Media Regulatory Capacity in Chile and Argentina, 1983–2026: An Episode-Level Dataset

This dataset records institutional episodes in the Chilean (11 March 1990 – 11 March 2026) and Argentine (10 December 1983 – 11 March 2026) media regulatory regimes, each coded by the state of regulatory capacity it produced. It supports a companion article, under review, on authoritarian constitutional legacies and media regulatory capacity in Chile and Argentina. The unit of analysis is the institutional episode: a dated, attributable formal act that creates, modifies, validates, annuls or attempts to modify a rule of the media regime (statutes, constitutional amendments and drafts, decrees, constitutional and judicial rulings, administrative acts, and bills that reached a formal vote or archive). Each episode is assigned to one or more of five regime components (spectrum and licensing; ownership and concentration; the public broadcaster; content and pluralism obligations; the digital perimeter) and coded in one of four capacity states: operative; formal or declarative; validated but unexecuted; annulled or reduced. Antecedent (pre-window) and post-window rows are included for context and flagged; they are excluded from counts. A separate environmental panel records concentration, audience, advertising, public-funding and territorial-coverage indicators, kept apart from the episode count by design. Files: media_regulatory_capacity_CL_AR_v0.9.xlsx (six sheets: README, Coding_scheme, Episodes, Summary with formula-based counts, Environmental_panel, Sources); episodes_CL_AR_v0.9.csv (UTF-8 export of the Episodes sheet); README_dataset_v0.9.md. The dataset contains 91 rows (58 Chile, 33 Argentina), of which 80 fall within the analytical windows and 49 carry a capacity-state code; 37 environmental indicators; 94 sources, including primary legal texts, constitutional and judicial rulings, regulator records and registers, and official reports. Version 0.9 is a draft for review. Rows carrying a verification_flag contain a date, figure or classification still to be confirmed in a primary source; version 1.0 will close those flags. Known discrepancies between secondary sources are recorded in the README and in the relevant rows. Coding was performed by a single researcher; every row carries a field for evidence against its assigned state so that each classification can be examined. Generative AI tools (Anthropic Claude Opus 5 and Claude Fable 5.1) were used to assist in compiling and formatting the dataset from the primary sources cited in it and in cross-checking citations; all coding decisions were taken and verified by the author, who is solely responsible for the content.