(iPoster) Feasibility and Feedback: Public Comments and Oversight in U.S. Rulemaking

• In Event: Poster Sessions: IR Theory; Experiments; Presidents and Executive Politics; & Political Cognition, Crisis, and Conflict Attitudes
  In Poster Session: Poster Session: Presidents and Executive Politics

Abstract

Federal agencies generate the bulk of binding policy via administrative rules, and they exercise substantial discretion between proposing and finalizing rules during notice and comment. We develop a three actor model—President, agency, and interest groups—in which OIRA acts as the President’s proxy. OIRA imposes a presidential feasibility set: a proposal or final rule can advance only if it is at least as close to the President’s ideal as the status quo. Public comments create a no neglect band from the comment record; rules that fall outside this band are more likely to be vacated in court. Anticipating both constraints, the agency chooses (i) the location of the proposal in policy space and (ii) the length of the comment window. Model results suggest: (1) re centering—policy placement shifts toward the President’s ideal after administration changes; (2) responsiveness—final rules move toward the comment median when dispersion is low, with weaker pull under high polarization; and (3) endogenous procedure—agencies lengthen comment windows when the expected intersection of the feasibility set and the no neglect band is small (i.e., neglect risk is high). We evaluate these predictions using 22,336 proposed rules (2012–2025) from 39 U.S. agencies, including the full text of each proposal, all public comments (texts and counts), and the length of the comment period. Using large language models, we embed proposal texts, comment texts, and measures of presidential ideal points in a shared policy space to compute distances, comment medians, and dispersion. Empirically, we find: proposals re center toward new presidential ideals at transitions; agencies extend comment windows when predicted neglect risk (large proposal–comment gaps and/or high dispersion) is greater; and—where post comment outcomes are observable—final rules shift toward the comment median much more when dispersion is low compared to when dispersion is high. These patterns validate the model’s predictions.

Authors

• Fred Gui, Louisiana State University

• Muzhi Liu, Columbia University