Diaspora Remittance Leverage as Institutional Reform: Natural Experiment Evidence from Hispaniola

Here's a comprehensive Zenodo description explaining all files and their use in Google Colab:

Replication Materials: Haiti-Dominican Republic Natural Experiment

Overview

This repository contains complete replication materials for "Shared Island, Divergent Institutions: The Dominican Republic as Counterfactual Evidence for Legal Infrastructure Investment in Postcolonial Civil Law Systems." All analyses are reproducible in Google Colab with zero local installation required.

Repository Contents

📊 Primary Data

• IPRI- 2007 - 2025 data[25].xlsx (437 KB)
  Complete IPRI dataset (2007-2025) for 144 countries including Haiti, Dominican Republic, and all comparator nations. Contains overall IPRI scores plus component indices (Legal Protection [LP], Physical Property Rights [PPR], Intellectual Property Rights [IPR]).

📓 Analysis Notebooks (Google Colab Ready)

1. HISPANIOLA.ipynb (973 KB) - Main Analysis Notebook
Core statistical analysis including:

• Country-specific trajectory analysis with R² values
• Beta/sigma convergence tests
• Rank stability analysis (Spearman correlation)
• Descriptive statistics and summary tables

How to use in Colab:

# 1. Upload to Google Colab
# 2. Mount Google Drive or upload IPRI xlsx file
# 3. Run all cells sequentially
# Generates: summary_statistics.csv, dr_haiti_comparison_data.csv

2. DIAGRAMS.ipynb (879 KB) - Publication Figures
Creates publication-quality figures for manuscript:

• Figure 1: IPRI Divergence trajectories (2007-2025)
• Figure 2: PPR Collapse comparison
• Regional comparison visualizations
• Dominican-Haiti side-by-side comparison

How to use in Colab:

# Requires: IPRI xlsx file, summary_statistics.csv
# Generates: Figure1_IPRI_Divergence.png/.pdf, Figure2_PPR_Collapse.png/.pdf
# All figures saved in publication-ready format (300 DPI)

3. ADVANCED.ipynb (1.05 MB) - Econometric & ML Analysis
Advanced methodological analyses:

• K-means clustering (k=4, silhouette validation)
• Machine Learning (Random Forest, Gradient Boosting)
• Unit root tests (ADF, KPSS)
• Synthetic control method (2011 treatment)
• Difference-in-differences (with limitations note)
• Component mechanism decomposition (LP/PPR/IPR)

How to use in Colab:

# Most comprehensive notebook - runs ALL analyses
# Requires: IPRI xlsx file
# Runtime: ~3-5 minutes
# Generates: All PNG figures + haiti_dr_analysis_summary.csv

📈 Generated Figures

Publication Figures (Manuscript):

• Figure1_IPRI_Divergence.png/.pdf - Main trajectory comparison
• Figure2_PPR_Collapse.pdf/.png - Physical property rights component breakdown

Supplementary Analysis Figures:

• haiti_dr_trajectories.png - Linear trends with R² values
• convergence_tests.png - Beta/sigma/rank stability (3-panel)
• clustering_analysis.png - K-means results with elbow method
• ml_analysis.png - Random Forest/Gradient Boosting performance
• synthetic_control.png - Counterfactual DR construction
• did_analysis.png - Difference-in-differences visualization
• mechanism_decomposition.png - LP/PPR/IPR gap breakdown
• regional_comparison.png - Caribbean & Latin America context
• dominican_haiti_comparison.png - Detailed side-by-side

📋 Data Tables (CSV)

• summary_statistics.csv - Descriptive statistics by country
• dr_haiti_comparison_data.csv - Year-by-year comparison dataset
• haiti_dr_analysis_summary.csv - Complete analysis results table

📖 Documentation

• README.md - Detailed methodology, data sources, variable definitions

🚀 Quick Start in Google Colab

Option 1: Run Complete Analysis (Recommended)

# 1. Open ADVANCED.ipynb in Google Colab
# 2. Upload IPRI- 2007 - 2025 data[25].xlsx when prompted
# 3. Run all cells (Runtime > Run all)
# 4. Download all generated figures from Files panel

