近年來，永續揭露的項目與篇幅持續擴張，企業被要求揭露的 ESG 資訊愈來愈多。然而揭露得多，是否就代表市場能從中讀到更多有用的訊息？在漂綠疑慮升高、揭露成本上升，且監理趨勢開始強調「應聚焦於具實質意義之資訊」的背景下，這個問題對企業、投資人與監理機關都日益迫切。
本研究以台灣上市櫃公司為對象，資料取自 TEJ 資料庫與臺灣證券交易所 ESG 數位平台，以股價崩跌風險與年報酬率作為觀察市場反應之依據。方法上結合傳統迴歸與三種樹基礎機器學習模型（GBRT、XGBoost、Random Forest），將兩套方法所辨識之重要與顯著特徵交叉比對，並橫跨四個應變數綜合判斷，藉以辨識在不同方法與應變數下反覆顯現關聯的揭露項目。
研究結果給出的答案相當一致，並不是每一種 ESG 揭露都有用。真正與崩跌風險或報酬產生關聯者，集中於少數可驗證、可量化的硬性指標，如範疇排放及其第三方驗證、職業災害比率、薪資水準與董事會結構；大量以「是否揭露」或「揭露字數」衡量的敘事型項目，在兩套方法與各應變數中幾乎都未顯現關聯。所以揭露的數量與形式，未必等同於其所攜帶的資訊內容。財務相關的變數仍然是與崩跌風險、報酬關聯最強的因素，ESG 揭露的關聯相對有限。
本研究以原始揭露資料取代第三方綜合評分，並結合迴歸與機器學習進行跨方法、跨應變數之交叉檢視，據以區分重要、邊際與訊號較微弱之揭露項目，較不重要的項目本身亦是一項值得參考的結果，可供重點揭露與永續揭露制度設計的簡單參考。本研究樣本期間僅涵蓋四個年度，相關結論仍有待未來以更長期間重新檢驗。

Sustainability disclosure has expanded rapidly in both scope and length, yet whether disclosing more conveys more decision-useful information to the market remains unclear. This study examines which ESG disclosure items are associated with stock price crash risk and annual stock returns among Taiwan-listed firms from 2021 to 2024. Using firm-level raw disclosure data from the Taiwan Stock Exchange ESG platform together with financial data from TEJ, it pairs panel regression with three tree-based machine learning models (GBRT, XGBoost, and Random Forest), cross-checking important and significant features across four dependent variables (Crash, NCSKEW, DUVOL, and annual Return). The results are consistent: not every ESG disclosure carries information. Associations concentrate in a small set of verifiable, quantifiable hard indicators, including emission verification, the occupational injury rate, salary levels, and board structure, whereas narrative disclosures measured by mere presence or word count show almost no association under either method. Financial variables remain the strongest correlates of both crash risk and returns. The study concludes that the volume and form of disclosure are not equivalent to their information content, and that the weaker standalone signal of narrative items is itself informative, provided it is not equated with unimportance, for corporate disclosure focus and for regulatory design.

Adams, R. B., & Ferreira, D. (2009). Women in the boardroom and their impact on governance and performance. Journal of Financial Economics, 94(2), 291–309. https://doi.org/10.1016/j.jfineco.2008.10.007
Ben-Nasr, H., & Ghouma, H. (2018). Employee welfare and stock price crash risk. Journal of Corporate Finance, 48, 700–725. https://doi.org/10.1016/j.jcorpfin.2017.12.007
Bose, S., Lim, E. K., Minnick, K., Schorno, P. J., & Shams, S. (2025). Does carbon risk influence stock price crash risk? International evidence. Journal of Business Finance & Accounting. https://doi.org/10.1111/jbfa.12834
Boustanifar, H., & Kang, Y. K. (2022). Employee satisfaction and long-run stock returns, 1984–2020. Financial Analysts Journal, 78(3), 129–151. https://doi.org/10.1080/0015198X.2022.2049127
Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5–32. https://doi.org/10.1023/A:1010933404324
Brown, S., Hillegeist, S. A., & Lo, K. (2004). Conference calls and information asymmetry. Journal of Accounting and Economics, 37(3), 343–366. https://doi.org/10.1016/j.jacceco.2004.02.001
Campbell, J. Y., & Thompson, S. B. (2008). Predicting excess stock returns out of sample: Can anything beat the historical average? Review of Financial Studies, 21(4), 1509–1531. https://doi.org/10.1093/rfs/hhm055
Chawla, N. V., Bowyer, K. W., Hall, L. O., & Kegelmeyer, W. P. (2002). SMOTE: Synthetic minority over-sampling technique. Journal of Artificial Intelligence Research, 16, 321–357. https://doi.org/10.1613/jair.953
Chen, J., Chan, K. C., Dong, W., & Zhang, F. (2017). Internal control and stock price crash risk: Evidence from China. European Accounting Review, 26(1), 125–152. https://doi.org/10.1080/09638180.2015.1117008
