Condensed courses, which deliver the same instructional content over a shorter time frame, have been increasingly used by colleges to expand scheduling flexibility. Emerging evidence suggests that these formats often improve course performance, but little is known about whether their effects differ across instructional modalities. Using administrative data from a statewide community college system, we compare outcomes across condensed and traditional-length sections within modality. We find that condensed formats improve current and subsequent outcomes when delivered in person, while online condensed courses show no comparable benefits and negative effects on performance in subsequent coursework.
We examine whether US interior immigration enforcement policies deter deportees from reentering the United States and whether these effects vary by individual characteristics. Using data from the Mexican Survey of Migration to the Northern Border and exploiting variation in E-Verify mandates and 287(g) agreements across states over time, we find that these policies reduce the likelihood of reentry. We show that each policy disproportionately affects deportees with specific characteristics: E-Verify mandates have larger effects on deportees with stronger labor market attachment, while 287(g) agreements have larger effects on those with lower human capital or who have committed a serious crime.
This paper examines whether mismeasurement due to markups is responsible for the decline in US manufacturing productivity growth. It estimates markups using a model to estimate the user cost of capital to split capital payments into regular returns and economic profit. I do not find that the slowdown is due to mismeasurement. Correcting for markups tends to strengthen the slowdown. Labor input growth has been faster than output growth. The markup correction increases the impact of labor changes. The results are consistent with the slowdown resulting from slower technical change in investment products.
We investigate the aggregate elasticity of substitution between clean and dirty energy using time-series data from the United States. To match the data, a standard macro climate-economy model requires an elasticity that is close to one in the long run but well below one in the short run. Impulse response functions show that the elasticity converges slowly to its long-run value after a shock to dirty energy prices. Our findings suggest that carbon taxes reduce emissions more slowly than standard models predict.
In general, chained deflators for IT goods will have a commingling problem that causes them to overstate quality growth. We base this on a comparison of chained superlative indexes with a multilateral index that does not have the commingling problem—the weighted time product dummy index. Empirical results from three recent studies suggest that this can occur often—we saw this in 9 of the 12 categories studied—and that the gaps can be nontrivial. We argue that the recently developed multilateral methods may be a better strategy in price measurement for this sector.
Export controls have become a central instrument of international economic policy, governing trade in dual-use and sensitive technologies among allies and rivals alike. Despite their prominence, little is known about how multilateral export control regimes affect trade between participating countries. Using long-run gravity estimates for 1965–2010, we show sizable increases in bilateral trade among joint members in multilateral export control regimes relative to nonmembers. The effects are heterogeneous across regimes and comparable in magnitude to preferential trade agreements. Our findings highlight export controls as institutions of regulatory coordination that shape aggregate trade patterns in modern global production networks.
We study the effects of colonial market access on the loss of lands for Indigenous nations, using the rapid expansion of the US railroad network in the nineteenth century as a quasi-natural experiment. We find that increased market access to Indigenous homelands led to a hastening of land dispossession. We find that both reductions in transport costs that improved connectivity to large population centers and the westward advancement of mass settlement were important channels. Taken together, our paper provides evidence that, unlike for settlers' outcomes, railroad-induced market access improvements may not have been beneficial for Indigenous peoples.
Our paper simulates a model of trade, diplomacy, and wars, calibrated to be relevant for analyzing the conflict between Russia and Ukraine. The quantitative results show that Ukraine’s decoupling from Russia after the 2014 annexation of Crimea reduced its economic exposure to war but also increased the likelihood of conflict by lowering its opportunity cost of war. We then simulate sanction scenarios and find that the 2024 package, if credibly announced in 2021 and made contingent on further aggression, would have significantly raised Russia’s cost of war and could have deterred the 2022 invasion.
This study explores whether the salience of the primary investigator's (PI) racial identity impacts how study participants evaluate and utilize information about a diverse pool of workers. Our research is motivated by the literature on pervasive anti-Black hiring discrimination and by a new literature testing whether centralized hiring practices and oversight can counteract discriminatory beliefs about Black workers (Bertrand and Mullainathan 2004; Quillian et al. 2017; Bertrand and Duflo 2017; Pager, Bonikowski, and Western 2009; Kline, Rose, and Walters 2022). Specifically, we hypothesize that study participants who are randomly assigned to a Black PI may engage in less discriminatory behavior relative to those assigned to a White PL This hypothesis is driven by two primary mechanisms: (i) Participants may derive utility from presenting themselves favorably to their supervisor (Brunnermeier and Parker 2005; Koszegi 2006; Burks et al. 2013) and (ii) the presence of a supervisor can anchor how employees evaluate worker performance (Bazerman, Giuliano, and Appelman 1984; Schoorman 1988). To test our hypothesis, we conducted an online study with a nationally representative sample of 3,000 US-based participants on Prolific. In the study, participants served as recruiters and were tasked with evaluating a set of eight stylized resumes from a separate pool of Amazon Mechanical Turk (MTurk) workers. During their review, the racial identity of their supervisor (one of the PIs of this study) was revealed. Our results suggest that the presence of a Black supervisor can affect stated beliefs about productivity differences across racial groups. Specifically, we show that Black workers are perceived as more productive under a Black supervisor than under a White supervisor. We find suggestive evidence that these participants update more in response to signals suggesting that Black workers are more productive than White workers; this belief-updating channel may be one contributor to why participants assigned to Black supervisors evaluate Black workers more favorably.
