Stable But Escapable: Heterogeneous Property Rights Trajectories in Spanish Civil

Dataset Description This dataset accompanies "Symbolic Institutional Traps: Institutional Divergence in Spanish-Speaking Latin America" (under review at Constitutional Political Economy). It provides comprehensive panel data analysis of property rights institutions across 17 Spanish-speaking Latin American countries from 2007-2025, using the International Property Rights Index (IPRI). Version 2 (November 2025) includes substantial expansion in response to peer review: panel regression analysis, trajectory clustering, and temporal dynamics that demonstrate institutional divergence within the same legal origin family. What's New in Version 2 Expanded Analysis Panel data coverage: 17 countries × 19 years = 319 observations Fixed effects regressions: Pooled OLS, country fixed effects, period effects Trajectory clustering: K-means and hierarchical clustering identify 4 distinct paths Country-specific time trends: Annual growth rates for each country Statistical tests: Variance ratio tests for divergence, ANOVA for cluster differences Key Findings Four trajectory clusters identified: Escapers: Uruguay (+38%), Paraguay (+39%), Costa Rica (+18%), Dominican Republic (+17%) Decliners: Venezuela (-29%), El Salvador (-15%), Mexico (-4%) High-Trapped: Chile (stable at ~6.0) Low-Trapped: Argentina, others (stable at mid-levels) Significant divergence: Variance increased substantially 2007-2025 Legal origins determinism falsified: Same legal family (Spanish civil law), radically different outcomes Files in This Dataset Panel Analysis Outputs (Version 2) Data Files (CSV) institutional_changes_2007_2025.csv (17 rows) Country-level changes between 2007 and 2025 Columns: country_code, ipri_2007, ipri_2025, ipri_change, ipri_pct_change, ppr_2007, ppr_2025, ppr_change, ppr_pct_change country_specific_slopes.csv (17 rows) Annual growth rates estimated from country-specific time trends Columns: country, slope (annual change), annual_change, total_change_2007_2025, r_squared cluster_membership.csv (17 rows) Trajectory cluster assignments for each country Columns: Country, Cluster (Escapers/Decliners/High-Trapped/Low-Trapped), Total_Change_2007_2025, Pct_Change, Annual_Slope, Mean_Level, Volatility Regression Results panel_regression_summary.txt Complete output from all panel regression models Models: Pooled OLS, Fixed Effects (time trend), Fixed Effects (period dummies), PPR Fixed Effects Includes coefficients, standard errors, t-statistics, R-squared values Figures (PNG, 300 DPI) figure1_temporal_dynamics.png Two-panel plot showing IPRI and PPR trajectories for all 17 countries (2007-2025) Use for: Visualizing divergence over time figure2_change_scatter.png Scatter plot of 2007 vs 2025 IPRI scores with 45-degree reference line Use for: Showing which countries improved/declined relative to starting point figure3_country_slopes.png Horizontal bar chart of annual growth rates by country Use for: Comparing rates of institutional change figure4_optimal_clusters.png Two-panel plot: Elbow curve and silhouette scores for k-means clustering Use for: Justifying choice of k=4 clusters figure5_cluster_analysis.png Four-panel visualization: Level vs Change, Slope vs Volatility, Start vs End positions, Average % change by cluster Use for: Understanding cluster characteristics figure6_dendrogram.png Hierarchical clustering dendrogram showing country relationships Use for: Validating k-means results with alternative method figure7_trajectories_by_cluster.png Four-panel plot showing temporal trajectories grouped by cluster Use for: Visualizing within-cluster similarity and between-cluster differences Version 1 Files (April 2025) Agricultural_Land_GINI_Coefficients.xlsx - Land inequality data Clustered_Countries.xlsx - Preliminary clustering results Correlation_Matrix.xlsx - Bivariate correlations Descriptive_Statistics_Spanish_Latin_America.xlsx - Summary statistics heat_map.png - Correlation heatmap IPRI_with_Spanish_Speaking_Dummy.xlsx - IPRI data with Spanish dummy Uruguay_vs_Venezuela_Comparison.xlsx - Extreme case comparison VDEM_with_Spanish_Speaking_Dummy.xlsx - V-Dem democracy indicators WGI_with_Spanish_Speaking_Dummy.xlsx - World Governance Indicators Complete Replication Instructions for Google Colab Prerequisites Google account (for Google Colab access) The IPRI dataset: IPRI- 2007 - 2025 data[25].xlsx (available from Property Rights Alliance or contact authors) Basic familiarity with Python (helpful but not required) Step-by-Step Replication Step 1: Set Up Google Colab Navigate to Google Colab Sign in with your Google account Click File → New notebook to create a blank notebook Name your notebook: Symbolic_Institutional_Traps_Replication Step 2: Upload