Replication Package: Structural conditions sustaining Mexico's post-consumer PET recovery — System dynamics model, calibration data, and reproduction scripts

This replication package contains all input data, model files, and analysis scripts needed to reproduce the calibrated system dynamics model and the results reported in the manuscript "Structural conditions sustaining post-consumer PET recovery: Informal-sector ceilings and the limits of policy substitution — evidence from Mexico" (Madrid Rico and Salinas Navarro, 2026). Contents: - 01_Models/: Vensim PLE 10.4.0 baseline (v11) and sensitivity (v12) models, plus PySD-translated Python files - 02_Data/: Primary source datasets (INEGI MEI Tabulados 2003–2024 NAICS sector 56; CONEVAL ITLP-IS Pobreza Laboral 2024; ECOCE Estadísticas 2024 and PET Descargable reports; Coca-Cola FEMSA derived values) - 03_Scripts/: Python reproduction script (reproduce.py) with requirements.txt - 04_Expected_Outputs/: PNG figures of calibrated simulation, longitudinal validation, sensitivity analysis, Monte Carlo, and initial-condition robustness, plus the SD model diagram (SVG) - 05_Documentation/: Manifest with full file listing - README.md: top-level documentation and reproduction instructions The model is calibrated against four independent empirical sources: INEGI MEI (household microdata 2003–2024), CONEVAL ITLP-IS (labor poverty index 2024), ECOCE 2023 annual report (industry-aggregate recovery), and Coca-Cola FEMSA 2024 sustainability report (industrial-buyer prices and capacity). The aggregate recovery rate of 0.63 is reproduced as an emergent model output, not as a fitted target. All software dependencies are open-source: Vensim PLE (free), PySD (BSD-3-Clause), Python ≥ 3.10, NumPy, SciPy, Matplotlib, Pandas.