The Immigration and Nationality Act of 1965 imposed, for the first time in US history, annual numerical restrictions on lawful immigration from the Western Hemisphere. This paper introduces and implements a method for estimating highly granular local unauthorized rates in the wake of the surge in Latin American immigration that ensued. Using within-state, cross-county variation in these estimated unauthorized rates for foreign-born Mexican women and multiple comparison groups, we then estimate how maternal legal status affects the health of US-born—and thus US-citizen—children. We find that maternal authorization increases birth weight in the second generation.
We examine worker-employer disputes over unemployment insurance (UI) claims and their relationship to benefit nonpayment. Administrative UI records allow us to use workers who file multiple claims to control for unobserved worker heterogeneity. We observe strong employer effects on whether claims are disputed, and weak evidence that employers dispute Black workers’ claims more often than White workers’ claims. But disputes are only weakly related to nonpayment, and nonpayment associated with employer disputes does not differ by race or ethnicity. We speculate that the UI agency mediates employer disputes to mitigate racial differences in UI nonpayment.
This paper uses the 2015 US Transgender Survey to estimate the association between gender-affirming hormone therapy (GAHT) initiation and the risk of running away among transgender adolescents during the first year of treatment. Our event study approach compares transgender adolescents who started GAHT with those who initiated GAHT a year later. Results indicate that GAHT is associated with a 2 percentage point reduction in the risk of running away from home, which amounts to a 20 percent decrease in the risk of running away. The association is largest when GAHT begins at younger ages (i.e., 14–15).
We discuss the limitations of using unit quantity data to measure output in the estimation of production functions. We introduce a simple model, the quality-equivalent Cournot model, where revenue or deflated revenue is the preferable measure of output as it allows consistent estimation of output elasticities and markups. In contrast, because reported units are not comparable across firms, relying on quantity data for productivity analysis or to generate output elasticities used in the production approach to markup estimation may be problematic.
This paper examines changes in joint labor supply among married US households during the period of trend reversal in aggregate labor supply from the late 1990s to the early 2020s. Married women, particularly those with children, exhibit the most significant increases in work hours, while their husbands experience the largest declines. Using a two-earner life-cycle model, we find that the rise in female wages relative to male wages is central to these patterns, in part by inducing fathers to reallocate time toward childcare.
A long literature in economics and public health has examined the relationship between socioeconomic status (SES) and health, often relying on area-based measures when individual-level data are unavailable. Using comprehensive Dutch administrative data, we build on work estimating income-health gradients (e.g., Chetty et al. 2016) and study how the level of aggregation systematically shapes estimates by inducing so-called ecological or atomistic biases. Individual and area income proxy for different channels (e.g., personal resources versus local surroundings) but are obviously correlated. Aggregating to the area level reduces meaningful variation and attenuates nonlinearities. We find that area-level estimates exaggerate the income gradient relative to individual-level estimates and are less robust to including individual-and area-level controls, respectively. Moreover, interaction analyses reveal that area exposure is highly unequal: The health of low-income households is far more sensitive to neighborhood conditions than that of high-income households. Together, these findings shed light on the sources of SES-health gradients and show how data aggregation affects both interpretation and empirical estimates.
After registering a 3 percent annual rate from 1948 to 2010, the growth of labor productivity in US manufacturing since 2010 has come to a cold stop, with a zero 2010–2025 annual growth rate. What caused the disappearance of productivity growth in the core of the US economy? We point to the channels by which an invasion of imports, led by China in the 2000–2010 decade, eroded the competitiveness of US producers. Imports competed away sales, closed plants, cut capacity utilization, slashed profits and investment, postponed installation of modern technology, and diverted resources from R&D and innovation.
We study how competitive banks design liabilities that function as money in an economy with asymmetric information. Assets differ in risk exposure, and sufficiently risky assets cannot circulate in decentralized trade. Banks commit to state-contingent payoffs and transform asset payoffs through bundling and tranching. In equilibrium, banks issue money as a liquid but risky liability that circulates in payments as well as an illiquid residual claim. Liquidity creation (i.e., money provision) relies on holding illiquid, risky assets and is enhanced by bundling safe and risky assets. Regulation of bank balance sheets should internalize these monetary provision and risk transformation roles.
Using a mandatory, purpose-designed Census Bureau survey of approximately 28,500 establishments, we provide new evidence on industrial AI adoption in US manufacturing. Despite widespread digitization, only 22.8 percent of plants report any AI use as of 2021; intensity-weighted adoption is far lower. Adoption correlates with more-recent digital infrastructure—cloud computing and predictive analytics—rather than legacy on-premises IT or descriptive analytics. Structured production-process management and size are significant predictors. Cost and lack of applicable use case are the most cited barriers, followed by expertise. Prior productivity does not predict use, pointing to organizational readiness as a key barrier to AI diffusion.
Risky health behaviors are private and rarely observed. This paper introduces a novel dataset of over 170,000 men’s classified advertisements published between 1975 and 1992 in The Advocate, the largest and longest-running national LGBT magazine in the United States. Using high-frequency text data, this paper documents significant changes in the content of ads during the early years of the HIV/AIDS epidemic, most notably a pronounced rise in the take-up of safe sex language following the first reports of AIDS. These findings offer rare insight into how perceived health risks shaped sexual behaviors during the very early years of the epidemic.
