Research
Writing sample
“The Effect of Artificial Intelligence Exposure on Labor Productivity”
- Two-way fixed-effects panel of 59 U.S. industries, 1997–2024 (1,649 observations)
- Builds a time-varying AI Industry Exposure index from annual occupational employment shares
Abstract
Does exposure to artificial intelligence raise labor productivity? I study 59 U.S. industries from 1997 to 2024, combining a time-varying measure of industry exposure to AI, built from occupational ability scores and annual employment shares, with an economy-wide index of AI capability. In two-way fixed-effects regressions, neither early-stage (2016–2019) nor generative-AI-era (2020–2024) exposure has a statistically significant effect on labor productivity. The preferred estimate for the generative-AI era is 0.023 (standard error 0.364), and the confidence intervals include both sizable gains and sizable losses. Because most of the identifying variation comes from 2022–2024, the results show that any productivity effect of generative AI was not yet detectable in industry data through 2024, not that no effect exists. Significant positive estimates appear only in specifications without year fixed effects or with a lagged dependent variable, which I argue are not credible.
Work in progress
“Code Enforcement, Housing Violations, and Economic Outcomes in New York City”
Examines the economic impact of housing code enforcement on landlords and corporate property owners, to inform tax and penalty policy that improves housing quality, tenant outcomes, and accountability. Uses over 2.65 million NYC enforcement records (2000–2023); DCWP issued about $1.44 billion in penalties, of which an estimated 42.6 percent was collected.
Other research
“Determinants of Renewable Energy Adoption: A Global Panel of 100+ Countries”
Builds a multi-source global panel and applies causal inference methods to identify the policy, economic, and institutional drivers of renewable energy adoption in emerging and developed markets.