Social Capital, Trust, and Firm Performance: The Value of Corporate Social Responsibility during the Financial Crisis
社會資本、信任與公司績效：金融危機期間企業社會責任的價值

“The present financial crisis springs from a catastrophic collapse in confidence . . . Financial markets hinge on trust, and that trust has eroded.” —Joseph Stiglitz (2008)
"目前的金融危機源於信心的災難性崩潰......。.金融市場取決於信任，而這種信任已經受到侵蝕"。約瑟夫-斯蒂格利茨 (2008)
“The fundamental problem isn’t lack of capital. It’s lack of trust. And without trust, Wall Street might as well fold up its fancy tents.”
"最根本的問題不是缺乏資金，而是缺乏信任。而是缺乏信任。如果沒有信任，華爾街也許還可以收起它華麗的帳篷"。
-Former U.S. Labor Secretary Robert Reich (2008)
-前美國勞動部部長 Robert Reich (2008)
*Karl V. Lins is at the University of Utah. Henri Servaes is at London Business School, CEPR, and ECGI. Ane Tamayo is at the London School of Economics and Political Science (LSE). The authors have no conflicts of interest to disclose. We would like to thank Taylor Begley, Colin Clubb, Joao Cocco, Mike Cooper, James Dow, Alex Edmans, Christopher Hennessy, Ioannis Ioannou, Ralph Koijen, Jean-Marie Meier, Yuval Millo, Michael Roberts (the Editor), Kelly Shue, Rui Silva, Hannes Wagner, Yao Zeng, an anonymous Associate Editor, an anonymous referee, and seminar participants at City University, Erasmus University, ESSEC, HEC Paris, INSEAD, King’s College, London Business School, London School of Economics, Tilburg University, University of Bristol, University of Edinburgh, University of Leicester, University of Melbourne, University of New South Wales, University of Southampton, University of Sydney, WHU Otto Beisheim, the French Finance Association, London Business School Summer Finance Symposium, the International Accounting Research Symposium at the Fundación Ramón Areces, the International Corporate Governance Conference at Hong Kong Baptist University, and the University of Cambridge Financial Accounting Symposium for helpful comments and discussions. We would also like to thank the ECGI for the 2016 Standard Life Investments Finance Working Paper Prize. Dimas Fazio provided excellent research assistance.
*Karl V. Lins 在猶他大學任教。Henri Servaes 在倫敦商學院、CEPR 和 ECGI 工作。Ane Tamayo 就讀於倫敦政治經濟學院 (LSE)。作者無利益衝突需揭露。我們感謝 Taylor Begley、Colin Clubb、Joao Cocco、Mike Cooper、James Dow、Alex Edmans、Christopher Hennessy、Ioannis Ioannou、Ralph Koijen、Jean-Marie Meier、Yuval Millo、Michael Roberts（編輯）、Kelly Shue、Rui Silva、Hannes Wagner、Yao Zeng、一位匿名副編輯、一位匿名審閱人，以及城市大學、伊拉斯谟大學、ESSEC、巴黎高等商學院、英國工商管理學院、英國國王學院、倫敦商學院、倫敦經濟學院、蒂爾堡大學、布里斯托爾大學的研討會參加者、愛丁堡大學、萊斯特大學、墨爾本大學、新南威爾斯大學、南安普頓大學、雪梨大學、WHU Otto Beisheim、法國金融協會、倫敦商學院夏季金融研討會、Fundación Ramón Areces 的國際會計研究研討會、香港浸會大學的國際公司治理會議，以及劍橋大學的財務會計研討會。我們還要感謝 ECGI 頒發 2016 年標準人壽投資金融工作論文獎。Dimas Fazio 提供了出色的研究協助。
“The global financial and economic crisis has done a lot of harm to the public trust in the institutions, the principles and the concept itself of the market economy.”
「全球金融和經濟危機對公眾對市場經濟的機構、原則和概念本身的信任造成了很大的傷害」。
-OECD Secretary General Angel Gurria (2009)
-經合組織秘書長 Angel Gurria (2009)
“Something important was destroyed in the last few months of 2008. It is an asset crucial to production, even if it is not made of bricks and mortar… This asset is trust.”
"在 2008 年最後的幾個月，一些重要的東西被摧毀了。它是對生產至關重要的資產，即使它不是由磚塊和灰泥製成......這種資產就是信任"。
-Paolo Sapienza and Luigi Zingales (2012, p. 123)
-Paolo Sapienza 和 Luigi Zingales (2012, p. 123)
The financial crisis highlighted the importance of trust for well-functioning markets and financial stability, but discussions on the role of trust and, more generally, social capital in economic life are not new. Already in 1972, Arrow argued that “virtually every commercial transaction has within itself an element of trust” ( p . 357), and suggested that much of the economic backwardness in the world might be due to the lack of mutual confidence. In line with this view, Putnam (1993) shows that higher social capital societies, in which trust is greater, display higher economic development (see also Fukuyama (1995), La Porta et al. (1997), and Knack and Keefer (1997)). Focusing on capital markets, Guiso, Sapienza, and Zingales $(2004,2008)$$(2004,2008)$(2004,2008)(2004,2008) document that trust derived from greater social capital allows for more stock market participation. These studies and other related work demonstrate the importance of social capital and trust from a macroeconomic perspective. However, the extent to which social capital and trust impact firm performance is relatively unexplored in the literature. The objective of this paper is to address this question.
金融危機突顯了信任對於市場良好運作和金融穩定的重要性，但關於信任以及更廣泛的社會 資本在經濟生活中的作用的討論並非新鮮事物。早在 1972 年，Arrow 就認為「幾乎每一項商業交易本身都包含信任的元素」(第 357 頁)，並指出世界上許多經濟落後的現象可能是由於缺乏互信所致。與此觀點一致，Putnam (1993) 指出，社會資本較高的社會，信任度較高，經濟發展也較高（參見 Fukuyama (1995)、La Porta et al. (1997) 以及 Knack and Keefer (1997)）。Guiso, Sapienza, and Zingales $(2004,2008)$$(2004,2008)$(2004,2008)(2004,2008) 以資本市場為重點，記錄了由更大的社會資本所產生的信任可以讓更多人參與股票市場。這些研究和其他相關工作從宏觀經濟角度證明了社會資本和信任的重要性。然而，社會資本和信任對公司績效的影響程度在文獻中還相對較少。本文的目的就是要解決這個問題。

