AI reproduces economics papers, with discrepancies flagged in nearly 80%
An NBER working paper says an LLM can reproduce economics research at scale; flagged discrepancies are automated detections, not proof of error.
An NBER working paper used a large language model to reproduce economics papers in bulk, flagging discrepancies in 3,460 of 4,452 published replication packages.
The workflow was built by economists Matthew Schwartz, Isaiah Andrews and Jesse M. Shapiro. It first reproduces the original calculations, then checks them against published findings and runs automated sensitivity analysis. The authors report that 496 papers saw calculations sped up by more than a factor of 10, and 923 received extensions consistent with the original aims.
The paper is NBER working paper w35782; Tyler Cowen covered it on Marginal Revolution on October 1. The flagged "discrepancies" are the workflow's automated detections, not proof the original papers were wrong, and the paper has not been peer reviewed.
Sources:https://www.nber.org/papers/w35782https://marginalrevolution.com/marginalrevolution/2026/10/merging-llms-and-economics-research.html