Unobserved heterogeneity in borrower underreporting would complicate inference on relationships between credit card use and other behaviors, such as financial distress, consumption paths, or portfolio choice. This paper offers some reassurance and several new stylized facts. Accounting for differences in definitions between household and industry measures reduces debt underreporting to a factor of two. Underreporting is less severe for general-purpose than for other cards. The true underreporting factor has remained stable over 15 years even as 26 million households entered the market. Households report charges and account holding relatively accurately.

View the Full Discussion Paper