Empirical identification of the effect of trust and, more generally, social capital on firm performance is challenging. First, social capital is a broadly defined concept, often encompassing trust and cooperative norms (e.g., Scrivens and Smith (2013)), and hence its measurement is not straightforward. Second, without exogenous variation in firm-level social capital, it is difficult to attribute changes in performance to changes in social capital.
以經驗識別信任以及更廣泛而言的社會資本對公司績效的影響具有挑戰性。首先，社會資本是一個定義廣泛的概念，通常包含信任和合作規範 (例如 Scrivens 和 Smith (2013))，因此其衡量並不直接。其次，如果公司層級的社會資本沒有外生變化，就很難將績效變化歸因於社會資本的變化。

To address the first challenge, we focus on a firm’s Corporate Social Responsibility (CSR) activities as a measure of its social capital, following recent work in economics (Sacconi and Degli Antoni (2011)) suggesting that a firm’s CSR activities are a good proxy for its social capital, and also the widespread view among practitioners and corporations that a firm’s CSR activities generate social capital and trust. ${}^1$${}^1$^(1){ }^{1}
為了應對第一個挑戰，我們專注於以公司的企業社會責任 (CSR) 活動來衡量公司的社會資本，因為最近經濟學領域的研究 (Sacconi and Degli Antoni (2011)) 指出，公司的企業社會責任活動是公司社會資本的良好替代物，而且實踐者和企業也普遍認為公司的企業社會責任活動會產生社會資本和信任。 ${}^1$${}^1$^(1){ }^{1}

To address the second challenge, we employ the 2008-2009 financial crisis, a period during which public trust in corporations, capital markets, and institutions declined unexpectedly. ${}^2$${}^2$^(2){ }^{2} If a firm’s social capital helps build stakeholder trust and cooperation (Putnam (1993)), it should pay off when being
為了應對第二個挑戰，我們採用了 2008-2009 年的金融危機，在這段期間，公眾對企業、資本市場和機構的信任出乎意料地下降。 ${}^2$${}^2$^(2){ }^{2} 如果公司的社會資本有助於建立利害關係人的信任與合作（Putnam (1993)），那麼它應該會在被利害關係人信任時得到回報。

trustworthy is more valuable, such as in an unexpectedly low-trust period. From a shareholder perspective, if high social capital firms are perceived as more trustworthy, investors may place a valuation premium on these firms when overall trust in companies is low (see Guiso, Sapienza, and Zingales (2008)), as in the 2008-2009 financial crisis. From a stakeholder perspective, the reciprocity concept often discussed in studies of social capital (i.e., the idea that “I will be good to you because I believe you will be good to me at some point in the future”) suggests that stakeholders (e.g., employees, customers, suppliers, and the community at large) are more likely to help high social capital firms weather a negative shock, given that such firms displayed greater attention to, and cooperation with, stakeholders in the past.
例如，在出乎意料的低信任度時期，值得信任的公司更有價值。從股東的角度來看，如果高社會資本的公司被認為更值得信賴，那麼當公司的整體信任度較低時，投資者可能會對這些公司給予估值溢價（見 Guiso, Sapienza, and Zingales (2008)），就像 2008-2009 年的金融危機一樣。從利益相關者的角度來看，社會資本研究中經常討論的互惠概念（即 「我會對你好，因為我相信你在未來某個時候會對我好」）表明，利益相關者（如員工、客戶、供應商和社區）更有可能幫助高社會資本公司抵禦負面衝擊，因為這些公司在過去表現出對利益相關者更多的關注和合作。

To test whether firm-level social capital pays off during a crisis of trust, we examine the performance of 1,673 nonfinancial firms with CSR data available on the MSCI ESG Stats database (formerly known as KLD) over the August 2008 to March 2009 financial crisis period. In regressions that control for a wide variety of factors and firm characteristics (including governance and transparency), we find that firms that entered the crisis period with high CSR ratings have significantly higher (between four and seven percentage points) crisis-period stock returns than those that entered it with low CSR ratings. The economic importance of social capital in explaining stock returns is at least half as large as the effect of cash holdings and leverage, financial variables previously shown to affect crisis-period returns (Duchin, Ozbas, and Sensoy (2010) and Almeida et al. (2012)). This result highlights the importance of expanding the focus beyond financial capital when attempting to understand the determinants of firm-level performance during a crisis of trust.
為了測試公司層級的社會資本是否會在信任危機期間獲得報酬，我們檢視了 1,673 家在 MSCI ESG Stats 資料庫（前稱 KLD）中有企業社會責任資料的非金融公司在 2008 年 8 月至 2009 年 3 月金融危機期間的表現。在控制多種因素和公司特徵（包括治理和透明度）的回歸中，我們發現以高 CSR 評級進入危機期的公司，其危機期股票回報顯著高於以低 CSR 評級進入危機期的公司（4 到 7 個百分點之間）。社會資本在解釋股票報酬率方面的經濟重要性，至少是現金持有量和槓桿效應的一半，而現金持有量和槓桿效應是先前已被證實會影響危機期間報酬率的金融變數 (Duchin, Ozbas, and Sensoy (2010) and Almeida et al. (2012))。這個結果突顯出，在嘗試了解信任危機期間公司層級績效的決定因素時，將焦點擴大到金融資本以外的重要性。

