Research
Private Information and Price Regulation in the US Credit Card Market (2022). R&R (2nd Round), Econometrica
Abstract: The 2009 CARD Act limited credit card lenders' ability to raise borrowers' interest rates on the basis of new information. Pricing became less responsive to public and private signals of borrowers' risk and demand characteristics, and price dispersion fell by one third. I estimate the efficiency and distributional effects of this shift toward more pooled pricing. Prices fell for high-risk and price-inelastic consumers, but prices rose elsewhere in the market and newly exceeded willingness to pay for over 30% of the safest subprime borrowers. On net, average traded prices fell and consumer surplus rose at all credit scores. Higher consumer surplus was partly driven by a fall in lender profits, and partly by the Act's insurance value to borrowers who could retain favorable pricing after adverse changes to their default risk. The relatively high level of pre-CARD-Act markups was crucial for realizing these surplus gains.
Tax Refund Uncertainty: Evidence and Welfare Implications (2022), with Sydnee Caldwell and Daniel Waldinger. Conditionally Accepted at AEJ: Applied Economics
Abstract: Transfers paid through annual tax refunds are a large but uncertain source of income for poor households. We document that low-income tax-filers have substantial subjective uncertainty about these refunds. We investigate the determinants and consequences of refund uncertainty by linking survey, tax, and credit bureau data. On average, filers' expectations track realized refunds. More uncertain filers have larger differences between expected and realized refunds. Filers borrow in anticipation of their refunds, but more uncertain filers borrow less, consistent with precautionary behavior. A simple consumption-savings model suggests that refund uncertainty reduces the welfare benefits of the EITC by about 10 percent.
Deleting a Signal: Evidence from Pre-Employment Credit Checks (2022), with Alexander W. Bartik. Accepted at Review of Economics and Statistics
Abstract: We study the removal of information from a market, such as a screening tool from labor markets or payment history from lending markets. Theoretically we show the effects of removal depend on a measure of signal precision scaled by the precision of other available information. This generates implications for incidence across groups: so long as the informed side of the market benefits from conveying any information at all, a signal ban harms the group for which, all else equal, the banned signal is most precise and other information sources are most noisy. Empirically, we illustrate these mechanisms in the context of banning the use of credit reports to screen job applicants. We estimate such bans have decreased job-finding rates for Black job-seekers by 3 percentage points, despite Black individuals having poorer average credit than other groups. We find this effect emerges primarily because other screening tools, such as referrals and interviews, have around 70% higher standard deviation of signal noise for Black relative to white job-seekers.
Information Design in Consumer Credit Markets (2022), with Laura Blattner and Jacob Hartwig.
Abstract: Over 30m US adults do not use formal consumer credit. How many of these are inefficiently excluded because they lack a credit history or have a poor credit score? We develop a framework to characterize the efficiency-maximizing system of credit histories and credit scoring, subject to the constraints imposed by the severity of adverse selection, and by the ability of credit histories to predict future risk. We find US consumer credit features a moderate amount of adverse selection and persistent consumer types. This adverse selection generates substantial welfare loss: a majority of today's non-borrowers would be first-best efficient to lend to. Credit reporting helps alleviate the costs of adverse selection, with the current US system recovering roughly two-thirds of the welfare that would be lost in a no-credit-reporting counterfactual, relative to a full-information first-best. We find that requiring histories to be shorter -- or to forget past default sooner -- would induce some market unraveling but also would help non-borrowing consumers escape the ``no history trap.''
How Costly is Noise? Data and Disparities in Consumer Credit (2022) with Laura Blattner.
Abstract: We establish three facts about how credit scores drive disparities for disadvantaged groups in consumer credit markets. First, we show that widely used credit scores drive disparities in classification errors. That is, among disadvantaged groups, there are more applicants who are predicted to repay a loan but do not repay and more applicants predicted not to repay a loan who would have repaid. Second, we nd that disparities in errors arise not because of (statistical) bias against a particular group, but rather because of noise, or higher variance of errors for disadvantaged groups. Third, we demonstrate that three plausible solutions do not help to reduce noise in credit scores. Developing better prediction technology based on machine learning techniques, building group-specific credit scoring models, and adding non-traditional credit bureau data all fail to eliminate the noise disparity we document. Rather, we provide evidence that reducing the noise disparity requires addressing higher credit report data sparsity for the disadvantaged groups.
Unpacking the Black Box: Regulating Algorithmic Decisions (2021), with Laura Blattner and Jann Spiess.
Abstract: We characterize optimal oversight of algorithms in a world where an agent designs a complex prediction function but a principal is limited in the amount of information she can learn about the prediction function. We show that limiting agents to prediction functions that are simple enough to be fully transparent is inefficient as long as the bias induced by misalignment between principal’s and agent’s preferences is small relative to the uncertainty about the true state of the world. Ex-post algorithmic audits can improve welfare, but the gains depend on the design of the audit tools. Tools that focus on minimizing overall information loss, the focus of many post-hoc explainer tools, will generally be inefficient since they focus on explaining the average behavior of the prediction function rather than sources of mis-prediction, which matter for welfare-relevant outcomes. Targeted tools that focus on the source of incentive misalignment, e.g., excess false positives or racial disparities, can provide first-best solutions. We investigate the empirical relevance of our theoretical findings using an application in consumer lending.
Zoning for Profits: How Public Finance Shapes Land Supply in China (2023), with Zhiguo He, Yang Su, Anthony Lee Zhang, and Fudong Zhang.
Abstract: Public finance and real estate are uniquely intertwined in China, where local governments also serve as monopolist sellers of land. We shed new light on how land sale decisions and land prices depend on local governments’ financing objectives. First, we document the large (ten-fold) price premium paid for residential-zoned relative to industrial-zoned land and show this price premium can be explained by the greater future tax revenues generated by industrial land; the choice to sell land as industrial rather than residential generates an IRR of 7.70%, which is comparable to local governments’ cost of capital in bond markets. Second, local governments are sensitive to financing constraints: industrial land supply decreases with governments’ bond yields. Third, local governments’ land sales are sensitive to the intergovernmental tax sharing, such that industrial land sales increase with the share of taxes captured by local governments. Thus, shocks to local public finances can be expected to affect the Chinese real estate market and vice versa.
The Arity of Disparity: Updating Disparate Impact for Modern Fair Lending (2023), with Spencer Caro.
Abstract: Discrimination and protections against it are both rapidly evolving. To address issues raised by algorithmic screening and by recent economic research on competing notions of fairness, we formalize the last step in a disparate impact (DI) claim of discrimination — the Least Discriminatory Alternative (LDA) analysis — as an explicit constraint on choices over screening models and data inputs. The constraint restricts model-induced disparities in both parity and accuracy relative to a budget that depends straightforwardly on overall model performance. We also show how this trade-off leads to balancing other notions of fairness that are spanned by weighted combinations of parity and accuracy. We illustrate examples and develop a legal argument focused on DI under the Equal Credit Opportunity Act (ECOA). Our approach resolves the tension between DI's traditional focus on parity and ECOA's statutory emphasis on credit-worthy consumers, and we discuss implications for new frontiers in credit underwriting such as the use of machine learning and alternative data.