Option 2: Reproduce Manuscript Figures Only

# 1. Open DIAGRAMS.ipynb in Google Colab
# 2. Upload IPRI xlsx file
# 3. Run all cells
# 4. Download Figure1 and Figure2 (PNG/PDF)

Option 3: Basic Descriptive Analysis

# 1. Open HISPANIOLA.ipynb in Google Colab
# 2. Upload IPRI xlsx file
# 3. Run all cells
# 4. Download CSV tables

📦 Dependencies (Auto-Installed in Colab)

All notebooks include automatic package installation:

• pandas, numpy - Data manipulation
• matplotlib, seaborn - Visualization
• scipy, statsmodels - Statistical tests
• scikit-learn - Machine learning & clustering
• linearmodels - Panel data econometrics
• arch - Unit root tests
• openpyxl - Excel file reading

No manual installation required - all dependencies install automatically when you run the first cell.

🔬 Methodological Notes

Data Availability

• Haiti: 2013-2025 (11 years) - IPRI range: 2.13-3.40
• Dominican Republic: 2007-2025 (19 years) - IPRI range: 3.96-5.03
• Treatment year: 2011 (Ciudad Juan Bosch Law 189-11)

Key Limitations

1. DID Constraint: Haiti data begins 2013, so no pre-2011 overlap for true difference-in-differences. We use synthetic control method as primary causal inference strategy.
2. Sample Size: N=16 countries for clustering (14 comparators + 2 target countries).
3. Unit Roots: Both Haiti & DR exhibit non-stationary time series (ADF p>0.05), consistent with systematic trends.

Reproducibility

• All random processes use random_state=42 for exact replication
• All figures saved at 300 DPI for publication quality
• Complete analysis runtime: ~3-5 minutes on Google Colab free tier

📊 Key Findings (Replicated from Notebooks)

Trajectory Analysis:

• Haiti: -0.0563 points/year (R²=0.541, p=0.0099) - 37% decline
• Dominican Republic: +0.0318 points/year (R²=0.359, p=0.0067) - 17% improvement
• Divergence rate: 0.0881 points/year (~0.88 points per decade)

Clustering:

• Haiti assigned to unique Cluster 3 (alone, worst trajectory)
• DR assigned to Cluster 2 (with Costa Rica, Panama, Jamaica)
• ANOVA F=18.01, p<0.001 (highly significant)

Synthetic Control:

• Treatment effect: +0.050 IPRI points (post-2011)
• Optimal weights: Nicaragua (42%), El Salvador (29%), Mexico (20%), Chile (8%)

Machine Learning:

• Test R² = 0.988 (Gradient Boosting)
• Feature importance: LP (47%) > IPR (38%) > PPR (14%)

📧 Contact & Citation

For questions about replication or data:

• Scott Brown (Universidad de Puerto Rico): scott.brown@upr.edu
• Adam Welker (Utah Valley University)
• Paul LaTortue (Université de Notre Dame d'Haiti)
• Theany Calderon-Abreu (Universidad de Puerto Rico)

Suggested Citation:

Welker, A., LaTortue, P., Calderon-Abreu, T., & Brown, S. (2025). 
Shared Island, Divergent Institutions: The Dominican Republic as 
Counterfactual Evidence for Legal Infrastructure Investment in 
Postcolonial Civil Law Systems [Data set]. Zenodo. 
https://doi.org/10.5281/zenodo.17695109

📜 License

All materials licensed under Creative Commons Attribution 4.0 International (CC BY 4.0).
You are free to share and adapt with appropriate attribution.

Copyright © 2025 The Authors.

🔗 Related Materials

• Preprint: [Link when available]
• Journal submission: Constitutional Political Economy (under review)
• Companion paper: 17-country Spanish civil law panel analysis
• Code repository: All notebooks maintained on Zenodo with version control

This repository provides complete transparency for peer review and enables exact replication of all analyses, figures, and tables in the manuscript.