To alleviate concerns that the stock market outperformance we observe is due to some factor other than a shock to trust, we conduct three further tests. First, we investigate the association between CSR and stock returns during the Enron/Worldcom crisis of the early 2000s, a period during which widespread revelation of fraud undermined investor confidence in the U.S. stock market. We find that high-CSR firms also earned excess returns relative to low-CSR firms during this period. Second, we investigate whether our results are driven by the decline in the supply of credit that firms faced during the financial crisis, rather than by a decline in market-wide trust. Specifically, we test whether CSR is related to stock returns in the period July 2007 through July 2008, when there was a shock to the credit supply but no shock to the importance of trust. We find no significant relation between CSR and stock returns during this earlier period of the crisis. Third, we examine whether the relation between CSR and crisis-period returns is stronger in high-trust regions, as identified in the 2006 General Social Survey. We find that this is indeed the case.
為了減輕對於我們觀察到的股市超額表現是由信任衝擊以外的其他因素造成的疑慮，我們進行了三個進一步的測試。首先，我們調查了 2000 年代初安隆/世通危機期間企業社會責任與股票回報之間的關係，在這段期間，廣泛揭露的欺詐行為削弱了投資者對美國股市的信心。我們發現，在這段期間，高 CSR 公司相對於低 CSR 公司也獲得了超額回報。其次，我們研究了我們的結果是否是由於公司在金融危機期間面臨的信貸供應下降，而非市場整體信任度下降所導致。具體來說，我們測試了在 2007 年 7 月到 2008 年 7 月這段期間，企業社會責任是否與股票回報有關，在這段期間，信貸供應受到衝擊，但信任的重要性並沒有受到衝擊。我們發現在這段較早的危機期間，企業社會責任與股票回報並無顯著關係。第三，我們檢視 2006 年一般社會調查所指出的高信任地區，企業社會責任與危機期間回報的關係是否較強。我們發現情況確實如此。

It is possible, of course, that high-CSR firms also outperform low-CSR firms during noncrisis periods (e.g., Edmans (2011)). To assess this possibility, we examine whether the superior performance of high-CSR firms extends to
當然，在非危機時期，高CSR公司的表現也可能優於低CSR公司（例如，Edmans (2011)）。為了評估這種可能性，我們檢視高CSR公司的優異表現是否延伸至

periods of economic growth or economic recovery using firm fixed effects models that test the relation between CSR and firm performance before, during, and after the crisis. These models show that CSR has a positive impact on returns only during the crisis period, and that this effect is not due to time-invariant unobservable firm characteristics.
在經濟成長或經濟復甦期間，使用公司固定效應模型來測試企業社會責任與公司在危機前、危機期間和危機後的表現之間的關係。這些模型顯示，企業社會責任只有在危機期間才會對報酬率產生正面影響，而且這種影響並非由時間不變的不可觀察公司特徵所造成。

We next seek to identify the mechanisms behind the outperformance of highCSR firms by examining firms’ profitability and productivity as well as their capital raising during the crisis. We find that high-CSR firms have higher profitability and gross margins, and experience higher sales growth, than other firms during the crisis. They also have higher sales per employee and are able to raise more debt. These results are consistent with stakeholder and investor commitment to help firms deemed to be more trustworthy during the crisis.
接下來，我們將透過檢視企業的獲利能力、生產力以及在危機期間的資本募集，來找出高CSR企業表現優異背後的機制。我們發現，在危機期間，與其他公司相比，高CSR 公司具有更高的盈利能力和毛利率，並經歷了更高的銷售增長。它們的員工人均銷售額也較高，並且能夠籌集更多的債務。這些結果與利害關係人和投資人在危機期間幫助被視為更值得信賴的公司的承諾一致。

Collectively, the findings that investors assign a premium to high-CSR firms during a crisis of trust and that real effects take place at the firm level during this time indicate that greater social capital maps into higher returns at the microeconomic level. From a firm’s perspective, our results indicate that the benefits that accrue to firms from building social capital through CSR activities outweigh the costs of these activities when trust declines unexpectedly. As such, investment in social capital can be thought of as an insurance policy that pays off when investors and the economy at large face a severe crisis of confidence and when the reward for being identifiably trustworthy increases markedly. Our results thus highlight an enhanced insurance benefit of CSR that goes beyond the notion that CSR acts as insurance against idiosyncratic firm-specific legal risk (see, e.g., Godfrey, Merrill, and Hansen (2009), Minor (2015), and Hong and Liskovich (2016)).
總括而言，投資者在信任危機期間給予高 CSR 公司溢價，以及在此期間公司層面產生實際效 應的研究結果顯示，更高的社會資本在微觀經濟層面上映射出更高的回報。從公司的角度來看，我們的結果顯示，當信任度意外下降時，公司透過企業社會責任活動建立社會資本所獲得的利益，會超過這些活動的成本。因此，社會資本投資可被視為一種保險政策，當投資者和整體經濟面臨嚴重信賴危機，以及可識別可信賴的報酬顯著增加時，這種保險政策就會得到報酬。因此，我們的結果突顯了企業社會責任的增強保險效益，這超越了企業社會責任可作為特定公司法律風險保險的概念（例如，請參閱 Godfrey、Merrill 和 Hansen (2009)、Minor (2015) 以及 Hong 和 Liskovich (2016)）。

