Replication Data and Code for Predicting Cooperative Survival: A First Empirical Test of the Trinity Law (S = T × C × DB) in Community-Based Health Insurance

Version 5.1 (19 July 2026) — corrections to the package. No data file, analysis, or result has changed; the pipeline reproduces the same 49 verification checks, all passing, as version 5. This version resolves the following: scripts/REPLICATE_ALL.R now holds an identical copy of the master script. In version 5 that file contained the replication script of a different paper, and the master script's usage note directed readers to it. The usage note now reads source("REPLICATE_ALL.R"), run from the repository root, matching the quick-start instruction in the README. The EU FP7 grant is identified as HEALTH-F2-2009-223518, project "Community-Based Health Insurance in India"; version 5 printed an incorrect number. LICENSE.txt added, carrying the CC BY 4.0 terms and the suggested citation. docs/FILE_MANIFEST.md documents all 23 data files, the role each plays in replication, and its provenance. It replaces an internal build record carried over from preparation. CODEBOOK.md extended from 9 files to all 23. The pipeline's verification section reports a file inventory: every documented file present, every present file documented. The replication log records the repository folder name in place of the absolute path of the machine that produced it. This repository contains all data and replication code for the paper "Predicting Cooperative Survival: A First Empirical Test of the Trinity Law (S = T × C × DB) in Community-Based Health Insurance," submitted to Annals of Public and Cooperative Economics. The Trinity Law states that the sustainability of voluntary cooperative institutions is the product of three conditions — trust (T), consensus (C), and dual benefit (DB) — implying non-compensability: if any factor is zero, sustainability is zero regardless of the others. This paper reports the first empirical test of the law against field data in any cooperative domain, using community-based health insurance (CBHI) as the test bed. The repository contains 22 data files from two independent studies spanning two countries (India and Nepal) and nine sites: Pillar 1 (EU FP7 Grant 261007, 2009–2013): 9,201 household records, 141 Self-Help Group enrollment records, 96 CHAT deliberation process records, and village-level trust data from three Indian sites and two Nepali districts. Pillar 2 (ECCP baseline, 2005–2006): 4,931 household records from seven Indian sites in four states, temporally and methodologically independent of Pillar 1. A single call to REPLICATE_ALL.R reproduces all computational results reported in the paper and its Statistical Appendix. The replication script runs 49 numerical checks; all pass. No synthetic or fabricated data are present in this repository. All files trace to authenticated field-collected household surveys.