Accountability in Time: Evolution in Access to Information Institutions

• In Event: Agents of Accountability: Monitors and Claimants in Accountability Processes

Abstract

We study how social accountability institutions evolve to be more responsive either to elite or grassroots claimants. We analyze Mexico’s access-to-information system using big data methods complemented by field-based qualitative evidence. We test between two main models of change, each of which features self-reproducing properties. A tendency towards an inclusive grassroots accountability model may result from gradual processes of skill attainment by government bureaucrats and buy-in by citizen users. A tendency toward an elite-biased brokered accountability model may be caused by a process of learning, in which elite claimants deploy resource advantages to navigate the system more effectively and government agents discriminate in favor of elites based on textual clues. We develop measures and tests to characterize different government agencies in different time periods according to these models, and to assess how changes play out over time. Our primary analyses consider nearly two million individual information requests and responses filed under Mexico’s access-to-information law. We combine topic models of the substance of requests, measures of the characteristics of both requests and responses based on expert hand-coding and machine learning approaches and associated meta-data such as requester geographic location and responding federal government agency. We pair these quantitative findings with process tracing based on elite interviews and focus groups with civil society organizations that use public information in their campaigns. We find evidence both of change in response to external political and legal shocks, as well as more gradual processes of learning and drift. In particular, we find evidence of requesters becoming more sophisticated over time and officials becoming more responsive to sophisticated requesters, both consistent with a brokered accountability model.

Authors

• Brian Palmer-Rubin, Marquette University

• Dan Berliner, London School of Economics

• Benjamin E. Bagozzi, University of Delaware

• Aaron S. Erlich, McGill University