本研究主要探討企業社會責任與現金股利之間的關聯性，並利用傾向分數配對方法(Propensity Score Matching method)控制可能會影響落實企業社會責任的因子，再進一步探討企業社會責任本身是否對現金股利仍具有正向影響性。由於企業社會責任由許多構面所組成，我們將繼續探討企業社會責任各個構面的落實是否對股利政策具有顯著的影響效果，並將研究落實企業社會責任是否會為企業帶來保險效果。本研究利用傾向分數配對方法，根據公司的傾向分數是否相近來進行樣本的配對，藉由分析落實企業社會責任公司與其反事實樣本(其特性變數和落實企業社會責任公司極為相似但實際上未落實企業社會責任的公司)來降低選擇性偏誤與內生性問題。本研究以2007年至2016年之上市上櫃公司(不包括金融類股)為樣本，實證結果發現落實企業社會責任的公司現金股利發放較多，且企業社會責任之評分與現金股利亦呈正向影響。此外，各項構面對現金股利發放皆有正面影響，然而落實企業社會責任並無顯著保險效果。同時以傳統解決選擇偏誤問題的Heckman(1979)兩階段估計法作為傾向分數配對法之穩健性檢定，實證結果與前述結果類似。 / This research examines the impact of corporate social responsibility (CSR) on firm’s dividends and the impact between different components on dividends. (i.e. Corpo-rate Governance, Corporate Commitment, Social participation, Environmental Pro-tection). This research suggests that corporate social responsibility have positive im-pact on dividends. We also examine the insurance effect of CSR activities. In this research, we use propensity score matching method to prevent from the contamina-tion of CSR-effect and the selection bias. This research uses all listed company at stock exchange market and over the counter data which financial stocks are exclud-ed and defines the winners of Excellence in Corporate Social Responsibility Award which conferred by Common Wealth Magazine as firms with CSR during 2007 to 2016. The findings support the view of relationship between CSR and dividends is positive. Moreover, each component has positive effect on dividends, while the in-surance effect does not exist. We also use the Heckman two-stage model (1979) as robust test, and find the similar results as using propensity score matching method.