Chen, J., Hong, H., & Stein, J. C. (2001). Forecasting crashes: Trading volume, past returns, and conditional skewness in stock prices. Journal of Financial Economics, 61(3), 345–381. https://doi.org/10.1016/S0304-405X(01)00066-6
Cho, M., Kim, H. D., & Kim, Y. (2023). Audit committee accounting financial expertise and stock price crash risk. International Review of Financial Analysis, 90, 102944. https://doi.org/10.1016/j.irfa.2023.102944
de Franco, C., Geissler, C., Margot, V., & Monnier, B. (2019). ESG investments: Filtering versus machine learning approaches. SSRN Working Paper. https://doi.org/10.2139/ssrn.3415911
Deng, S., Zhu, Y., Duan, S., Fu, Z., & Liu, Z. (2022). Stock price crash warning in the Chinese security market using a machine learning-based method and financial indicators. Systems, 10(4), 108. https://doi.org/10.3390/systems10040108
Dimson, E. (1979). Risk measurement when shares are subject to infrequent trading. Journal of Financial Economics, 7(2), 197–226. https://doi.org/10.1016/0304-405X(79)90013-8
Dumitrescu, A., & Zakriya, M. (2021). Stakeholders and the stock price crash risk: What matters in corporate social performance? Journal of Corporate Finance, 67, 101871. https://doi.org/10.1016/j.jcorpfin.2020.101871
Edmans, A. (2011). Does the stock market fully value intangibles? Employee satisfaction and equity prices. Journal of Financial Economics, 101(3), 621–640. https://doi.org/10.1016/j.jfineco.2011.03.021
Enders, C. K. (2010). Applied missing data analysis. Guilford Press.
European Commission. (2025a). Corporate sustainability reporting. Directorate-General for Financial Stability, Financial Services and Capital Markets Union. https://finance.ec.europa.eu/financial-markets/company-reporting-and-auditing/company-reporting/corporate-sustainability-reporting_en
European Commission. (2025b, July 30). Commission presents voluntary sustainability reporting standard to ease burden on SMEs. https://finance.ec.europa.eu/publications/commission-presents-voluntary-sustainability-reporting-standard-ease-burden-smes_en
European Financial Reporting Advisory Group. (2024, December 17). EFRAG releases the voluntary sustainability reporting standard for non-listed SMEs. https://www.efrag.org/en/news-and-calendar/news/efrag-releases-the-voluntary-sustainability-reporting-standard-for-nonlisted-smes
Fawcett, T. (2006). An introduction to ROC analysis. Pattern Recognition Letters, 27(8), 861–874. https://doi.org/10.1016/j.patrec.2005.10.010
Friede, G., Busch, T., & Bassen, A. (2015). ESG and financial performance: Aggregated evidence from more than 2000 empirical studies. Journal of Sustainable Finance & Investment, 5(4), 210–233. https://doi.org/10.1080/20430795.2015.1118917
Fu, X., Wu, X., & Zhang, Z. (2021). The information role of earnings conference call tone. Journal of Business Ethics, 173(3), 557–575. https://doi.org/10.1007/s10551-019-04326-1
Garcia-Laencina, P. J., Sancho-Gomez, J. L., & Figueiras-Vidal, A. R. (2010). Pattern classification with missing data: A review. Neural Computing and Applications, 19(2), 263–282. https://doi.org/10.1007/s00521-009-0295-6
Gibson Brandon, R., Krueger, P., & Schmidt, P. S. (2021). ESG rating disagreement and stock returns. Financial Analysts Journal, 77(4), 104–127. https://doi.org/10.1080/0015198X.2021.1963186
Hu, J., Li, S., Taboada, A. G., & Zhang, F. (2020). Corporate board reforms around the world and stock price crash risk. Journal of Corporate Finance, 62, 101557. https://doi.org/10.1016/j.jcorpfin.2020.101557
Hutton, A. P., Marcus, A. J., & Tehranian, H. (2009). Opaque financial reports, R2, and crash risk. Journal of Financial Economics, 94(1), 67–86. https://doi.org/10.1016/j.jfineco.2008.09.008
Jiang, F., Ma, T., & Zhu, F. (2024). Fundamental characteristics, machine learning, and stock price crash risk. Journal of Financial Markets, 69, 100908. https://doi.org/10.1016/j.finmar.2024.100908
