Data and code for "Gateway or Amplifier? How Climate Finance Readiness Operates Through Divergent Institutional Pathways"
Description
This deposit contains the data and code to reproduce all quantitative results, tables, and figures in the article "Gateway or Amplifier? How Climate Finance Readiness Operates Through Divergent Institutional Pathways: Evidence from Green Climate Fund and Adaptation Fund Recipients." The study examines whether climate finance readiness support activates or substitutes for domestic institutional capacity across countries eligible for Green Climate Fund (GCF) and/or Adaptation Fund (AF) support over 2015-2024. The primary file, readiness_analytical_dataset_vFROZEN_2026-08-06.xlsx (sheet "data"), is a country-level analytical build with one row per country. It contains the access outcomes (participation; approved-project count; total approved project finance), readiness measures (disbursed-plus-closed and closed-only grant funding, USD), and controls: institutional quality (first principal component of three Worldwide Governance Indicators dimensions), GDP per capita, population, the ND-GAIN Index, GCF Direct Access Entity (DAE) and AF National Implementing Entity (NIE) accreditation, and UN region, SIDS, LDC, and FCAS classifiers. Full variable definitions, sources, and transformations are in Appendix A of the article. Saint Kitts and Nevis is excluded (missing ND-GAIN value); South Sudan, the Cook Islands, Niue, and the State of Palestine are excluded from models using the ND-GAIN control because they are unscored by the ND-GAIN Index, giving a frozen analytical sample of N = 137 (129 participating countries for the conditional count and finance models). ND-GAIN provenance: relative to the pre-freeze working data, North Macedonia was corrected from 0 to 52.4736, Cabo Verde from 26.87 to 51.8665, and the four unscored countries were set to missing rather than zero. Also included: cluster_membership_FINAL_vFROZEN.csv (k-means cluster assignments); twelve R scripts (01-12) that regenerate every table and figure; the qualitative coding protocol (S12) and source register (S13) documenting the 481 official documents analysed; and the article figures. The underlying quantitative data derive from publicly available sources: GCF and AF project and readiness databases, the World Bank Worldwide Governance Indicators, the ND-GAIN Country Index, the World Bank World Development Indicators, and UN-OHRLLS classifications. The third-party qualitative documents are cited with access dates but are not redistributed for copyright reasons.
Files
Steps to reproduce
Software: R (tested with version 4.5.1). Required packages: readxl, dplyr, tidyr, MASS, MatchIt, sandwich, lmtest, pscl, logistf, fpc, cobalt, modelsummary, flextable, e1071, ggplot2, svglite. Each script installs any missing packages on first run. Setup: place all deposit files in a single working directory and start R there (or setwd() to it). Every script reads readiness_analytical_dataset_vFROZEN_2026-08-06.xlsx from the working directory, so no file paths need editing. Fixed random seeds are set in the propensity-score-matching and clustering scripts, so matched pairs and cluster assignments reproduce exactly. Run the scripts in numbered order: 01_main_results_reproduce.R - Table 1 (descriptives); baseline hurdle model, native scale (Table 3); propensity-score-matching ATT and expanded-covariate check (Tables 4 and S11); two-stage residual inclusion diagnostic (Table 5); k-means cluster sizes (Table S4); case-country institutional-quality values (Table 2). 02_table3_standardized_hurdle.R - Table 3, standardized panel (= Table S6). 03_supplementary_S1_S10.R - Supplementary Tables S1-S5 and S8-S10. 04_supplementary_S7_firth.R - Table S7 (Firth penalized participation logit). 05_psm_expanded_S11.R - Table S11 (expanded-covariate PSM). 06_clusters_2SRI.R - cluster membership (Figure 4, Table S4), case-country IQ (Table 2), 2SRI diagnostic. 07_appendixC_mediation.R - Appendix C mediation (N = 137). 08_figure2_balance.R, 09_figure3_ATT_forest.R, 10_figure4_cluster.R - Figures 2, 3, and 4. 11_figureB1_distributions.R - Figure B1. 12_appendixB_vif_correlations.R - Appendix B, Table B1 (VIF) and Table B2 (Spearman correlations). Expected results: the analytical sample is N = 137 (129 participants). Readiness is positively and significantly associated with project count (breadth) but not with first-time participation (not identifiable in this near-universal-participation sample) or, robustly, with project finance (depth). K-means cluster sizes are 40 / 44 / 53 by capacity.
Institutions
- Korea UniversitySeoul, Seoul