the IPRI Data File On the left sidebar, click the folder icon (Files) Click the upload button (page icon with up arrow) Select and upload: IPRI- 2007 - 2025 data[25].xlsx Wait for the upload to complete (you'll see the file appear in the file list) IMPORTANT: The file path in Colab will be /content/IPRI- 2007 - 2025 data[25].xlsx Note the space after "IPRI-" and before "2007" The script expects this exact path Step 3: Install Required Python Packages Create a new code cell and run: !pip install linearmodels statsmodels scikit-learn seaborn matplotlib pandas numpy scipy openpyxl -q Wait for installation to complete (~30-60 seconds). You should see "Successfully installed" messages. Step 4: Run the Analysis Create a new code cell Copy the entire contents of CPE_Panel_Analysis_Complete.py from this dataset Paste into the code cell Click the play button (▶) or press Shift+Enter to run Wait for analysis to complete (~2-3 minutes) What you'll see: Installation messages "Data loaded successfully!" Summary statistics tables Seven figures appearing sequentially Regression results Cluster analysis results "ANALYSIS COMPLETE!" message Step 5: Download Results All output files are automatically saved in Colab's file system. To download: Option A: Download individually Click the folder icon on the left sidebar Locate each output file: figure1_temporal_dynamics.png through figure7_trajectories_by_cluster.png institutional_changes_2007_2025.csv country_specific_slopes.csv cluster_membership.csv panel_regression_summary.txt Right-click each file → Download Option B: Download all at once Create a new code cell and run: from google.colab import files import os # Create list of output files output_files = [ 'figure1_temporal_dynamics.png', 'figure2_change_scatter.png', 'figure3_country_slopes.png', 'figure4_optimal_clusters.png', 'figure5_cluster_analysis.png', 'figure6_dendrogram.png', 'figure7_trajectories_by_cluster.png', 'institutional_changes_2007_2025.csv', 'country_specific_slopes.csv', 'cluster_membership.csv', 'panel_regression_summary.txt' ] # Download each file for filename in output_files: if os.path.exists(filename): print(f"Downloading {filename}...") files.download(filename) else: print(f"Warning: {filename} not found") Step 6: Verify Replication Compare your outputs to the files in this Zenodo dataset: CSV files should match exactly: Same countries in same order Same numerical values (minor floating-point differences acceptable) Figures should match closely: Same overall patterns and trends Colors and labels identical Minor differences in rendering acceptable Regression results should match: Coefficients within 0.001 Standard errors within 0.001 R-squared values within 0.001 Troubleshooting Error: "FileNotFoundError" Cause: File path incorrect Solution: Verify your uploaded file is named exactly: IPRI- 2007 - 2025 data[25].xlsx with the space after "IPRI-" Alternative: Modify line 54 in the script to match your actual filename Error: "No module named 'linearmodels'" Cause: Packages not installed Solution: Run the pip install command again from Step 3 Error: "MemoryError" or Colab crashes Cause: Insufficient RAM (rare with this dataset) Solution: Click Runtime → Restart runtime, then try again Figures look different Cause: matplotlib version differences or random seed in clustering Solution: Patterns and clusters should still be identical; cosmetic differences are acceptable Different cluster assignments Cause: K-means uses random initialization Solution: The script uses random_state=42 for reproducibility, but if results differ, overall patterns should still match (e.g., Uruguay still in "Escapers", Venezuela still in "Decliners") Expected Runtime Total execution time: 2-3 minutes Installation: ~30 seconds Data loading: <5 seconds Panel regressions: ~30 seconds Clustering: ~20 seconds Visualization: ~60 seconds System Requirements Browser: Chrome, Firefox, or Safari (latest versions) Internet connection: Required throughout (Colab runs on Google servers) Storage: ~10 MB for input data + ~3 MB for outputs RAM: Colab provides sufficient memory (12GB standard tier) Data Sources Primary Data Source International Property Rights Index (IPRI) 2007-2025 Source: Property Rights Alliance URL: https://www.propertyrightsalliance.org/ Coverage: 17 Spanish-speaking Latin American countries Variables: Overall IPRI, Legal and Political Environment (LP), Physical Property Rights (PPR), Intellectual Property Rights (IPR) Countries Included Argentina, Bolivia, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, Venezuela Time Period Annual observations from 2007-2025 (19 years) Methodology Panel Regression Models Model 1: Pooled OLS IPRI_it = β₀ + β₁(time_trend) + ε_it Model 2: Fixed Effects IPRI_it = α_i + β₁(time_trend) + ε_it