数字ではなく語り：中国排出量取引制度への参入と年次報告書の炭素含有量

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

🔬研究者:中国ETSの政策イベントを用いた開示行動の因果分析手法と、大規模年報テキストデータ構築の再現可能な枠組みを提供する。

🏢実務担当者:排出量取引制度の対象拡大が自社の年報・統合報告書における炭素開示の質と量に与える影響を予測する材料になる。

🏛政策担当者:炭素市場参入が企業のナラティブ開示と定量開示の乖離を生む可能性を示し、開示義務設計における数値検証の重要性を提起する。

This package reproduces every table and figure in the accompanying manuscript from one command. It contains the analysis code, the unit tests, and the constructed datasets built from 4,777 annual reports filed by 763 Chinese A-share firms for financial years 2019 to 2025. The study asks what happens to two margins of carbon reporting when a firm is brought inside China's national emissions trading system: broad carbon-related disclosure, and the itemized allowance quantities that the Ministry of Finance has required in the notes to the annual report since 2020. The identified event is the 2024 expansion of the market to steel, cement and aluminium smelting. Contents. Analysis and figure scripts under src/; unit tests under tests/; constructed datasets under out/, including page-level carbon disclosure measures under three successively stricter keyword dictionaries, the firm-year panel, the sample definition, extracted operating data, and the narrative text measures; regenerated tables and figures under out/tables/ and out/figures/; the archived policy instruments and the daily carbon market quotations under raw/. Reproduction. Create the environment from requirements.txt, then run "bash run_all.sh". The script runs the tests, rebuilds the sample and the panel, regenerates every table into out/tables/ and every figure into out/figures/, and exits with status 0. It takes a few minutes. Not included. The annual report corpus itself is approximately 18 GB and is not redistributed here. It is public and can be retrieved from CNINFO, the disclosure platform designated by the China Securities Regulatory Commission; the retrieval script is included as src/fetch_reports.py. The extraction products that the regressions depend on are archived in this package, so all reported results reproduce without downloading the corpus. Sources. Annual reports from CNINFO. Institutional parameters from documents published by the Ministry of Ecology and Environment and the Ministry of Finance, archived under raw/policy/. Daily national carbon market quotations from the Shanghai Environment and Energy Exchange. Financial statement items and year-end market data from the statements filed with the annual reports. No commercial database was used at any stage. Note on language. Keyword strings inside the code remain in Chinese because they are the patterns matched against the source documents. Comments, docstrings and documentation are in English. Company and sector names inside the data files are in Chinese because that is how they appear in the source filings.

• openalex https://doi.org/10.17632/c287n97rz5.1first seen 2026-09-14 04:39:49 · last seen 2026-09-14 04:39:55