Replication code and variable documentation: digital transformation, debt structure rebalancing, and financial distress risk in Chinese listed firms

This archive contains the estimation code and variable documentation for a study of how digital transformation affects financial distress risk among Chinese non-financial listed firms (2007-2024), in which the liability-side channel - debt structure rebalancing - is examined within a discrete-time logit hazard framework. The code/ folder holds the scripts that construct the firm-year analysis panel and reproduce every result in the manuscript: panel construction (the digital-transformation intensity measure, funding-source shares, the source-concentration HHI, leverage-decomposition components, and the ST event indicator); the descriptive statistics and continuous-distress regressions (Tables 1-2), the baseline and extended discrete-time hazard estimations (Table 3) and the same-sample contribution decomposition (Table 4); the debt-structure margin and component regressions, including the leverage decomposition (Table 5); the heterogeneity splits (Table 6); and the full robustness battery with the winsorisation and risk-set checks in Table 7 - Firth penalized logit, propensity-score matching, entropy balancing, inverse-probability weighting, sub-period and fully lagged re-estimations, competing-risk hazards, scale and window sensitivity, a 500-replication within-year placebo permutation, and the firm-level bootstrap of the attenuation shares - plus the script that draws Figure 1. The docs/ folder holds the annotated estimation code and the variable-definition and sample-construction documentation (manuscript S1 File, S2 Table and S3 Table). The underlying data are third-party licensed (CSMAR/CNRDS) and cannot be redistributed; the scripts contain no embedded records. Running the archive against an identically structured panel file reproduces every table and figure in the manuscript. Environment: Python 3.13 with statsmodels, linearmodels, numpy, pandas, and matplotlib. See README.txt for the complete file list.