is reduced to a factor of two. Similarly, Zinman finds a disparity between the number of accounts reported by consumers in the SCF and the number of accounts reported by issuers in the G.19. He concludes that while “households report substantially fewer total accounts than issuers do, the gap narrows to a factor of less than 1.5 if one compares household reports to the number of active accounts in industry data.”

In assessing the usefulness of SCF credit card data to researchers, Zinman finds that the adjusted SCF underreporting factor has held steady over a 15-year period (from 1989 to 2004), despite a significant increase in consumers’ use of credit cards. During this period, 26 million new households entered the general-purpose credit card market and the proportion of overall credit card debt held on general-purpose cards rose from 74 percent to 88 percent. Despite these changes in consumers’ use of credit cards, his analysis indicates that adjusted underreporting remained stable at a factor of two.

Finally, Zinman draws on the behavioral economics literature to suggest possible explanations for why households underreport credit card balances. These include the notion of a stigma associated with reporting a high amount of debt, effects of “survey fatigue,” normal forgetfulness, survey respondents’ limited attention spans, and respondents’ failure to consider other family-held credit cards in the household. Zinman points out that, over time, important heterogeneity at the micro level might potentially be obscured.

Since credit cards are plausibly the marginal source of borrowing and consumption for many U.S. households, having accurate macro- and microeconomic data on household credit card use is particularly valuable to researchers and policymakers. By comparing existing data from two commonly used sources, Zinman reaches a positive finding: that households likely underreport their credit card data by a factor of two, not three as previously believed, and that this finding also appears relatively constant over time. In Zinman’s words, “The results offer some reassurance for researchers interested in using micro data to study the role of credit card use in households’ finances and their financial conditions.”