Publications
Review of Economic Studies (Published June 9, 2025)
with Juliane Begenau, Saki Bigio, and Matias Vieyra
Abstract
Bank capital requirements are based on book values, which are slow to reflect losses. In this paper, we develop a dynamic model of banks to study the interaction of regulation and delayed accounting. Our model explains four stylized facts: book and market values diverge during crises, the market-to-book ratio predicts future profitability, book leverage constraints rarely bind strictly even as market leverage fans out during crises, and banks delever gradually after net-worth shocks. We show how delayed accounting can allow the regulator to achieve better outcomes than immediate (mark-to-market) accounting. In an estimated version of the model, the optimal regulation couples faster loan-loss recognition with a modest relaxation of the book leverage constraint.
American Economic Review Papers and Proceedings 106(5): 282-88, 2016
with Raj Chetty, Nathan Hendren, Frina Lin, and Ben Scuderi
Abstract
We show that differences in childhood environments play an important role in shaping gender gaps in adulthood by documenting three facts using population tax records for children born in the 1980s. First, gender gaps in employment rates, earnings, and college attendance vary substantially across the parental income distribution. Notably, the traditional gender gap in employment rates is reversed for children growing up in poor families: boys in families in the bottom quintile of the income distribution are less likely to work than girls. Second, these gender gaps vary substantially across counties and commuting zones in which children grow up. The degree of variation in outcomes across places is largest for boys growing up in poor, single-parent families. Third, the spatial variation in gender gaps is highly correlated with proxies for neighborhood disadvantage. Low-income boys who grow up in high-poverty, high-minority areas work significantly less than girls. These areas also have higher rates of crime, suggesting that boys growing up in concentrated poverty substitute from formal employment to crime. Together, these findings demonstrate that gender gaps in adulthood have roots in childhood, perhaps because childhood disadvantage is especially harmful for boys.
Working Papers
with David Hughes
Abstract
Misallocation of inputs across firms has been proposed as a reason for low levels of development in some countries. However, existing work has largely relied on strong assumptions about production functions in order to estimate the cost of misallocation. We show that, for arbitrary production functions, the cost of misallocation can be expressed as a function of the variance of marginal products. Using an RCT that gave grants to microenterprises, we estimate heterogeneous returns to capital by baseline characteristics, and provide a lower bound on the total variance of returns to capital. This lower bound is a nonlinear function of the parameters from a linear IV model, and we show that standard methods (e.g. the delta method or projection) fail in this setting. We provide novel econometric tools that provide uniformly valid confidence intervals for nonlinear functions of parameters. We find evidence for sizable losses from misallocation of inputs across the firms we study, although the magnitude depends critically on which inputs we allow to be reallocated. We estimate that optimally reallocating capital would increase output by 22%, while optimally reallocating all inputs would increase output by 301%.
with Miguel Faria-e-Castro and Julian Kozlowski
Abstract
We show how to use credit registry microdata to estimate the cost of capital, and how this affects capital allocation efficiency in the United States. Our measure of the cost of capital accounts for the interest rate, expected default probability, recovery rates, and, for floating-rate loans, the expectation of future rates. We find that, on average, the lender's cost of capital closely tracks the five-year Treasury rate, with a spread of 1.5%. Misallocation depends on dispersion in the social cost of capital, which equals the lender's cost of capital plus an agency friction. We find that in normal periods, the implied misallocation is small, resulting in an output loss of only 0.5%. However, the dispersion in the cost of capital rose dramatically during the COVID-19 pandemic, driven by agency frictions. By integrating microdata with a corporate finance framework, this study highlights the resilience of U.S. credit markets under typical conditions, and underscores the inefficiencies that can arise during a crisis.
with Anthony Yu
Older Draft with Additional Results
Abstract
We study the effects of the typical merger in the consumer packaged goods industry, a sector making up over 10% of United States gross domestic product. Using an event-study design and linked retail scanner data from hundreds of mergers, we find that mergers raise prices at the target by approximately 1%. Under nested CES demand, we provide sufficient statistics to recover average consumer welfare effects as a function of effects on price, product availability, and exit. Accounting for availability and exit is quantitatively important. The decline in consumer welfare is equivalent to a 1.9% price increase at the target firm.
