研究旨在探討ESG(環境、社會和公司治理)分數對上市公司價值的影響。在傳統投資中大多數投資者仰賴公司財務報表表現來做出決策，然而隨著全球永續意識抬頭，除了財務指標層面，還必須考量到非財務指標對於公司價值的影響。本研究中使用到的財務指標除了有常見的ROA以及ROE，新增了一個較少見的財務指標—Tobin’s Q，進而使用非財務指標—ESG分數(環境構面分數、社會構面分數、公司治理分數)，想要探討ESG分數對於財務績效之間是否具有影響以及並且利用機器學習模型建立預測模型。將ESG分數納入多元迴歸模型後檢測是否跟財務績效指標具有顯著相關性，並利用羅吉斯迴歸模型獲得預測分類準確率，而機器學習模型利用超參數選擇及梯度提升法，將迴歸模型結果進行改善並得到最佳的預測模型。近年來隨著各大企業對於ESG逐漸重視，ESG指標已成為一種新型態評估企業績效的數據及指標，而公司對於永續發展的作為則會影響到ESG分數，進而影響到財務績效指標表現以及一間公司的價值。
    研究結果顯示，使用迴歸模型得到環境分數、社會分數、公司治理分數對於Tobin’s Q都有正向的顯著相關，然而ROA及ROE則不顯著，羅吉斯迴歸得到的預測準確率並不高，因此透過機器學習模型進行準確率改善，最終在三個財務績效指標都獲得相當高的改善成效。透過分析結果，可以得知使用Tobin’s Q指標來衡量公司價值及盈利能力是較好的選擇，而一間公司年度ESG分數確實能夠影響到公司未來發展，高ESG評分的公司在市場價值及盈利能力方面表現更好。綜合以上結果，永續發展對於可持續發展的重要性已是企業必須優先考量到的因素，因為這將大大的影響到ESG分數。

    This study aims to explore the impact of ESG (Environmental, Social, and Governance) scores on the value of listed companies. In traditional investments, most investors rely on a company's financial statement performance to make decisions. However, with the rising awareness of global sustainability, it is necessary to consider the impact of non-financial indicators on a company's value in addition to financial indicators. The financial indicators used in this study include the commonly used Return on Assets (ROA) and Return on Equity (ROE), as well as the less commonly used Tobin's Q ratio. Additionally, non-financial indicators—ESG scores (environmental score, social score, governance score)—are used to explore whether ESG scores have an impact on financial performance. A machine learning model is employed to establish a predictive model. The inclusion of ESG scores in the multiple regression model is used to test for significant correlations with financial performance indicators. Logistic regression is used to obtain prediction accuracy for classification, and the machine learning model improves the regression model results using hyperparameter selection and gradient boosting methods to achieve the best predictive model.In recent years, as major companies have increasingly emphasized ESG, ESG indicators have become a new type of data and metrics for evaluating corporate performance. A company's sustainability practices can affect its ESG scores, thereby influencing financial performance indicators and the company's value. 

    The results of the study show that using the regression model, environmental scores, social scores, and governance scores all have a positive significant correlation with Tobin's Q. However, ROA and ROE are not significant. The prediction accuracy obtained from logistic regression is not high, so the machine learning model is used to improve accuracy. Significant improvements in all three financial performance indicators are achieved through the machine learning model. The analysis results indicate that using Tobin's Q to measure company value and profitability is a better choice, and a company's annual ESG score can indeed affect its future development. Companies with high ESG scores perform better in terms of market value and profitability. In summary, sustainable development is a priority factor that companies must consider, as it significantly impacts ESG scores.

