Private Information and Price Regulation in the US Credit Card Market (2017). Job Market Paper.
Abstract: The 2009 CARD Act limited credit card lenders' ability to raise borrowers' interest rates on the basis of information learned during lending relationships. This paper estimates the efficiency and distributional effects of these restrictions using account-level data from a near-universe of US credit cards. I show that two forces drove these restrictions' effects. First, I show that the Act constrained lenders from adjusting interest rates after learning new information about default risk, which I find exacerbated adverse selection among existing borrowers and caused (partial) market unraveling on new accounts. Second, the Act constrained lenders from adjusting rates in response to new information about demand, which reduced lender rents from inelastic borrowers. To study the net effect of these two forces, I estimate a model that features time-varying consumer risk, flexible correlation between risk and demand characteristics, and differentiated lenders who acquire private information about borrowers over time. When I impose the CARD Act's pricing restrictions in the model, I find that equilibrium market unraveling is most severe for subprime consumers, but lower lender rents are important throughout the market, so that on net, the Act's restrictions allow consumers of all credit scores to capture higher surplus on average. Total surplus inclusive of firm profits rises among prime consumers, whereas gains in subprime consumer surplus are greatest among borrowers who were recently prime.
Credit Reports as Résumés: The Incidence of Pre-Employment Credit Screening (2017), with Alexander W. Bartik. Submitted.
Abstract: We study recent bans on employers' use of credit reports to screen job applicants - a practice that has been popular among employers, but controversial for its perceived disparate impact on racial minorities. Exploiting geographic, temporal, and job-level variation in which workers are covered by these bans, we analyze these bans' effects in two datasets: the panel dimension of the Current Population Survey (CPS); and data aggregated from state unemployment insurance records. We find that the bans reduced job-finding rates for blacks by 7 to 16 percent, and increased subsequent separation rates for black new hires by 3 percentage points. Results for Hispanics and whites are less conclusive. We interpret these findings in a statistical discrimination model in which credit report data, moreso for blacks than for other groups, send a high-precision signal relative to the precision of employers' priors.
Research in Progress
A Bankrupt System? Debt Collection Lawsuits and Wage Garnishment, with Hoai-Luu Nguyen.
Abstract: While there is an extensive literature on consumer bankruptcy, comparatively little is known about bankruptcy's close substitute, informal default: debtors can seek to avoid repayment by dodging collection efforts and forcing lenders to pursue a debt in court. We study informal default by conducting a novel, large-scale data collection of debt-related lawsuits from county and state courts, and by then joining these data with county-level credit bureau data nationwide. We ask three closely related questions. First, how common is informal default, and what drives debtor substitution between it and formal bankruptcy? Second, after informal default, which creditors rely on the court system to enforce repayment, and which consumers are most likely to be sued? Finally, does the local court system efficiently allocate forced repayment – through means such as wage garnishment – to debtors who are most able to repay, or is repayment instead allocated to consumers least likely to have a lawyer or to defend themselves in court?
Tax Refund Expectations and Financial Behavior, with Sydnee Caldwell and Daniel Waldinger.
Abstract: Many economic models assume that consumption and savings decisions today depend on beliefs about future income. In this project we combine administrative (tax, credit report) and survey (durable consumption) data with novel survey information on filers’ expectations about the size of their tax refunds. We first document that while many households in our sample of low-income filers have correct mean expectations about their refunds, there is substantial uncertainty, particularly among EITC-eligible tax filers. Furthermore, the accuracy of expectations closely tracks self-reported uncertainty, as measured through probabilistic survey questions (Manski 2004). We then study how consumption and savings behavior responds to surprises in the size of tax refunds and to the resolution of uncertainty about tax refunds during tax filing. Revolving debt balances at the time of tax filing are on average roughly equal to refund size, and tax filers who expect larger refunds are more likely to use their refund to repay debt. Positive surprises about tax refunds affect self-reported savings rather than durable purchases. Households with more uncertainty resolved at the time of tax filing are on average equally likely to repay debt, save, or make durable purchases, suggesting little to no precautionary behavior.