While our focus is on the impact of social capital on firm performance during a shock to trust, our research design allows us to sidestep typical endogeneity concerns that make it difficult to identify whether CSR activities impact firm value, despite much research on this issue. ${}^3$${}^3$^(3){ }^{3} In our natural experiment, the exogenous financial shock disrupts the equilibrium, while levels of CSR remain fixed, at least in the short term. This allows us to directly observe how investors adjust their valuations of firms with differing attitudes toward CSR. Thus, this paper also makes a contribution to the literature investigating whether CSR is value-enhancing for shareholders. We recognize, however, that we do not have exogenous variation in the levels of CSR, which limits the inferences we can draw about the impact of CSR on performance during normal times.
雖然我們的重點是在信任衝擊期間社會資本對公司績效的影響，但我們的研究設計允許我們避開典型的內生性問題，這些問題使我們很難確定企業社會責任活動是否影響公司價值，儘管對這個問題有很多研究。 ${}^3$${}^3$^(3){ }^{3} 在我們的自然實驗中，外生的金融衝擊破壞了均衡，而企業社會責任的水準至少在短期內是固定的。這使我們能夠直接觀察投資者如何調整他們對企業社會責任持不同態度的公司的估值。因此，本文也對研究企業社會責任是否能提升股東價值的文獻有所貢獻。然而，我們承認，我們沒有企業社會責任水平的外生變化，這限制了我們對企業社會責任在正常時期對業績影響的推斷。

The remainder of the paper is structured as follows. Section I discusses in more detail the theoretical motivation behind our proxies and tests. Section II discusses our data and summary statistics. In Section III, we analyze whether CSR ratings impact stock returns during the crisis and conduct robustness
本文其餘部分的結構如下。第一節詳細討論我們的代用指標和測試背後的理論動機。第二節討論我們的數據和摘要統計。在第三節中，我們分析企業社會責任評級是否影響危機期間的股票回報，並進行穩健性分析。

tests. In Section IV, we investigate several mechanisms that may explain the excess performance of high-CSR firms. Section V concludes the paper.
測試。在第四節中，我們研究了幾種可能解釋高CSR公司超額績效的機制。第五節為本文的結論。

Over the last 20 years, the terms “social capital” and “trust” have become increasingly popular in the economics and finance literature (Putnam (1993, 2000), Knack and Keefer (1997), La Porta et al. (1997), Guiso, Sapienza, and Zingales (2004, 2008)). ${}^4$${}^4$^(4){ }^{4} Often used indistinctly, both concepts are somewhat abstract, although social capital is arguably the harder one to define due to its multidimensional nature.
在過去 20 年中，「社會資本」和「信任」這兩個詞在經濟學和金融學文獻（Putnam (1993, 2000)、Knack 和 Keefer (1997)、La Porta 等人 (1997)、Guiso、Sapienza 和 Zingales (2004, 2008)）中越來越流行。 ${}^4$${}^4$^(4){ }^{4} 這兩個概念經常被模糊使用，都有點抽象，不過社會資本因其多維性質而更難定義。

Trust is often understood as “the expectation that another person (or institution) will perform actions that are beneficial, or at least not detrimental, to us regardless of our capacity to monitor those actions… so that we will consider cooperating with him [the institution]” (Sapienza and Zingales (2012, p. 124), based on Gambetta (1988)). This definition highlights the probabilistic nature of trust (e.g., Gambetta (1988)), the concept of cooperation (e.g., Fukuyama (1995) and La Porta et al. (1997)), and the inability to monitor others’ actions ex ante (e.g., Dasgupta (1988)). ${}^5$${}^5$^(5){ }^{5}
信任通常被理解為「對另一個人（或機構）將採取對我們有利或至少無損的行動的期望，無論我們是否有能力監控這些行動......這樣我們就會考慮與他（機構）合作」（Sapienza 和 Zingales (2012, p. 124)，根據 Gambetta (1988)）。這個定義強調了信任的概率性（例如 Gambetta (1988)）、合作的概念（例如 Fukuyama (1995) 和 La Porta 等人 (1997)），以及無法事前監控他人行為（例如 Dasgupta (1988)）。 ${}^5$${}^5$^(5){ }^{5}

Social capital is a broader concept. For example, Putnam $(1993,2000)$$(1993,2000)$(1993,2000)(1993,2000) views social capital as “a propensity of people in a society to cooperate to produce socially efficient outcomes” (La Porta et al. (1997, p. 333)) and highlights “the norms of reciprocity and trustworthiness” that arise from connections among individuals. A recent OECD paper (Scrivens and Smith (2013)) decomposes social capital into four dimensions, with the intent of facilitating the development of empirical measures: (i) personal relationships, (ii) social network support, (iii) civic engagement, and (iv) trust and cooperative norms. ${}^6$${}^6$^(6){ }^{6} The notion of social capital that we explore, like much of the work in economics and finance (e.g., Putnam (1993, 2000), Fukuyama (1995), Knack and Keefer (1997), La Porta et al. (1997), Guiso, Sapienza, and Zingales (2004, 2008)), is mostly related to the last two interpretations of the OECD.
社會資本是一個更廣泛的概念。例如，Putnam $(1993,2000)$$(1993,2000)$(1993,2000)(1993,2000) 將社會資本視為 「一個社會中的人們為產生社會有效成果而進行合作的傾向」（La Porta et al. (1997, p.333)），並強調了個人之間的聯繫所產生的 「互惠和可信賴的規範」。經合組織（OECD）最近的一篇論文（Scrivens and Smith (2013)）將社會資本分解為四個層面，目的在於促進實證量度的發展：(i) 個人關係，(ii) 社會網路支援，(iii) 公民參與，以及 (iv) 信任與合作規範。 ${}^6$${}^6$^(6){ }^{6} 我們所探討的社會資本概念，就像經濟學與金融學的許多工作（例如，Putnam (1993、2000)、Fukuyama (1995)、Knack and Keefer (1997)、La Porta et al. (1997)、Guiso、Sapienza and Zingales (2004、2008)）一樣，多半與經合組織的後兩項詮釋有關。