目錄
第一章 緒論	1
第一節 研究背景與動機	1
第二節 研究目的	3
第三節 研究架構	4
第四節 研究範圍與限制	6
(一) 研究對象與限制	6
(二) 研究範圍	6
第二章 文獻探討	7
第一節 ESG評分對上市公司的影響	7
第二節 非財務資訊披露的重要性	9
第三節 機器學習模型	11
第三章 研究方法	14
第一節 類別資料分析—卡方檢定法	14
第二節 機器學習模型	15
(一) 羅吉斯迴歸模型(Logistic Regression Model)	15
(二) 多元線性迴歸模型(Multiple Linear Regression Model)	17
(三) 支持向量機(Support Vector Machine)	18
(四) 極限梯度提升(XGBoost)	19
(五) 決策樹及隨機森林(Decision Tree & Random Forest)	19
第四章 研究結果與討論	21
第一節 樣本與資料來源	21
第二節 資料及變數選取標準	23
第三節 敘述性統計結果	25
第四節 類別資料分析—卡方檢定結果	26
第五節 機器學習模型結果	29
(一) 羅吉斯迴歸模型結果	29
(二) 多元線性迴歸模型結果	32
(三) 支援向量機模型結果	40
(四) 極限梯度提升模型結果	42
(五) 決策樹及隨機森林模型結果	45
第五章 結論與建議	52
參考文獻	54


表目錄
表4.3.1  經營績效之敘述性統計	25
表4.3.2  非財務性ESG指標之敘述性統計	25
表4.4.1  ROA之非財務指標ESG三大構面評分卡方結果	27
表4.4.2  ROE之非財務指標ESG三大構面評分卡方結果	27
表4.4.3  Tobin’s Q之非財務指標ESG三大構面評分卡方結果	28
表4.5.1  二元ROA之混淆矩陣(羅吉斯迴歸)	29
表4.5.2  二元ROE之混淆矩陣(羅吉斯迴歸)	29
表4.5.3  二元Tobin’s Q之混淆矩陣(羅吉斯迴歸)	30
表4.5.4  財務績效指標與非財務績效指標之勝算比	31
表4.5.5  ROA之所有可能迴歸法	34
表4.5.6  ROE之所有可能迴歸法	35
表4.5.7  Tobin’s Q 之所有可能迴歸法	36
表4.5.8  Tobin’s Q之環境構面分數迴歸結果	37
表4.5.9  Tobin’s Q之社會構面分數迴歸結果	37
表4.5.10  Tobin’s Q之公司治理構面分數迴歸結果	37
表4.5.11  ROA之環境構面分數迴歸結果	38
表4.5.12  ROA之社會構面分數迴歸結果	38
表4.5. 13  ROA之公司治理構面分數迴歸結果	38
表4.5.14  ROE之環境構面分數迴歸結果	38
表4.5.15  ROE之社會構面分數迴歸結果	39
表4.5.16  ROE之公司治理構面分數迴歸結果	39
表4.5.17  模型預測誤差值	40
表4.5.18  二元ROA之混淆矩陣(SVM)	41
表4.5.19  二元ROE之混淆矩陣(SVM)	41
表4.5.20  二元Tobin’s Q混淆矩陣(SVM)	41
表4.5.21  ROA之XGBoost模型訓練與測試損失值	42
表4.5.22  二元ROA之混淆矩陣(XGBoost)	42
表4.5.23  ROE之XGBoost模型訓練與測試損失值	43
表4.5.24  二元ROE之混淆矩陣(XGBoost)	43
表4.5. 25  Tobin’s Q之XGBoost模型訓練與測試損失值	44
表4.5.26  二元Tobin’s Q之混淆矩陣(XGBoost)	44
表4.5.27  二元ROA混淆矩陣(決策樹)	46
表4.5.28  隨機森林模型性能評估	46
表4.5.29  隨機森林模型參數調優結果	46
表4.5.30  二元ROE混淆矩陣(決策樹)	48
表4.5.31  隨機森林模型性能評估	48
表4.5.32  隨機森林模型參數調優結果	48
表4.5.33  二元Tobin’s Q混淆矩陣(決策樹)	50
表4.5.34  隨機森林模型性能評估	50
表4.5.35  隨機森林模型參數調優結果	50
表4.5.36  模型預測準確率結果統整	51


圖目錄
圖1.3.1 研究架構圖	5
圖4.5.1  ROA之決策樹結果	45
圖4.5.2  ROE之決策樹結果	47
圖4.5.3  Tobin’s Q之決策樹結果	49

Aydoğmuş, M., Gülay, G. and Ergun, K. (2022). "Impact of ESG performance on firm value and profitability." Borsa Istanbul Review 22: S119-S127.

