Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence (with Shakked Noy)
Science (2023) [Tweet]

Press: [MIT News] [Nature] [NYT] [TechReview] [Vox] [Wired] [WSJ 1, 2]


We examine the productivity effects of a generative artificial intelligence technology—the assistive chatbot ChatGPT—in the context of mid-level professional writing tasks. In a preregistered online experiment, we assign occupation-specific, incentivized writing tasks to 444 college-educated professionals, and randomly expose half of them to ChatGPT. Our results show that ChatGPT substantially raises average productivity: time taken decreases by 0.8 SDs and output quality rises by 0.4 SDs. Inequality between workers decreases, as ChatGPT compresses the productivity distribution by benefiting low-ability workers more. ChatGPT mostly substitutes for worker effort rather than complementing worker skills, and restructures tasks towards idea-generation and editing and away from rough-drafting. Exposure to ChatGPT increases job satisfaction and self-efficacy and heightens both concern and excitement about automation technologies.

Improving Local Labor Market Definitions Using the Louvain Community Detection Algorithm
Economics Letters (2022)


Well-defined commuting zones are essential for accurate research on US local labor markets. To develop commuting zones, one must construct edge weights – a measure of commuting flows between counties – and then use the edge weights to partition counties into clusters. I improve upon currently used “ERS” commuting zones in two ways. First, it is unclear if ERS commuting zones use the best edge weights. Therefore, I test multiple edge weights. Second, the algorithm to produce ERS commuting zones requires specifying a theoretically-unguided cutoff parameter; results may be sensitive to the parameter choice. Instead, I use the Louvain algorithm, which optimizes for “modularity”, a graph-intrinsic parameter that is greater when there is higher intra-commuting zone flow and lower inter-commuting zone flow. I call my new delineations “TS Louvain”, which uses the ERS commuting flow definition to construct edge weights, and “Sum Louvain”, which uses the total number of commuters as edge weights. Compared to ERS, TS Louvain and Sum Louvain have 0.05 to 0.15 greater modularity, Sum Louvain has a 0.01 to 0.02 higher share of people who work and live in the same commuting zone, and in a case study, TS Louvain produces greater estimates and t-statistics. These metrics suggest that these new commuting zones improve upon the existing delineations. Researchers can access these commuting zone definitions at

Working Papers

Interpreting TSLS Estimators in Information Provision Experiments (with Vod Vilfort)

May 2024, Arxiv


To estimate the causal effects of beliefs on actions, researchers often conduct information provision experiments. We consider the causal interpretation of two-stage least squares (TSLS) estimators in these experiments. In particular, we characterize common TSLS estimators as weighted averages of causal effects, and interpret these weights under general belief updating conditions that nest parametric models from the literature. Our framework accommodates TSLS estimators for both active and passive control designs. Notably, we find that some passive control estimators allow for negative weights, which compromises their causal interpretation. We give practical guidance on such issues, and illustrate our results in two empirical applications.

Older Work

Effect of Public Library Access on K-12 English Language Arts Performance
MIT Undergraduate Economic Journal (2021)


Does public library access affect K-12 students’ English Language Arts performance? Using data from the California Standardized Testing and Reporting exam and the Institute of Museum and Library Services, this paper estimates the impact of having a public library nearer or farther from a school. I measure that for schools that do not have a public library within a 10 mile radius, introducing a public library within 5 miles increases the percentage of students that are proficient in English Language Arts in a school by around 7 percentage points. However, for schools that already have a library nearby, marginal changes in distance of a school to a library have little to no effect. Combined with other empirical specifications, these results suggest that introducing a nearby public library is an effective public investment for schools very distant from a library, but ineffective otherwise.

Wirtschaftswunder: Incentives Aligned for Growth — A Comparative Study of West Germany and the UK 1950-1980

MIT Undergraduate Economic Journal (2020)


I analyze post-war growth in Germany and the UK. I argue that the strong growth of Germany and stagnant growth of the UK can be attributed to their degree of alignment of employee, management, and bank incentives, and the ability of agents to check each others’ rent-seeking. These were much better aligned in Germany than in the UK, leading to better worker productivity, financing, and longer-term decision-making, driving Total Factor Productivity (TFP) and capital growth.