The civic engagement aspect of social capital refers to the activities through which agents contribute positively to the community and social life (e.g., volunteering, political participation, donations; Guiso, Sapienza, and Zingales (2011), Scrivens and Smith (2013)). Civic engagement can engender positive
社會資本的公民參與方面指的是代理人對社區和社會生活做出積極貢獻的活動（如志願服務、政治參與、捐贈；Guiso, Sapienza, and Zingales (2011)，Scrivens and Smith (2013)）。公民參與可以產生正面的

outcomes by, for example, fostering trust and norms of cooperation, such as reciprocity. ${}^7$${}^7$^(7){ }^{7}
例如，通過培養信任和合作規範（如互惠）來達到結果。 ${}^7$${}^7$^(7){ }^{7}

Trust and cooperative norms comprise factors (social norms, including reciprocity, and shared values) that shape the way that agents behave towards each other and as members of society. Under this definition, social capital is viewed as an enabler of collective action and cooperation, and thereby leading to positive outcomes (e.g., economic growth, government performance, and environmental stewardship). The channels through which positive outcomes are derived include: (i) reductions in transaction costs (by reducing the need for formal contracts in the presence of information asymmetry (Knack and Keefer (1997)) and (ii) potentially more efficient allocation of resources.
信任與合作規範包括塑造代理人彼此之間以及作為社會成員的行為方式的因素（社會規範，包括互惠和共同價值觀）。根據此定義，社會資本被視為集體行動和合作的推動力，從而產生積極的結果（如經濟增長、政府表現和環境管理）。產生正面結果的渠道包括(i) 降低交易成本（在資訊不對稱的情況下，減少對正式合約的需求 (Knack and Keefer (1997)）；(ii) 可能更有效率地分配資源。

All of the above concepts are, of course, interconnected. For example, civic engagement can generate trust and cooperation, which in turn can foster further civic engagement; likewise, cooperation can build trust and vice versa. Furthermore, social capital can accrue at different levels, such as societal, institutional, and individual levels. Hence, some individuals or institutions, including firms, can invest more in social capital than others (see Coleman (1990), Leana and Van Buren (1999), and Glaeser, Laibson, and Sacerdote (2002)).
當然，上述所有概念都是相互聯繫的。例如，公民參與可以產生信任與合作，進而促進公民的進一步參與；同樣地，合作可以建立信任，反之亦然。此外，社會資本可以在不同層面累積，例如社會、機構和個人層面。因此，一些個人或機構（包括公司）可以比其他個人或機構投資更多的社會資本（見 Coleman (1990)、Leana 和 Van Buren (1999) 以及 Glaeser、Laibson 和 Sacerdote (2002)）。

To measure social capital at the firm level, we focus on a firm’s CSR activities. We motivate this metric by noting that definitions of CSR, which generally involve aspects of civic engagement, shared beliefs, and disposition towards cooperation between the firm and its stakeholders, tend to map directly into the theoretical foundations of social capital. For example, one definition commonly used by academics and practitioners, proposed by the World Business Council for Sustainable Development (2000), is that “CSR is the commitment of a business to contribute to sustainable economic development, working with employees, their families, the local community and society at large to improve the quality of life.”
為了量度公司層面的社會資本，我們著重於公司的企業社會責任活動。企業社會責任的定義通常涉及公民參與、共同信念，以及企業與利益相關者之間的合作態度等方面，這些定義往往直接映射到社會資本的理論基礎中。例如，學者和實踐者常用的一個定義是由世界可持續發展工商理事會（2000）提出的，即「企業社會責任是企業對可持續經濟發展做出貢獻的承諾，與員工、員工家庭、當地社區和社會一起改善生活品質」。

The belief that CSR activities can help build social capital and trust is widespread among corporate managers. For example, in two recent CEO surveys conducted by PricewaterhouseCoopers (2013, 2014), CEOs indicate having plans to increase their firms’ engagement in CSR activities to restore stakeholder trust after the crisis. In contrast, academic work linking social capital, trust, and CSR is scarce but a recent book edited by Sacconi and Degli Antoni (2011) presents a series of analytical studies showing that firms can build social capital and trust through CSR investments.
企業社會責任活動有助於建立社會資本和信任的信念在企業管理者中非常普遍。例如，在普華永道會計師事務所最近進行的兩次 CEO 調查（2013 年、2014 年）中，CEO 表示有計劃增加公司參與企業社會責任活動，以在危機後恢復利害關係人的信任。相比之下，將社會資本、信任和企業社會責任聯繫起來的學術工作並不多，但 Sacconi 和 Degli Antoni (2011) 最近編輯的一本書提出了一系列分析研究，顯示企業可以透過企業社會責任投資建立社會資本和信任。

Other recent work also supports the claim that CSR builds social capital and enhances stakeholder trust in and cooperation with high-CSR firms. Eccles,
最近的其他研究也支持企業社會責任能建立社會資本、增進利害關係人對高社會責任公司的信任及合作的說法。Eccles、

