Research

We examine the impact of political instability on scientific output. To do so, we analyze the 2013 U.S. federal government shutdown’s impact on on federally funded research in Antarctica. Leveraging a novel dataset with a difference-in-differences design and exploiting the timing of the decision, we document an 11% decline in the number of publications among affected researchers, as well as altered collaboration patterns, which we corroborate with qualitative survey evidence. Together, the results suggest that even brief episodes of short-lived political instability can have enduring deleterious effects on science.

Abstract removed for now because we're eliciting people's priors on the study!

Artificial intelligence is advancing at a pace once thought unimaginable, yet we still lack clear tools to understand how these breakthroughs map onto the world of work. This paper introduces a novel framework that systematically links AI benchmark progress - the scoreboards that track frontier capabilities - to the occupational tasks that define human labor. Unlike patents, surveys, or deployment data, which are often lagged, opaque, or subjective, benchmarks are transparent, replicable, and updated in near real time. Using O*NET as a bridge, we connect benchmark trajectories across domains-including language, reasoning, vision, and multimodal tasks-to 52 human abilities, and translate these into occupation-level indices of AI exposure. The result is a dynamic, task-level methodology that allows us to track and forecast where automation pressures are likely to emerge. By repositioning benchmarks from technical scoreboards to economic indicators, this study offers a fresh lens for anticipating the future of work and shaping policy responses.

How exposed are different occupations to advances in artificial intelligence (AI)? Existing approaches typically infer exposure from patents, expert surveys, or static mappings between AI capabilities and tasks, but each has important limitations. I introduce a complementary framework that aggregates evidence from randomized controlled trials (RCTs) in economics and the social sciences, which directly test the causal impact of AI tools on worker performance. Drawing on RCTs across domains such as writing, coding, forecasting, and management, I map their results to O*NET tasks and then aggregate to the occupational level. The outcome is the 'RCT Exposure Index', a novel, empirically grounded measure of AI exposure that captures heterogeneity across tasks and occupations. By anchoring exposure in experimental evidence rather than proxies, the RCT Exposure Index provides a real-time lens on how AI is reshaping the landscape of work.