How did the US become a land of opportunity? Previous historical research on intergenerational mobility has focused on father-son income correlations, masking the role of mothers. We introduce a new mobility measure that incorporates both parents' human capital, develop a latent variable method leveraging literacy as a proxy, and construct a representative linked panel that includes women. We find that intergenerational mobility—in both human capital and income—rose sharply from the 19th to the 20th century. Initially, maternal human capital was most predictive of children's outcomes. However, as schooling expanded, this reliance declined and intergenerational mobility rose. America's investment in mass education has therefore been central to its rise as a mobile society.
This article studies the long-run effects of slavery and restrictive Jim Crow institutions on Black Americans' economic outcomes. We track individual-level census records of each Black family from 1850 to 1940 and extend our analysis to neighborhood-level outcomes in 2000 and surname-based outcomes in 2023. We show that Black families whose ancestors were enslaved until the Civil War have considerably lower education, income, and wealth than Black families whose ancestors were free before the Civil War. The disparities between the two groups have persisted substantially because most families enslaved until the Civil War lived in states with strict Jim Crow regimes after slavery ended. In a regression discontinuity design based on ancestors' enslavement locations, we show that Jim Crow institutions sharply reduced Black families' economic progress in the long run.
High-income business service workers dominate the economies of major US cities, and their spending supports many local consumer service jobs. As a result, business services' high remote work potential poses a risk to consumer service workers who could lose an essential source of revenue if business service workers left big cities to work from elsewhere. We use the COVID-19-induced increase in remote work to provide empirical evidence for this mechanism and its role in shaping the pandemic's economic impact. Our findings have broader implications for the distributional consequences of the transition to more remote work.
Artificial intelligence (AI) reshapes workers' comparative advantage by altering the tasks they perform and the skills those tasks require. We develop a dynamic task-based model to quantify the general-equilibrium effects of task-specific technical change. Workers have multidimensional skills, choose occupations, and accumulate skills on the job; occupations combine tasks, and productivity depends on how workers' skills match task requirements. We develop a computationally efficient procedure to estimate the model using panel data and a new database of task-level skill requirements. We apply the model to AI, allowing it to augment, automate, and simplify tasks. We find that AI narrows wage inequality and raises average wages across scenarios ranging from slow to rapid AI progress. The key equalizing force is simplification: by lowering tasks' skill requirements, AI lets lower-skill workers compete for previously inaccessible jobs. Adoption costs, highest for lower-skill workers, dampen but do not eliminate the decline in inequality.
The World War II GI Bill was the largest education subsidy in US history and a cornerstone of the postwar US transition to a knowledge economy. Although formally race-blind, the program's decentralized administration left implementation to local officials and segregated institutions, with sharply different consequences for Black and white veterans. This paper quantifies the GI Bill's impact on Black and white Americans' economic outcomes across two generations, using a regression discontinuity around WWII service eligibility cutoffs and a new data linkage from veterans in the 1940 and 1950 censuses to their sons' outcomes in the 2000s. The GI Bill widened racial inequality, doubling white veterans' college completion while steering Black veterans into often-fraudulent vocational programs with no earnings returns. The disparities persisted across generations, increasing the white-Black gap in sons' adult-neighborhood outcomes, including a 5-percentage-point (47 percent) widening of the racial college gap. Unequal returns to the same eligibility account for the intergenerational gap, with no contribution from prewar differences in socioeconomic status or geography. In sum, access to the GI Bill was not nearly the same economic opportunity for Black Americans as it was for white Americans, highlighting that race-blind policy does not guarantee racial equality.
Two Steps Forward, One Step Back: Racial Income Gaps among Women since 1950
This paper studies the evolution of Black-white income gaps among women since 1950. I document that the gap in incomes of Black and white women narrowed substantially in the 1960s, around the end of Jim Crow. While the South was the epicenter of racial inequality during Jim Crow, its Black-white gaps have since converged with other regions. The improvements in the Black-white gap were shared among Black women across the income distribution. However, there were two distinct drivers. At the bottom and middle of the income distribution, the pre-1980 compression of the distribution's lower tail narrowed the Black-white gap. In contrast, Black women at the top experienced a substantial improvement in the rank they occupied in the white distribution, narrowing the Black-white gap despite rising inequality at the distribution's higher tail.
We model the economics of recursive self-improvement (RSI) and assess its plausibility and impacts. First, we build a sequence of increasingly rich models of AI progress to highlight the feedback loops behind RSI. We represent our models as directed graphs and show that net acceleration in AI capabilities depends on the product of elasticities across each feedback loop. Second, we distinguish between "narrow" and "broad" AI capabilities, capturing the possibility that AI systems improve narrowly at optimizing AI R&D benchmarks without improving at broader economically valuable tasks. Third, we document existing estimates of key parameters, and provide a wish list of empirical objects that AI companies can measure and feasibly share publicly. Finally, we calibrate the model with existing data. A back-of-the-envelope calculation suggests that feedback loops are not currently strong enough to generate a self-sustaining acceleration, though they appear to be strengthening. We conclude by assessing the plausibility and implications of such an acceleration.
A task-level database of scientific research: a comprehensive map of what researchers actually do, broken down by domain, field, subfield, and topic. SciNet enables rigorous, task-level analysis of scientific work by mapping the granular activity structure of science across 5 domains, 30 fields, 302 subfields, and 4,516 topics, with 26,371 released task statements.
Despite large estimated productivity gains from AI across a wide range of tasks, the labor market effects are not well understood. This paper develops a methodology to estimate workers' comparative advantage across tasks and the labor market effects of task-biased technical change. We propose and estimate a dynamic general equilibrium model in which workers build multi-dimensional skills and can switch occupations based on their evolving comparative advantage. The model is informed by existing detailed data on tasks in the US labor market and new data on their skill requirements.
This dataset comprises crosswalks to link census records for men and women in the United States from 1850 to 1950. It covers 42 million Americans who are linked across 186 million census records. The dataset tracks individuals across multiple censuses despite name changes (e.g., due to marriage) by combining historical census records with Social Security Number application data. The resulting panels are representative, making them particularly valuable for including women in the study of intergenerational mobility.
