research

How do collateral requirements impact consumer borrowing behavior? Using administrative loan application and performance data from the U.S. Federal Disaster Loan Program, we exploit a loan amount threshold above which households must post their residence as collateral. Our bunching estimates suggest that the median borrower is willing to give up 40% of their loan amount to avoid posting collateral. Exploiting time variation in the threshold, we estimate collateral causally reduces default rates by 36%. Finally, we structurally estimate households' attachment to their homes, net of any equity, and find a median value of \$11,000. Attachment creates a wedge between lender and borrower valuation of collateral of 15%. Our results explain high perceived default costs in the mortgage market, and document the importance of collateral for reducing moral hazard in consumer credit markets.

We estimate and trace a credit demand curve for households that recently experienced damage to their homes from a natural disaster. Our administrative data include over one million applicants to a federal recovery loan program for households. We estimate extensive-margin demand over a large range of interest rates. Our identification strategy exploits 24 natural experiments, leveraging exogenous, time-based variation in the program's offered interest rate. Interest rates meaningfully affect consumer demand throughout the distribution of rates. On average, a 1 percentage point increase in the interest rate reduces loan take-up by 26%. We find a large impact of applicants' credit quality on demand and evidence of monthly payment targeting.

In catastrophe-exposed insurance markets, a large share of what homeowners pay reflects the cost of correlated tail risk. Local insurers cannot diversify these losses on their own and instead transfer them to globally diversified reinsurers, but reinsurance is expensive and volatile. We study a regulatory reform that decreased the capital cost of reinsuring tail risk. After the cost decrease, reinsurance use expanded by 25%, the reinsurance market became less concentrated, and the model-implied price of reinsurance fell by a third. By reducing required capital, the reform also increased insurers' exposure to reinsurer non-payment risk. We show this dramatically impacted the primary insurance market in hurricane-exposed areas of Florida: additional insurers entered and the cost of insuring wind risk fell sharply. To quantify welfare, we estimate an equilibrium model of the Florida wind-insurance market. The model implies that the reform reduced treated insurers' marginal cost of supplying wind coverage by about 10% and lowered equilibrium premiums by about 14%. Counterfactual simulations imply consumer-surplus gains of \$186 per household per year, split roughly equally across direct cost pass-through, strategic markup adjustment, and expanded product availability. Our results imply that capital costs required to cover tail events, over and above average losses, meaningfully contribute to consumers' premiums and that policies that reduce those capital costs can decrease prices and improve availability for homeowners insurance.

This paper examines the moral hazard effects of cost-sharing subsidies on the Affordable Care Act's Health Insurance Exchanges. Exploiting a sharp discontinuity in subsidy generosity at 150% of the federal poverty level, we compare healthcare spending for individuals just above and below this threshold using a regression discontinuity design and data from the Medical Expenditure Panel Survey. We find that individuals just below 150% federal poverty level who receive the most generous subsidies spend approximately \$1860 more annually on healthcare compared to those just above the threshold receiving less generous subsidies, implying an elasticity of −0.52. Several analyses suggest this discontinuity reflects moral hazard rather than adverse selection or health differences across the income threshold. The results highlight the significant impact of moral hazard induced by generous cost-sharing subsidies, with important implications for the design of means-tested health insurance subsidies.

While many states have legalized medical cannabis, many unintended consequences remain under-studied. We focus on one potential detriment—the effect of cannabis legalization on automobile safety. We examine this relationship through auto insurance premiums. Employing a modern difference-in-differences framework and zip code-level premium data from 2014 to 2019, we find that premiums declined, on average, by \$22 per year following medical cannabis legalization. The effect is more substantial in areas near a dispensary and in areas with a higher prevalence of drunk driving before legalization. We estimate that existing legalization has reduced health expenditures related to auto accidents by almost \$820 million per year with the potential for a further \$350 million reduction if legalized nationally.

When consumers gain Medicaid, their cost of healthcare changes. The direction of this change determines how utilization changes. The previously uninsured see a stark decrease in the price of primary care after gaining public insurance. Due to charity care, they may face an increase in the price of emergency department care. The previously insured see a reduction in emergency department prices and decreased access to primary care. We examine the impact of the prior insurance status of the newly publicly insured on substitution between healthcare. We base our identification on California's LIHP and ACA Medicaid expansions. One challenge we face is estimating crowd-out. We use machine learning techniques to predict prior insurance status based on observable covariates in cross-sectional data. We find an increase in emergency department utilization caused entirely by those crowded-out whose access to primary care has decreased. We find the opposite utilization patterns for the previously uninsured.

We show that donors, not managers, drive risk-averse behavior in nonprofit organizations (NPOs). Theoretically, when donor recognition is tiered (e.g., naming rights vs. thank-you cards), donors concentrate rather than diversify giving, making shadow donation capital costs sensitive to NPO-specific risk. This induces risk-averse actions even without risk-averse managers. We test our theory using hospitals and the staggered adoption of medical liability caps and find that reduced NPO risk increases donations. Effects on substitute bond-financing indicate this increase is supply- (donor-) driven, not demand- (manager-) driven. Liability caps lead nonprofit hospitals to expand risky investments faster than for-profits, improving patient health.

