本研究探討 ESG 表現與企業全要素生產力（Total Factor Productivity， TFP）之關聯，並以 2015 至 2024 年台灣上市非金融業公司為樣本，建立 671 家公司、6，710 筆公司－年度觀察值之平衡追蹤資料。相較於既有文獻多以 ROA、ROE、Tobin’s Q 或股票報酬衡量 ESG 的經濟效果，本文改以 TFP 作為核心被解釋變數，藉此檢驗 ESG 是否真正反映於企業資本與勞動投入之外的生產效率。本文採 Solow residual 成本份額法建構公司年度層級 TFP，並以雙向固定效果模型控制公司固定特質與年度共同衝擊。實證設計包含總 ESG 分數之二次項模型、E/S/G 分項二次項模型，以及以 ESG 為門檻變數之分段模型，藉以檢驗 ESG 與生產力之間是否存在線性、非線性、分項差異與區間斜率差異。實證結果顯示，總 ESG 分數與 TFP 之間雖存在統計關聯，但未呈現穩健正向效果；完整控制變數後，ESG 的平方項不具統計顯著性。ESG 門檻模型顯示，ESG = 57 前後的斜率存在顯著差異，但兩區間皆為負向，代表 ESG 總分效果不宜解讀為「分數愈高愈能提升生產力」。拆解 E、S、G 分項後，社會面向 S 呈現較明確的倒 U 型關係，轉折點約為 61.9 分；環境與治理面向則未呈現穩定統計證據。為回應刪除缺漏資料形成平衡樣本可能造成樣本選擇偏誤之疑慮，本文另以附加價值每員工之自然對數（LP2）作為替代生產力衡量，並採非平衡追蹤資料進行穩健性檢定。結果顯示，在 1，648 家企業、12，979 筆公司－年度觀察值下，ESG 與 LP2 呈顯著正向關聯；惟 ESG 平方項未達顯著，表示替代衡量下未支持顯著非線性效果。整體而言，本文結果指出，ESG 對生產力的影響取決於永續面向、投入水準與企業資源配置方向，而非總分高低本身；同時，ESG 與生產力之正向關聯在非平衡樣本與替代衡量下具有一定穩健性，但非線性效果之解釋宜採保守態度。;This study examines the relationship between ESG performance and firm-level total factor productivity (TFP) using Taiwanese listed non-financial firms from 2015 to 2024. The main balanced panel includes 671 firms and 6，710 firm-year observations. Unlike prior studies that mainly use ROA， ROE， Tobin’s Q， or stock returns， this study adopts TFP， constructed by the Solow residual cost-share approach， to capture firms’ internal production efficiency. Using two-way fixed effects models， the study estimates aggregate ESG， E/S/G dimension， and ESG threshold specifications. The results show that aggregate ESG does not have a robust positive effect on TFP， and its squared term becomes insignificant after full controls. The threshold model identifies ESG = 57 as a sample-specific breakpoint， but both interval slopes are negative. Among the E/S/G dimensions， only the social pillar shows a clearer inverted U-shaped relationship with TFP， with a turning point of approximately 61.9. To address potential sample selection bias from constructing a balanced panel， this study further uses the natural logarithm of value added per employee as an alternative productivity measure in an unbalanced panel of 1，648 firms and 12，979 observations. The LP2 robustness test provides supplementary evidence of a positive ESG-labor productivity association， while the nonlinear effect remains sensitive to the productivity measure. Overall， ESG’s productivity effect depends on ESG dimensions， engagement levels， and resource allocation， while nonlinear findings should be interpreted conservatively.