Ioannou, and Serafeim (2014) show that high-CSR firms implement processes that consistently engage with stakeholders over the long term. ${}^8$${}^8$^(8){ }^{8} Bénabou and Tirole (2010) argue that stronger stakeholder engagement via CSR can lessen the likelihood of short-term opportunistic behavior by managers, a view supported by empirical evidence in Gao, Lisic, and Zhang (2014) that executives of high-CSR firms are less likely to engage in insider trading than executives of low-CSR firms. In a similar vein, Kim, Park, and Wier (2012) find that socially responsible firms are less likely to manage earnings.
Ioannou 和 Serafeim (2014) 的研究表明，高 CSR 公司實施的流程能夠長期持續與利益相關者保持聯繫。 ${}^8$${}^8$^(8){ }^{8} Bénabou和Tirole（2010）認為，通過企業社會責任加強利益相關者的參與，可以降低管理者短期機會主義行為的可能性，Gao、Lisic和Zhang（2014）的實驗證據支持了這一觀點，即高社會責任公司的高管比低社會責任公司的高管更少參與內幕交易。與此類似，Kim、Park 和 Wier (2012) 發現，有社會責任的公司較不可能進行盈餘管理。

While we acknowledge the limitations of CSR as an all-encompassing measure of firm-level social capital (see Scrivens and Smith (2013) and Sapienza, Toldra-Simats, and Zingales (2013) for a discussion of social capital metrics), we note that (i) CSR is measureable, albeit inexactly, (ii) CSR can have a nonnegative payoff (see, e.g., Edmans (2011), Servaes and Tamayo (2013), and Flammer (2015)), and (iii) firm-level CSR can change through investment or depreciation. Taken together, these three features alleviate Solow’s (1995) reservations about social capital. ${}^9$${}^9$^(9){ }^{9}
儘管我們承認企業社會責任（CSR）作為公司層面社會資本的全面度量存在局限性（有關社會資本度量的討論，請參閱 Scrivens and Smith (2013) 和 Sapienza, Toldra-Simats, and Zingales (2013)），但我們注意到 (i) 企業社會責任是可以度量的，儘管並不精確；(ii) 企業社會責任可能具有非負報酬（請參閱 Edmans (2011)、Servaes and Tamayo (2013) 和 Flammer (2015)）；以及 (iii) 公司層面的企業社會責任可以通過投資或折舊發生變化、Edmans (2011)、Servaes 和 Tamayo (2013) 以及 Flammer (2015)）；(iii) 企業層級的 CSR 可以透過投資或折舊而改變。綜合來看，這三個特點減輕了Solow (1995)對社會資本的保留。 ${}^9$${}^9$^(9){ }^{9}

In this paper, we argue that if a firm’s social capital helps build stakeholder trust and cooperation, it should pay off more when being trustworthy is more valuable, such as during an unexpectedly low-trust period.
在本文中，我們認為如果公司的社會資本有助於建立利害關係人的信任與合作，那麼在值得信任的價值較高時，例如在意外的低信任時期，公司的社會資本應該會有較高的報酬。

From a shareholder perspective, Guiso, Sapienza, and Zingales (2008, p. 2557) posit that “the decision to invest in stocks requires not only an assessment of the risk-return trade-off given the existing data, but also an act of faith (trust) that the data in our possession are reliable and that the overall system is fair.” During an unexpected decline in the general level of trust, outside shareholders are likely to be more concerned that the financial information they previously relied upon to guide investment decisions may not be credible. As such, they will seek metrics such as social capital ratings that speak to a firm’s values and integrity, placing a valuation premium on firms that are deemed to be more trustworthy.
從股東的角度來看，Guiso, Sapienza, and Zingales (2008, p. 2557)認為「投資股票的決策不僅需要評估現有資料下的風險-回報權衡，還需要相信（信任）我們所擁有的資料是可靠的，整體系統是公平的」。在整體信任度意外下降時，外部股東可能會更擔心他們之前依賴以引導投資決策的財務資訊可能不可信。因此，他們會尋求社會資本評等等能反映公司價值與誠信的指標，對被視為更值得信賴的公司給予估值溢價。

From the perspective of other stakeholders (e.g., employees, customers, suppliers, and the community at large), much of their interaction with the firm occurs through implicit or incomplete contracts, which may not be honored by either party during a crisis. Social capital could facilitate these interactions by fostering trust and cooperation (Putnam (1993)) and by reducing the need for formal contracts (Knack and Keefer (1997)). For example, stakeholders may perceive that the probability of breaching (implicit) contracts is lower for high-social-capital firms due to shared values and cooperative norms. Likewise,
從其他利害關係人 (例如員工、客戶、供應商和社區) 的角度來看，他們與公司之間的互動大多是透過隱含或不完整的契約進行的，而在危機期間，任何一方都可能無法履行這些契約。社會資本可以促進信任與合作 (Putnam (1993))，並減少對正式契約的需求 (Knack and Keefer (1997))，從而促進這些互動。例如，利害關係人可能會認為，高社會資本公司因為有共同的價值觀與合作規範，所以違反（隱性）契約的機率較低。同樣地

stakeholders are more likely to “do whatever it takes” to help high-social-capital firms weather a crisis, given that such firms displayed greater attention to, and cooperation with stakeholders in the past. ${}^{10}$${}^{10}$^(10){ }^{10} This observation is consistent with the notion of reciprocity often discussed in studies of social capital and with prior work showing that stakeholders tend to cooperate more when they perceive firms to be trustworthy (e.g., for employees, see Guiso, Sapienza, and Zingales (2015); for customers, see Servaes and Tamayo (2013)). The benefits of social capital derived from stakeholder cooperation may be present during any crisis, but as Sapienza and Zingales (2012) emphasize, cooperation breaks down without trust. As such, firm-level social capital becomes even more relevant when the level of trust in corporations, institutions, and capital markets plummets, as occurred during the 2008-2009 financial crisis.