Where α_i are country-specific intercepts Model 3: Period Effects IPRI_it = α_i + β₁(post_2015) + β₂(post_2020) + ε_it Model 4: Physical Property Rights PPR_it = α_i + β₁(time_trend) + ε_it All models use clustered standard errors at the country level. Trajectory Clustering Features for clustering: Total change (2007-2025) Linear trend slope Volatility (standard deviation of annual changes) Mean level (2007-2025) Method: K-means clustering with k=4 Features standardized (z-scores) Optimal k determined by elbow method and silhouette scores Random state fixed (42) for reproducibility Validation: Hierarchical clustering (Ward linkage) confirms k-means results Statistical Tests Variance Ratio Test (F-test): Tests for divergence H₀: Variance₂₀₂₅ = Variance₂₀₀₇ H₁: Variance₂₀₂₅ > Variance₂₀₀₇ ANOVA: Tests for differences between clusters H₀: All cluster means equal H₁: At least one cluster differs Convergence Test: Correlation between initial level and change Negative correlation → convergence Positive correlation → divergence Software and Dependencies Python Version Python 3.8 or higher Required Packages pandas >= 1.3.0 numpy >= 1.21.0 matplotlib >= 3.4.0 seaborn >= 0.11.0 scikit-learn >= 0.24.0 scipy >= 1.7.0 statsmodels >= 0.13.0 linearmodels >= 4.25 openpyxl >= 3.0.0 All packages are automatically installed by the replication script. Analysis Script File: CPE_Panel_Analysis_Complete.py Lines of code: ~700 Execution time: 2-3 minutes Platform: Google Colab (recommended) or local Jupyter notebook License This dataset is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0). You are free to: Share: Copy and redistribute the material Adapt: Remix, transform, and build upon the material Under the following terms: Attribution: You must give appropriate credit, provide a link to the license, and indicate if changes were made Full license text: https://creativecommons.org/licenses/by/4.0/ Contact Scott BrownProfessor of FinanceUniversity of Puerto Rico, Río PiedrasResearch Affiliate, University of Granada (Erasmus+) Email: [e-mail]: https://orcid.org/0000-0001-6432-9516 For questions about: Data: Contact Scott Brown Replication: Open an issue on [GitHub repository if you create one] Methodology: See paper (under review at Constitutional Political Economy) Acknowledgments This research was conducted as part of a revise-and-resubmit process at Constitutional Political Economy. We thank the editor and anonymous reviewers for constructive feedback that substantially improved the empirical analysis. Data collection was supported by the University of Puerto Rico and the University of Granada (Erasmus+ mobility program). Version History Version 2 (November 2025) Added panel data analysis (2007-2025) Added trajectory clustering (k-means and hierarchical) Added country-specific time trends Added 7 publication-ready figures Added complete replication code Responds to Constitutional Political Economy R&R Version 1 (April 2025) Initial cross-sectional analysis Descriptive statistics and correlations Preliminary clustering Heat map visualization Funding No external funding was received for this research. Conflicts of Interest The authors declare no conflicts of interest. How to Get the IPRI Data The raw IPRI data used in this analysis is publicly available from: Property Rights AllianceWebsite: https://www.propertyrightsalliance.org/Data access: Free download, registration may be required Alternatively, contact the authors for the specific extract used in this analysis. Related Resources Property Rights Alliance: https://www.propertyrightsalliance.org/ World Bank Governance Indicators: https://www.worldbank.org/en/publication/worldwide-governance-indicators V-Dem Democracy Indices: https://www.v-dem.net/ Google Colab: https://colab.research.google.com/ Updates and Corrections If you discover errors or have suggestions for improvement, please contact the corresponding author. We welcome feedback from the research community. Last updated: November 22, 2025 FAIR Data Principles Compliance This dataset adheres to FAIR principles: Findable: Unique DOI, comprehensive metadata, indexed by multiple services Accessible: Free download, persistent URL, open license Interoperable: Standard file formats (CSV, XLSX, PNG, TXT), documented structure Reusable: Clear license (CC BY 4.0), complete documentation, replication code provided Keywords institutional economics, property rights, panel data, Latin America, trajectory clustering, legal origins, path dependence, institutional divergence, fixed effects, time series, Spanish civil law, IPRI, constitutional political economy This dataset supports open science practices by providing complete data, code, and documentation necessary for independent replication and extension of our findings.