Knowledge that Transforms

This paper documents the bias introduced by attrition of individuals from administrative data with an application to the labor market consequences of postsecondary education. Attrition due to crossstate migration is non-trivial, particularly for high-earners, graduates from selective universities, and certain majors. Consequently, the premium associated with graduating from a most selective university is 23% higher than in-state earnings suggests, though this magnitude differs across context. The impact of obtaining a 2-year CTE credential is also understated, as are earnings differences across majors. Differences in missingness are systematically related to bias in measurement; we evaluate approaches to quantifying that bias.

We study how public school teachers use paid sick leave. Most US sick leave schemes operate as individualized credit accounts: Paid leave is earned, and unused leave accumulates. We construct a unique dataset of daily leave balances and behavior among 982 teachers for 2010–2018. Sick leave use increases during flu season, and evidence indicates that the average teacher does not use sick leave for leisure, though some subsets of teachers (e.g., the young and inexperienced) do. Usage increases with leave balance; the elasticity is around 0.4. Further, teachers with higher balances are less likely to work sick, particularly during flu season.

We develop, validate, and quantify a tractable model of multinational firms that connects multinational production (MP) with foreign direct investment (FDI). Firms choose where to produce and how to finance the production. They can access external finance, but capital market imperfections prevent them from relying exclusively on it for affiliate production, giving rise to FDI. The model rationalizes the three-way relationship between MP, FDI, and financial market conditions that we document and leads to novel welfare implications. Quantification of the model highlights the relevance of these welfare implications and the importance of financial factors in shaping the activities of multinationals.

This study examines the impact of school choice on academic achievement. I use differences in the number of schools across similar Romanian towns, generating variation in school choice for local students, who compete for seats via test scores. I find that more school choice results in increased sorting of students by admission scores across different schools. Sorting widens achievement gaps between high- and low-admission score students. High-scorers having access to better teachers and peer effects are the primary factors explaining these widening gaps. Last, between-school competition via school choice does not increase average achievement levels.

We show that firms that face higher uncertainty adjust their investment less in response to monetary policy shocks. We find corroborating evidence of this differential effect from firm-level stock returns on FOMC announcement days. Our results are consistent with a real options (or wait-and-see) channel whereby higher uncertainty dampens the response to changes in business conditions. Consistent with this mechanism, the dampening effect is stronger for firms that face higher reversibility.

We present a new data set we built based on Swiss rich lists going back to 1989. We show, among other things, that 60% of the super-rich are heirs—a fraction twice as large as in the United States—and that wealth mobility at the very top has declined significantly. We find that top 0.01% wealth shares are higher than what previous estimates based on wealth tax statistics suggest. At the same time, we argue that rich list data lead to overestimating wealth inequality. Although rich lists are valuable to study the super-rich, we recommend using reported wealth figures with caution.

Aggregate and sectoral effects of public investment crucially depend on the interaction between the output elasticity to public capital and intermediate inputs. We uncover this fact through the lens of a New Keynesian production network. This setting doubles the socially optimal amount of public capital relative to the one-sector model without intermediate inputs, leading to a substantial amplification of the public-investment multiplier. We also document novel sectoral implications of public investment. Although public investment is concentrated in far fewer sectors than public consumption, its effects are relatively more evenly distributed across industries. We validate this model implication in the data.

Migrants shape market access: They reduce international trade frictions and they affect the geographical location of demand. This article incorporates both effects in a model of inter- and intranational trade and migration calibrated to U.S. states. It estimates the elasticity of exports and imports to migrants and shows that reducing U.S. migrant population shares to 1980s levels would increase import (export) trade costs by 7% (2.5%) and decrease U.S. natives’ real wages by more than 2%. States with higher exposure to migrant consumer demand than to migrant labor competition would suffer more, as would states with higher export and import exposure.

P-hacking is prevalent in reality but absent from classical hypothesis-testing theory. We therefore build a model of hypothesis testing that accounts for p-hacking. From the model, we derive critical values such that, if they are used to determine significance, and if p-hacking adjusts to the new significance standards, then spurious significant results do not occur more often than intended. Because of p-hacking, such robust critical values are larger than classical critical values. In the model calibrated to medical science, the robust critical value is the classical critical value for the same test statistic but with one-fifth of the significance level.

I investigate the social costs imposed by poor implementation of public infrastructure. Focusing on the period from 2005 to 2015 in Peru, when the government embarked on a nationwide initiative to expand sewerage systems, I leverage quasirandom variation in initiation of the implementation phase. By combining several sources of administrative data, I find that infrastructure development increased infant and under-5 mortality. These effects are driven by health and safety hazards associated with construction work, leading to increased deaths from accidents and waterborne diseases. The severity of these effects is more pronounced in areas where construction activity was more intense.