Working Papers
Job Market Paper
Abstract: I show that price regulation and increased oversight can unintentionally reduce credit access and raise prices by driving out lenders with advanced risk-screening models. These lenders concentrate in less-creditworthy market segments, where their superior information helps overcome adverse selection. This makes them both the best positioned to serve borrowers with lower credit scores and the most impacted by regulatory intervention. To show this, I study the sudden enforcement of rate caps and oversight in the U.S. nonbank personal loan market. Using a difference-in-differences design around state regulation, I show that rate caps reduce credit by 10%, increase prices by 8%, and result in 21% of lenders exiting the market. I develop a structural model that separates the effects of interest rate caps from the fixed costs of regulatory oversight. The model, incorporating borrower heterogeneity, screening technologies, and adverse selection, shows that these regulations hurt low-risk borrowers who appear risky on paper. Counterfactuals suggest that raising the cap from 21% to 28% and cutting fixed regulatory costs by 45% could improve credit access for subprime borrowers.
Publications
with Erica Jiang, Gregor Matvos, Tomasz Piskorski and Amit Seru (01/2022) [Link to Publication]
AEA Papers and Proceedings, 2022
Press: Insights by Stanford Business [Link]
Works-in-Progress
with Zunda Winston Xu
What drives state regulation of nonbanks? We examine the factors influencing state-level regulation of nonbank financial institutions. As nonbanks—particularly fintech lenders—capture a growing share of the consumer credit market, states have responded with diverse regulatory approaches. To analyze these responses, we construct one of the first comprehensive databases of enforcement actions against nonbanks across 47 states over 20 years. Leveraging this dataset, we find no evidence that state regulations are primarily motivated by consumer protection or aiding subprime borrowers. Instead, we find preliminary evidence suggesting that state regulators may target nonbanks to “protect” state-chartered banks from competition. To explore this further, we investigate the “demand” side, where state-chartered banks lobby, file complaints, or make political donations to reduce competitive pressure, and the “supply” side, where state regulators respond due to incentives like job prospects or financial repression. Regulators may also aim to maintain financial stability, concerned that excessive competition could lead state-chartered banks to take on riskier behaviors.
Credit scores are the public signal that prices most U.S. consumer lending, so a rule that changes what they report reshapes market equilibrium. The CARES Act required creditors granting forbearance to report missed payments as current, suppressing negative information on tens of millions of accounts. Using a 10% tradeline panel and a shift-share instrument built from 2019 mortgage-servicer composition, I show the suppression degraded the signal rather than improving finances: at the same score, forbearance borrowers default 1.5 percentage points above their score-implied risk. Because one score prices many markets, the degraded signal repriced products the policy never covered. Personal-loan APRs rose by roughly 0.9 percentage points, concentrated in the near-prime range. The burden falls on borrowers who never received forbearance, including those without a mortgage of their own: at the same observed score they pay higher rates, subsidizing recipients whose inflated scores pool with theirs. Reporting-rule design is thus a policy lever. An information-preserving alternative that flags forbearance while leaving the record intact already existed in the disaster-relief system and was declined. A model of credit-market equilibrium puts the cross-subsidy at $4.5~billion from accurate-history borrowers to recipients, almost entirely in the uncovered markets, and against a far smaller efficiency loss; a flagging design would have eliminated 49% of it at the same recipient benefit.
with Zunda Winston Xu
Interest rate caps are intended to protect borrowers from debt traps, but they can also restrict credit to households that rely on high-cost loans to smooth shocks, so the welfare effect of a uniform APR cap depends on why borrowers become trapped. We study this question in the U.S. market for high-cost installment loans using a credit-bureau panel covering roughly 15 million originations from 2013 to 2020, defining a debt-trap episode from the contractual amortization path: a borrower is trapped when a high-cost loan fails to pay down as promised and is reset before substantial amortization. We show that debt traps are persistent and concentrated among financially fragile borrowers, though not confined to the worst observed credit risks, and we present reduced-form evidence consistent with a behavioral margin and difficult to reconcile with a purely liquidity-based or purely time-consistent explanation. Using state-level changes in interest-rate caps, we show that trapped borrowers experience improved outcomes when high-cost credit is restricted, whereas non-trapped borrowers are made worse off by the resulting loss of credit access. We then develop a structural model of installment borrowing with heterogeneous patience, partial naivete, private risk, and endogenous lender exit, which implies that a uniform APR cap can raise welfare for behaviorally biased borrowers but lower it for others, so that policies targeting repeated refinancing or requiring minimum amortization can deliver consumer protection at lower cost.