評價方法 ${}^{10}$${}^{10}$^(10){ }^{10} 這一觀察與社會資本研究中經常討論的互惠概念一致，也與之前的研究顯示利益相關者在認為公司值得信賴時傾向於更多合作一致（例如，對於員工，請參見 Guiso、Sapienza 和 Zingales (2015)；對於客戶，請參見 Servaes 和 Tamayo (2013)）。在任何危機中，利益相關者合作所產生的社會資本效益都可能存在，但正如 Sapienza 和 Zingales (2012) 所強調的，沒有信任，合作就會破裂。因此，當企業、機構和資本市場的信任度驟降時，公司層級的社會資本就變得更加重要，就像 2008-2009 年金融危機期間發生的情況一樣。

To construct our sample, we gather information on firms’ CSR ratings from the MSCI ESG Stats Database, which contains environmental, social, and governance ratings of large publicly traded companies. ${}^{11}$${}^{11}$^(11){ }^{11} This database contains yearly ratings on roughly the 3,000 largest U.S. companies and has been used in numerous studies examining the effect of CSR on firm performance (e.g., Hong and Kostovetsky (2012), Deng, Kang, and Low (2013), Servaes and Tamayo (2013), Krüger (2015), and Borisov, Goldman, and Gupta (2016)). ${}^{12}$${}^{12}$^(12){ }^{12} ESG Stats classifies environmental, social, and governance performance into 13 different categories: community, diversity, employee relations, environment, human rights, product, alcohol, gambling, firearms, military, nuclear, tobacco, and corporate governance. As in Servaes and Tamayo (2013), we focus on the first five of these categories. We do not include the product category in our main analyses because it contains a number of elements that we consider to be outside the scope of CSR, such as product quality and innovation; our findings are unchanged, however, if we include the product category in our measure of CSR. Similarly, we do not consider in our tests the ESG Stats categories that
為了建構我們的樣本，我們從 MSCI ESG Stats 資料庫中收集公司的企業社會責任評級資訊，該資料庫包含大型上市公司的環境、社會和治理評級。 ${}^{11}$${}^{11}$^(11){ }^{11} 該資料庫包含了大約 3,000 家最大的美國公司的年度評級，並已被用於許多研究企業社會責任對公司業績影響的研究中（例如，Hong 和 Kostovetsky (2012)、Deng、Kang 和 Low (2013)、Servaes 和 Tamayo (2013)、Krüger (2015)，以及 Borisov、Goldman 和 Gupta (2016)）。 ${}^{12}$${}^{12}$^(12){ }^{12} ESG Stats 將環境、社會和治理績效分為 13 個不同的類別：社區、多元化、員工關係、環境、人權、產品、酒精、賭博、槍械、軍事、核能、煙草和公司治理。如同 Servaes 和 Tamayo (2013)，我們著重於前五個類別。我們在主要分析中不包括產品類別，因為它包含了許多我們認為不屬於企業社會責任的元素，例如產品品質和創新；然而，如果我們在衡量企業社會責任的標準中包括產品類別，我們的研究結果將維持不變。同樣，我們在測試中也不考慮 ESG Stats 類別，這些類別包括

penalize participation in the six industries that are considered controversial, as there is nothing incremental that firms operating in these industries can do to change their score except exit those industries (in addition, we control for industry in all of our tests). Finally, we do not include the ESG Stats corporate governance category in our main tests because governance is generally not part of a firm’s CSR remit. However, as the governance category in aggregate, or some of the individual governance category components, may be correlated with the trustworthiness of a firm, we examine this category in robustness tests.
懲罰參與六個被認為有爭議的行業，因為在這些行業經營的公司除了退出這些行業之外，沒有任何增量可以改變他們的得分（此外，我們在所有測試中都控制了行業）。最後，我們在主要測試中不包括 ESG Stats 公司治理類別，因為治理通常不是公司企業社會責任的一部分。然而，由於總體治理類別或個別治理類別的某些組成部分可能與公司的可信度相關，因此我們在穩健性測試中檢查了這一類別。

