ESG 風險管理對臺灣銀行業績效之跨期影響：市場與會計績效分歧之實證研究

Abstract

本研究探討2017 至 2024 年 24 家臺灣銀行環境、社會與公司治理(Environmental, Social, and Governance, ESG)風險管理對銀行財務績效之跨期影響，並深入解析市場類與會計類績效指標間之分歧現象。本研究採用臺灣經濟新報(TEJ)所建構之 TESG 永續發展指標，涵蓋事件雷達(Event Radar Score, ERS)、環境(ENV)、社會(SOC)及公司治理(GOV)四個構面，並以資產報酬率(ROA)、股東權益報酬率(ROE)、息稅前利潤率(EBIT)及 Tobin's Q 四項指標衡量銀行績效，建立實證矩陣，應用縱橫資料迴歸（Panel Data Regression）進行實證研究。將變數採滯後一期處理，以消除反向因果之疑慮，另以銀行層級之叢集穩健標準誤(cluster-robust standard errors)修正，確保統計推論之穩健性。 實證結果揭示臺灣銀行業存在顯著之「市場—會計分歧現象」：ESG 事件雷達(ERS)對 Tobin's Q 於極高顯著水準下呈穩健正向影響，惟對 ROA、ROE 及 EBIT 等會計類績效指標之影響皆不顯著，呼應 Friede et al.(2015)整合 2,200 篇研究所揭示之國際普遍規律。此外，實證結果顯示 ESG 效應呈現顯著之「條件性結構」(conditional structure)：ESG 市場溢價主要集中於民營銀行與金控銀行，而 ESG 對會計績效之改善效益則主要出現於公股銀行；公司治理(GOV)之經濟價值高度集中於金控銀行，於市場價值(Tobin's Q)與營運績效(EBIT)兩個層面皆顯著為正。 上述實證結果印證Pástor et al.(2021)之 ESG 風險溢價理論與 La Porta et al.(2002)之政府銀行所有權理論，並為金管會「綠色及轉型金融行動方案」之政策成效評估提供具有時代意義之實證依據。

This study examines the inter-temporal impact of Environmental, Social, and Governance (ESG) risk management on the financial performance of Taiwan's banking industry, with a particular focus on the divergent effects between market-based and accounting-based performance indicators. Using a balanced panel dataset of 24 Taiwanese banks over the period 2017–2024 (168 bank-year observations after one-period lagging), this research employs the TEJ TESG Sustainability Indicators, which encompass four ESG dimensions—the Event Radar Score (ERS), Environmental (ENV), Social (SOC), and Governance (GOV) pillars—to investigate their effects on four performance measures: Return on Assets (ROA), Return on Equity (ROE), Earnings Before Interest and Taxes margin (EBIT), and Tobin's Q, forming a complete “four-dimension × four-indicator” empirical matrix. To address the “cost-front-loaded, benefit-deferred” nature of ESG investments and to mitigate potential endogeneity concerns, all ESG variables and time-varying controls are lagged by one period. Model selection. Cluster-robust standard errors at the bank level are employed to address heteroskedasticity and serial correlation. The empirical findings reveal a robust “Market–Accounting Divergence” of ESG effects in Taiwan's banking sector: the Event Radar Score (ERS) exhibits a significantly positive impact on Tobin's Q at the 1% level, while its effects on ROA, ROE, and EBIT are statistically insignificant, echoing the international pattern documented in Friede et al.'s (2015) meta-analysis of over 2,200 studies. Furthermore, heterogeneity analyses—uncover a “conditional structure” of ESG effects: ESG market premiums are concentrated in private and FHC-affiliated banks, whereas accounting performance improvements driven by ESG are more salient in state-owned banks. In addition, the economic value of Governance (GOV) is highly concentrated in FHC-affiliated banks, exhibiting significantly positive effects on both market valuation (Tobin's Q) and operational performance (EBIT). These findings provide local empirical evidence for Pástor et al.'s (2021) ESG risk-premium theory and La Porta et al.'s (2002) government ownership theory in Taiwan's banking context, and offer policy implications for the Financial Supervisory Commission's Green and Transition Finance Action Plan.