Machine Learning in Economic History
An opinionated literature review from the frontier — mapping what ML can and cannot do for economic history, and charting the research agenda.
Status: Active — drafting
Machine learning has entered economic history unevenly: some areas have seen genuine methodological advances, others have adopted tools without the methodological infrastructure to use them well. This paper offers an opinionated review from the frontier — not a neutral survey, but a guide to what works, what doesn’t, and what the field should build next.
The paper positions economic history as a natural home for ML-driven measurement (recovering unobservable historical quantities) while arguing that the hard problems of causal identification remain irreducibly human.
With co-authors.