Dataset Supporting: When prudential regulation fails: technical reserves, hidden operational insolvency and financial fragility in Colombia's social health insurance system, 2019–2025

This repository contains the analytical dataset and replication materials used for the study evaluating financial resilience, technical reserve dynamics, and prudential regulation performance among Colombian health insurers (EPS) between 2019 and 2025. The database includes insurer-level longitudinal information compiled from publicly available administrative, financial, and regulatory sources in Colombia. Variables included in the dataset cover operational margins, liquidity ratios, solvency indicators, medical loss ratios (MLR), provider debt per affiliate, technical reserve composition, equivalent affiliates, and reserve burden indicators such as Technical Reserves relative to Operational Burden (TROB). The repository additionally includes: Data dictionaries and variable definitions Statistical replication scripts Event-study model specifications Threshold spline analysis scripts Robustness and sensitivity analyses Supplementary tables and figures used in the manuscript Data sources Data were compiled and harmonised from: Colombian National Health Superintendence (Superintendencia Nacional de Salud) Ministry of Health and Social Protection of Colombia ADRES administrative databases Public insurer financial statements Regulatory and prudential reporting systems Study design The dataset supports: Longitudinal panel analyses Fixed-effects regression models Difference-in-differences estimations Event-study analyses Threshold spline models Bootstrap sensitivity analyses Time coverage 2019–2025 Geographic coverage Colombia Unit of analysis Health insurer-year observations (EPS-year panel dataset) File contents panel_dataset.csv data_dictionary.xlsx replication_scripts.do / replication_scripts.R supplementary_materials.pdf README.txt Reproducibility statement All analytical procedures reported in the manuscript can be reproduced using the data and scripts provided in this repository. The dataset was anonymised and aggregated at insurer level prior to publication.