with Karthik Sastry
Abstract
Many prominent studies in macroeconomics, labor, and trade use panel data on regions to identify the local effects of aggregate shocks. These studies construct regional-exposure instruments as an observed aggregate shock times an observed regional exposure to that shock. We argue that the most economically plausible source of identification in these settings is uncorrelatedness of observed and unobserved aggregate shocks. Even when the regression estimator is consistent, we show that inference is complicated by cross-regional residual correlations induced by unobserved aggregate shocks. We suggest two-way clustering, two-way heteroskedasticity- and autocorrelation-consistent standard errors, and randomization inference as options to solve this inference problem. We also develop a feasible optimal instrument to improve efficiency. In an application to the estimation of regional fiscal multipliers, we show that the standard practice of clustering by region generates confidence intervals that are too small. When we construct confidence intervals with robust methods, we can no longer reject multipliers close to zero at the 95% level. The feasible optimal instrument more than doubles statistical power; however, we still cannot reject low multipliers. Our results underscore that the precision promised by regional data may disappear with correct inference.
with Paco Buera, Yongseok Shin, and Kuldeep Singh
Abstract
Risk is an important factor that affects investment decisions, especially for undiversified entrepreneurs in less developed economies. Yet standard macro models of financial frictions do not incorporate risk: short-term returns are known in advance, and investment is fully reversible. Thus, even if entrepreneurs are risk averse and credit constrained, they will invest all of their assets in the firm, until the marginal product of capital equals the interest rate. As a result, standard models often find that productive entrepreneurs quickly save their way out of credit constraints, limiting the effect of financial frictions on output and aggregate productivity. We incorporate risk into a model of financial frictions, by making investment partially irreversible. Productive entrepreneurs accumulate capital substantially more slowly than in the first-best, leading to a reduction in aggregate productivity. Credit can play a role in undoing these frictions if firms have an option to default. Default creates a state-contingent contract, in which the entrepreneur repays if productivity stays high and defaults if productivity falls; this encourages investment and improves welfare through risk-sharing with the bank.
Abstract
An important determinant of aggregate productivity is the selection channel: the process by which less efficient firms are driven out of the market by more efficient firms. Conventional wisdom suggests that markets in developing countries are more sclerotic, allowing inefficient firms to survive that would have exited in a developed country. I provide a tractable model to examine the importance of the selection channel, and show how to calibrate it to panel data on firms. I use this model to show that the effect of the selection channel on aggregate productivity is approximately equal to the average difference in log productivity between stayers and exiters, which can be measured easily in firm panel data. Results for Indonesia, Spain, Chile, and Colombia suggest that Indonesia could raise its aggregate productivity by roughly 30% if its firm exit process became as selective as Spain’s. However, cross-country estimates suggest that the selection channel is not an important explanation for cross-country differences in output per capita.
Work in Progress
Abstract
Both economists and the public are deeply interested in the degree to which a child’s adult income is determined by her parent’s income and race. Recent work (Chetty, Hendren, Jones, and Porter, 2020) has shown, for recent cohorts, that black boys have much lower incomes in adulthood than do white boys, even after controlling for parent income. Moreover, their estimates suggest that the black-white income gap is already at its steady state: the gap will remain at its current level unless upward mobility improves for black children. This begs the question: has the black-white mobility gap been improving or worsening over time? Estimating intergenerational mobility by race for earlier cohorts has been difficult to date because of data limitations: accurate estimates require large, high-quality data sets, and such linked data containing parent income, child income, and race are only currently available for recent cohorts. I provide a solution to this problem, using a GMM approach to combine panel data from the NLS and NLSY with cross-sectional data from the Census. The addition of the Census data imposes restrictions on the coefficients that allows for substantially more precise estimates over time.