Institutionalizing Innovation Information and Corporate Bond Credit Spreads: Evidence from China

Two files are provided: the original data file entitled "raw_english.dta" and the Stata do-file etitled "innovation_pilot_DID.do" that reproduces the paper; raw_english.dta is the original, uncleaned firm–year panel of Chinese A-share listed corporate bond issuers over 2007–2024, with English variable names and labels covering the outcome (credit spread), the treatment variables, the firm- and city-level controls, and the moderators/grouping variables, with values exactly as collected (no winsorization, screening, or construction), while innovation_pilot_DID.do reproduces every result in the paper from this file by first performing all data cleaning (firm-level treatment, pilot year, post and Treat×Post indicators, regional controls, 1st/99th-percentile winsorization, and the estimation sample) and then estimating the descriptive statistics, the baseline regressions, the event study and short-run/long-run specifications, the robustness, mechanism, and heterogeneity analyses, and the placebo tests and appendix tables, writing all tables and figures to the out/ folder; to run the do-file, users need only edit the file paths at the top — i.e., set the $datadir (and $work) globals so that the statement use "$datadir/raw_english.dta" points to the location where raw_english.dta is saved, after which the do-file reads the data from that path, performs the cleaning and estimation, and writes its output to $work/out (the required packages are listed in the do-file header).