Replication package: Beyond mean effects — a Bayesian structural analysis of patent portfolio adjustments by Chinese listed firms in response to the 2021 reform

Data and code to reproduce the main results, tables and figures of the paper "Beyond mean effects: a Bayesian structural analysis of patent portfolio adjustments by Chinese listed firms in response to the 2021 reform". The package includes:- The firm-year panel of 1,017 Chinese Main Board listed firms (9,731 firm-year observations, 2013–2024)- Provincial subsidy-program characteristics for 27 provinces- Python scripts (PyMC 6.0.1) for the hierarchical Bayesian share-equation model (m0), the difference-in-differences specifications (continuous, discrete, hierarchical), the identification-robustness battery (leave-one-province-out, wild cluster bootstrap, randomization inference, instrumental variables), and the figures- Pre-generated figures (5 PNG files at 95% credible intervals) Requires Python 3.12 and PyMC 6.0.1. Runtime: ~5 minutes for the full pipeline. See README.md and DATA_DICTIONARY.md for details. Cluster identifiers from a companion paper have been omitted from the public data file as they are not required to reproduce any main result.