Bigelli, M., Mengoli, S. and Sandri,S. (2023). "ESG score, board structure and the impact of the non-financial reporting directive on European firms." Journal of Economics and Business 127: 106133.

Cheng, S. and Huang, S. (2024). "ESG combined score effects on stock performance of S&P 500-listed firms." Finance Research Letters 66: 105686.

Gao, Y., Yun, T., Chen, B., Lai, H., Wang, X. and Wang, G. (2024). "Improving the accuracy of canopy height mapping in rubber plantations based on stand age, multi-source satellite images, and random forest algorithm." International Journal of Applied Earth Observation and Geoinformation 131: 103941.

Geng, Z., Duan, X., Li, J., Chu, C., and Han, Y. (2022). "Risk prediction model for food safety based on improved random forest integrating virtual sample." Engineering Applications of Artificial Intelligence 116: 105352.

Göttsche, M., Habermann, F. and Sieber, S. (2024). "The materiality of non-financial tax disclosure: Experimental evidence." Journal of International Accounting, Auditing and Taxation 54: 100600.

Hakkal, S. and Lahcen, A. (2024). "XGBoost To Enhance Learner Performance Prediction." Computers and Education: Artificial Intelligence 7: 100254.

Ke, T., Liao, Y., Wu, M., Ge, X., Huang, X. and Zhang, C. (2023). "Maximal margin hyper-sphere SVM for binary pattern classification." Engineering Applications of Artificial Intelligence 117: 105615.

Matzavela, V. and Alepis, E. (2021). "Decision tree learning through a Predictive Model for Student Academic Performance in Intelligent M-Learning environments." Computers and Education: Artificial Intelligence 2: 100035.

Todorovic, M., Stanisic, N., Zivkovic, M., Bacanin, N., Simic, V. and Tirkolaee, E. B. (2023). "Improving audit opinion prediction accuracy using metaheuristics-tuned XGBoost algorithm with interpretable results through SHAP value analysis." Applied Soft Computing 149: 110955.

Zhang, C. and Wang, Y. (2024). "Are ESG scores affected by the quality of non-financial disclosures?" International Review of Economics & Finance 93: 1431-1458.

Zhou, L., Si, Y.-W. and Fujita, H. (2017). "Predicting the listing statuses of Chinese-listed companies using decision trees combined with an improved filter feature selection method." Knowledge-Based Systems 128: 93-101.

Han, J-J., Kim. H. and Yu, J (2016). "Empirical study on relationship between corporate social responsibility and financial performance in Korea" Asian Journal of Sustainability and Social Responsibility 1:61-76.

Barnett, M. (2007). "Stakeholder Influence Capacity and the Variability of Financial Returns to Corporate Social Responsibility" The Academy of Management Review 32:794-816.

Brammer, S., Pavelin, S. and Brooks, C. (2006). "Corporate Social Performance and Stock Returns: UK Evidence from Disaggregate Measures" Financial Management 35:1-20.

Landi, G., and Sciarelli, M. (2019). "Towards a more ethical market: the impact of ESG rating on corporate financial performance" Social Responsibility Journal 15(1):11-27.

Khader, Q., Oraib, A and Linda, J. (2015) "Research on the impact of gender diversity on the economic performance of companies in Jordan" Economies Volume 10, Issue 4.

Oats, L., and Tuck, P. (2019) "Corporate tax avoidance: is tax transparency the solution?" Accounting and Business Research 49: 565-583.

Song, Y. and Ying, L. (2015) "Decision tree methods: applications for classification and prediction" Shanghai Archives of Psychiatry 27(2):130-135.

Albuquerque, R., Koskinen, Y. and Zhang, C. (2019) "Corporate Social Responsibility and Firm Risk: Theory and Empirical Evidence" Management Science 65(10):4451-4469.