Lying About Borrowing

We compare survey self-reports with administrative data and find that nearly 50% of recent borrowers do not report their high-interest consumer loans. Under-reporting appears to be correlated with several characteristics of interest, in particular gender: 62% of women interviewed by men underreport whereas 42% of women interviewed by women underreport. On the other hand, 40% of men underreport, irrespective of the gender of the interviewer. As such relying strictly on self-reported data may lead to biased inference, and we outline some methodological implications for identifying impacts of credit access on borrower behavior and outcomes. Matching female surveyors to female respondents appears to be a low-cost mitigating strategy, but clearly the best strategy is to make sure one has administrative data from a lender to measure actual borrowing history.