Jin, L., & Myers, S. C. (2006). R2 around the world: New theory and new tests. Journal of Financial Economics, 79(2), 257–292. https://doi.org/10.1016/j.jfineco.2004.11.003
Ju, C., Fang, X., & Shen, Z. (2025). ESG rating divergence and stock price crash risk. North American Journal of Economics and Finance, 76, 102323. https://doi.org/10.1016/j.najef.2024.102323
Karasan, A., Alp, O. S., & Weber, G. W. (2025). Machine learning approach to stock price crash risk. Annals of Operations Research, 350, 1053–1074. https://doi.org/10.1007/s10479-024-06042-4
Kim, Y., Li, H., & Li, S. (2014). Corporate social responsibility and stock price crash risk. Journal of Banking & Finance, 43, 1–13. https://doi.org/10.1016/j.jbankfin.2013.12.032
Kosmidou, K., Kousenidis, D., Ladas, A., & Negkakis, C. (2024). Climate-related performance and stock price crash risk. Financial Markets, Institutions & Instruments, 33(2), 113–148. https://doi.org/10.1111/fmii.12194
Krueger, P., Sautner, Z., & Starks, L. T. (2020). The importance of climate risks for institutional investors. Review of Financial Studies, 33(3), 1067–1111. https://doi.org/10.1093/rfs/hhz137
Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4765–4774. https://doi.org/10.5555/3295222.3295230
Pedersen, L. H., Fitzgibbons, S., & Pomorski, L. (2021). Responsible investing: The ESG-efficient frontier. Journal of Financial Economics, 142(2), 572–597. https://doi.org/10.1016/j.jfineco.2020.11.001
PwC. (2023, November 15). PwC's 2023 Global Investor Survey: Investors are calling for increased transparency. PricewaterhouseCoopers. https://www.pwc.com/gx/en/newsroom/press-releases/2023/pwc-2023-global-investor-survey.html
Rubin, D. B. (1976). Inference and missing data. Biometrika, 63(3), 581–592. https://doi.org/10.1093/biomet/63.3.581
Saito, T., & Rehmsmeier, M. (2015). The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PLoS ONE, 10(3), e0118432. https://doi.org/10.1371/journal.pone.0118432
Scikit-Learn Documentation. (2023). Cross-validation: Evaluating estimator performance. https://scikit-learn.org/stable/modules/cross_validation.html
Sperrin, M., Carpenter, J. R., Sisk, R., & Buchan, I. (2020). Explicitly modelling missing data improves risk prediction in clinical risk scores: An illustrative example with the Manchester Triage System. BMC Medical Informatics and Decision Making, 20(1), 1–9. https://doi.org/10.1186/s12911-020-01244-9
Stekhoven, D. J., & Bühlmann, P. (2012). MissForest—Non-parametric missing value imputation for mixed-type data. Bioinformatics, 28(1), 112–118. https://doi.org/10.1093/bioinformatics/btr597
U.S. Securities and Exchange Commission. (2023, September 25). Deutsche Bank subsidiary DWS to pay $25 million for anti-money laundering violations and misstatements regarding ESG investments (Press Release No. 2023-194). https://www.sec.gov/newsroom/press-releases/2023-194
van Buuren, S., & Groothuis-Oudshoorn, K. (2011). mice: Multivariate imputation by chained equations in R. Journal of Statistical Software, 45(3), 1–67. https://doi.org/10.18637/jss.v045.i03
Welch, I., & Goyal, A. (2008). A comprehensive look at the empirical performance of equity premium prediction. Review of Financial Studies, 21(4), 1455–1508. https://doi.org/10.1093/rfs/hhm014
Yang, M., Chen, S., & Maresova, P. (2024). Environmental corporate social responsibility and stock price crash risk: The role of environmental performance and ISO 14001. International Review of Economics and Finance, 96, 103627. https://doi.org/10.1016/j.iref.2024.103627
Zhang, J., Cui, C., Zheng, C., & Taylor, G. (2025). Artificial intelligence innovation and stock price crash risk. Journal of Financial Research, 48, 503–543. https://doi.org/10.1111/jfir.12412
金管會. (2023). 上市櫃公司永續發展行動方案（2023年）。金融監督管理委員會。https://www.fsc.gov.tw/ch/home.jsp?id=96&parentpath=0%2C2&mcustomize=news_view.jsp&dataserno=202303280001&dtable=News