Political Language in Economics.

This article investigates whether academic writing in economics reflects the political orientation of economists by applying machine learning techniques to a large corpus of economics research articles. Using observed political behavior—measured through campaign contributions and petition signings—linked to economists’ published texts, the authors develop a predictive model of partisanship that achieves significant out-of-sample accuracy and aligns with survey-based ideology measures. They find substantial sorting of economists by predicted ideology across research fields, journals, and institutions, and demonstrate that predicted political leanings correlate with empirical estimates of policy-relevant economic parameters, such that the implied policy prescriptions align with partisan preferences. While causality cannot be established, the study suggests that political preferences are detectable in ostensibly non-partisan academic economics and may influence empirical findings relevant to public policy.