We examine how private equity (PE) firms generate value in markets where prices are regulated and do not reflect costs. Using novel, comprehensive data from Arizona's ambulance industry, we find PE-owned companies increase their operating profits by 50% by cream-skimming: through strategically exploiting regulations, and avoiding minimum service requirements, PE firms are able to shift unprofitable customers to the government while retaining high-profit customers. They accomplish this by firing paramedics, which, due to nationwide staffing regulations, forces local fire departments to take high-cost runs. This strategic reallocation only occurs where PE firms overlap with a sufficient number of fire departments, which allows them to avoid minimum timing requirements. These operational changes increase total call time by 6%, contributing to a significant increase in traffic fatalities in Arizona, with effects magnified in areas where PE firms face non-PE competition, and a 7% increase nationally. Our findings demonstrate how regulatory arbitrage by PE in mixed public-private markets can create substantial negative externalities for public balance sheets and for public health.

We identify and quantify a new form of welfare loss in insurance markets. We first show theoretically that moral hazard from subsidies for cost-sharing combined with community rating mimics adverse selection and can unravel insurance markets. To quantify the potential welfare loss, we use exogenous variation in the number of subsidized enrollees on the ACA exchanges. We find that subsidy-induced moral hazard led to higher premiums, which has lowered enrollment among the unsubsidized by 7.6 percentage points. We estimate the welfare costs of this "moral hazard induced unraveling" to be around 25% of the welfare loss from existing adverse selection.

Many Americans remain financially underprepared for retirement. While automatic enrollment in employer-sponsored retirement plans has helped, target-date funds (TDFs) used as default investments have limitations. We propose target-date registered index-linked annuities (TD-RILAs) as a transparent, cost-effective alternative providing decreasing equity exposure over time. A theoretical analysis explores TD-RILAs' optimal structure and compares them to TDFs. A lab experiment examines investors' preferences, product transparency, default choices, and the role of information. We find that TD-RILAs are a suitable addition to retirement plans, potentially rivaling TDFs.

We develop a model of insurance pricing under heterogeneous lapse rates with asymmetric information about lapse likelihood within the context of an optional two-part tariff as a screening device for future policyholder behavior. We then test for consumer self-selection using policy-level data on life insurance backdating. We exploit randomness in the initial tariff size to separately identify the selection and sunk cost effects of backdating on lapse proclivity. We find that consumers who are less likely to lapse self-select into the two-part tariff pricing structure and we also document consumer behavior consistent with sunk cost fallacy.

Life insurance is often believed to feature advantageous, rather than adverse, selection since policyholders have lower mortality than non-policyholders. We introduce policy lapses as an alternative selection channel. We demonstrate theoretically that lapsing can induce adverse selection, even among naïve consumers. We then empirically validate our theory using administrative data and confirm that lapse rates respond significantly to price changes, leading to adverse selection and a downward-sloping cost curve. No comparable effect emerges in mortality rates. This adverse selection reduces overall demand by 14.2% and reduces welfare by \$1.3 billion annually. Furthermore, lapse-based selection exacerbates inequity, disproportionately benefiting wealthy consumers.

In this paper, we introduce and employ a multi-agent model of entry to analyze barriers to entry in the U.S. property casualty insurance industry. Our estimations are conducted at the market-year level, enabling a detailed exploration of the relative importance of barriers across three dimensions: geography, product, and time. Our findings reveal that entry barriers are both prevalent and significant in the U.S. property casualty insurance industry, with de novo entrants encountering the most challenges from product barriers across all markets. We observe that expanding across product lines is more costly than expanding across states. Particularly, New York and California exhibit the strongest state barriers. Among various product lines, our research identifies that insurance expertise in mortgage guaranty and medical professional liability insurance presents the most substantial barriers to entry. This study contributes valuable insights into the dynamics of entry barriers in the U.S. property casualty insurance industry, highlighting the importance of considering geographic and product-specific knowledge, as well as temporal factors, in the context of market entry. Through analyses across these dimensions, we enhance the understanding of the challenges faced by new entrants and shed light on the specific market conditions shaping entry barriers.

In the past 30 years, consumer credit card debt has expanded tremendously. We know that consumers willingly pay more for the same product when using credit cards versus cash, contrary to the classical rational agent model. Research suggests that it happens due to three imperfections in the classical model: imperfect self-knowledge, imperfect willpower, and imperfect recall. Traditional solutions to credit abuse address the first two imperfections; we examine the third. We propose reminding consumers, on every receipt, how much they have spent cumulatively. We test the effect of this proposal on spending via a controlled experiment. We find that printing this additional information on credit card receipts leads to a significant 9.6% reduction in overall spending compared to the status quo. We discuss the public policy implications of this finding as well as implementation issues.

Since the advent of the file-sharing program Napster in June of 1999, copyright infringement has plagued the recorded music industry. We review the evidence on piracy and its effect on record industry profits. We then model the behavior of file sharers and music producers under different remuneration and legal regimes using a stage game. We find that under certain conditions, the removal of copyright laws for recorded music is welfare improving. There is also a parameter space where public sector music distribution combined with a tax-payer funded subsidy of music production is welfare dominant.

Financial economists often use regressors "constructed" from values of other observations in the same dataset, with lagged and leave-out variables being common examples. We examine the use of such variables in common settings with fixed effects and show that it can induce bias and distort inference. We illustrate the severity of this problem via simulations and with patent examiner data using a leave-out instrument. Even when scrambling the patent examiners, thus removing any instrument validity, the bias leads to a first-stage F-statistic over 1,000. General and case-specific solutions are provided.

Under the federal regulations of the Brady Act, individuals using medical cannabis are prohibited from legally purchasing firearms. However, 36 states permit medical cannabis use, compelling individuals to choose between legally possessing cannabis or guns, but not both. This backdoor ban allows us to utilize the differential timing of cannabis legalization to identify the impact of gun purchase restrictions on gun crime. Using difference-in-differences and triple difference-in-difference models on FBI Uniform Crime Reports data, we find medical cannabis reduces assaults and robberies with firearms by 5%, relative to knife crimes. We find no impact on knife crimes alone, suggesting substitution away from crime instead of towards other weapons. Further analysis shows the effect is concentrated near cannabis dispensaries, supporting our mechanism.

We analyze the causal impact of the 2016 adoption of the Solvency II regulatory framework on equity risk, default risk, and stock liquidity of the newly-regulated insurers. Using nine proxies in a triple difference model, we find that the new regulation has caused significantly lower levels of firm risk and higher levels of liquidity for regulated European insurers as opposed to a control sample with entities from the U.S. and Japan. We conclude that the stricter capital adequacy and risk management requirements of Solvency II affect business models and how resulting risks are dealt with, and that the regular risk-based disclosures better inform capital market participants. This suggests that — at least from a capital market perspective — the new regulatory framework fulfills its key promises.

Thirty-eight states have legalized sports betting since 2018, yet policy debates ignore local governments. We ask whether state sports betting policy affects local fiscal health. Using municipal bond yields (2018-2024) and Census of Governments data (2017-2022), we estimate an imputation difference-in-differences model exploiting staggered legalization and cross-sectional variation in tax rates. Legalization improves local fiscal conditions, but higher state tax rates diminish these gains. Across all legalizing states, bond yields fall by approximately 10 basis points and municipal sales tax revenue rises by roughly \$44,000 per municipality. At the median tax rate (15%), the yield reduction is approximately 14 basis points and the revenue gain is roughly \$36,000. Each percentage point increase in the state tax rate erodes these benefits. We find no extensive-margin effect on intergovernmental transfers, suggesting effects flow through consumer spending rather than revenue-sharing.

Managerial, or discretionary, earnings opacity is the intentional lack of transparency to hide the intrinsic value of a firm. Opacity arises through two channels. The first is ex ante, when managers manipulate current expectations about future performance; and the second is ex post, when managers hoard negative news that corrects their initial estimates. Using insurance industry data and the 2015 FASB accounting standards update in a difference-in-differences framework, we investigate the relationship between discretionary opacity and stock price informativeness. We find that the 2015 FASB update increased informativeness through a reduction in delayed news release that reduced opacity.

A large share of political fundraising email is routed to spam. Inbox placement for Republican committee email varies by 30 percentage points across mailbox providers on a typical week and moves by comparable amounts within provider over time; higher spam placement tracks lower Google search interest for GOP senators and lower Republican Senate implied probabilities on Kalshi and Polymarket. Two natural experiments in 2025 pin down the fundraising cost: a 78-day episode in which Gmail, Outlook, and Yahoo routed emails containing WinRed donation links to spam while routing otherwise-identical ActBlue-linked emails to the inbox, and a separate Outlook-only placement drop in February-April. First-time donors on WinRed decline -74.6% in donation count during the June episode and -83.2% during the Outlook drop, with no analogous movement in the other DDD cells. The two episodes together cost 142,126 first-time donations over 167 days of filtering — roughly 850 missing first-time donations for every day a link-based filter was active, worth about \$700,000 in eight-month lifetime donor value per day. Pre-episode WinRed new donors give \$461 on their first gift and cumulatively \$827 over eight months, so the dollar cost of each missing first gift compounds well beyond the contemporaneous shortfall.

We investigate whether the U.S. patent examiner lottery, which is intended to ensure random assignment and equal treatment of applicants, can be strategically manipulated. Using data on six million applications from 2001-2020, we derive the null distribution of examiner leniency under true randomization and show that many law firms and in-house counsel systematically obtain far more lenient examiners than chance would permit. We document multiple mechanisms consistent with strategic manipulation, including leveraging institutional knowledge and the exploitation of publicly observable examiner workloads. These deviations from randomness have economically meaningful consequences: sorting firms on examiner leniency at publication produces sizable and statistically significant return spreads, implying that markets only partially incorporate assignment-based variation in approval likelihood. We develop a simple model of manipulation investments that explains which firms are most likely to game the patent lottery, highlight potential welfare consequences, and show empirical support for its predictions. Our findings challenge the fairness of the patent lottery and call into question empirical designs that rely on examiner randomization for causal identification.