For each of the five categories we consider, ESG Stats compiles data on both strengths and concerns. We are interested in capturing both elements; accordingly, we construct a net CSR measure that adds strengths and subtracts concerns. As the maximum number of strengths and concerns for any given category varies over time (e.g., the maximum number of strengths for community is seven in 2005, but only four in 2010), we scale the strengths (concerns) for each category by dividing the number of strengths (concerns) for each firm-year by the maximum number of strengths (concerns) possible for that category in that year. This procedure yields strength and concern indices that range from zero to one for each category-year. Our measure of net CSR involvement in each category-year is then obtained by subtracting the concerns index from the strengths index. The net CSR index per category therefore ranges from -1 to +1 . Finally, to obtain our primary explanatory variable, a firm’s total net CSR index (CSR hereafter), we combine the net CSR indices for the categories of community, diversity, employee relations, environment, and human rights. This is a net measure across our set of stakeholder-oriented categories, and ranges from -5 to +5 . There is substantial variation in this measure across firms and industries. For example, in 2006, in the Apparel Retail Industry, GAP has a score of 0.40 while Limited Brands scores -0.53 ; in Chemicals, the score for Air Products and Chemicals is 0.16 , while that for Celanese is $-1.36{}^{13}$$-1.36{}^{13}$-1.36^(13)-1.36{ }^{13}
對於我們考慮的五個類別中的每個類別，ESG Stats 都會編製有關優點和關注事項的資料。我們有興趣捕捉這兩個元素；因此，我們建構了一個淨 CSR 量度，增加優勢並減去顧慮。由於任何給定類別的優勢和關注的最大數目會隨時間變化（例如，社區的最大優勢數目在 2005 年為 7，但在 2010 年只有 4），因此我們將每個公司年度的優勢（關注）數目除以該年度該類別可能的最大優勢（關注）數目，從而縮放每個類別的優勢（關注）數目。此程序產生每個類別年份的優勢和關注指數，範圍從 0 到 1。然後從優勢指數中減去關注指數，就得到每個類別年度的企業社會責任淨參與程度。因此，每個類別的淨 CSR 指數範圍從 -1 到 +1 不等。最後，我們將社區、多元化、員工關係、環境和人權等類別的 CSR 淨指數合併，得到我們的主要解釋變數，即公司的 CSR 總淨指數（以下簡稱 CSR）。這是我們以利益相關者為導向的類別的淨衡量，範圍從 -5 到 +5。在不同的公司和產業中，這個指標有很大的差異。例如，2006 年，在服裝零售行業，GAP 的得分為 0.40，而 Limited Brands 的得分為 -0.53；在化工行業，Air Products and Chemicals 的得分為 0.16，而 Celanese 的得分為 $-1.36{}^{13}$$-1.36{}^{13}$-1.36^(13)-1.36{ }^{13} 。

We obtain stock return data from CRSP and accounting data from Compustat. We remove financial firms from our sample due to the extensive amount of government support given to such firms during the crisis. We also remove micro-cap stocks (those with a market capitalization below $\mathrm{\$}250$$\mathrm{\$}250$$250\$ 250 million as of year-end 2007) because these stocks tend to have low liquidity and high bid-ask spreads, and are subject to more price pressure effects of trading, all of which would likely be more pronounced during the financial crisis.
我們從 CRSP 取得股票回報資料，並從 Compustat 取得會計資料。由於政府在危機期間對金融公司提供了大量支持，因此我們將金融公司從樣本中剔除。我們也剔除了微型市值股票（截至 2007 年底市值低於 $\mathrm{\$}250$$\mathrm{\$}250$$250\$ 250 百萬的股票），因為這些股票的流動性往往較低，買賣差價較高，受到的交易價格壓力效應也較大，而這些效應在金融危機期間可能會更加明顯。

As in Lins, Volpin, and Wagner (2013), we define the financial crisis as the period from August 2008 to March 2009. August of 2008 preceded the September 2008 Lehman Brothers bankruptcy, while March of 2009 is when the S&P 500 hit its lowest point of the crisis. This period also corresponds to the time of a severe decline in trust as suggested by Sapienza and Zingales (2012) (see also
如同 Lins、Volpin 和 Wagner (2013)，我們將金融危機定義為 2008 年 8 月至 2009 年 3 月。2008 年 8 月是 2008 年 9 月雷曼兄弟破產之前，而 2009 年 3 月則是標準普爾 500 指數跌至危機最低點之時。這段期間也是 Sapienza 和 Zingales (2012) 所說的信任度嚴重下降的時期（另見

Tonkiss (2009)). The decline in trust later in 2008 is also corroborated by the Trust Barometer developed by Edelman, the world’s largest independent public relations firm, which conducts global surveys of trust in business, government, NGOs, and the media-they report that trust in business in the United States declined from $58\mathrm{\%}$$58\mathrm{\%}$58%58 \% in early 2008 to $38\mathrm{\%}$$38\mathrm{\%}$38%38 \% in early 2009. ${}^{14}$${}^{14}$^(14){ }^{14}
Tonkiss (2009)）。全球最大的獨立公關公司愛德曼（Edelman）所發展的「信任晴雨表」（Trust Barometer）也證實了2008年後期信任度的下降，該公司對企業、政府、非政府組織和媒體進行全球信任度調查--他們的報告指出，美國企業的信任度從2008年初的 $58\mathrm{\%}$$58\mathrm{\%}$58%58 \% 下降到2009年初的 $38\mathrm{\%}$$38\mathrm{\%}$38%38 \% 。 ${}^{14}$${}^{14}$^(14){ }^{14}

The main stock return measures for each firm are Raw Crisis-Period Return, which is the firm’s raw buy-and-hold return from August 2008 through March 2009, and Abnormal Crisis-Period Return, which is the raw return minus the expected return, based on the market model estimated over the 60 -month period ending in July 2008. To avoid problems with outliers, we winsorize these returns at the 1st and 99th percentiles. We relate these return measures to our CSR measure for the year 2006 to guard against the possibility that by yearend 2007 firms may have already changed their CSR policies in anticipation of the crisis ahead. ${}^{15}$${}^{15}$^(15){ }^{15}
每家公司的主要股票回報指標為原始危機期間回報，即公司自 2008 年 8 月至 2009 年 3 月的原始買入並持 有回報，以及異常危機期間回報，即原始回報減去預期回報，該預期回報是根據截至 2008 年 7 月的 60 個月期間的市場模型估算得出的。為了避免異常值的問題，我們在第 1 個百分位數和第 99 個百分位數將這些回報率勝數化。我們將這些回報量與 2006 年的企業社會責任量相關聯，以避免在 2007 年年底時，企業可能已經因預期未來的危機而改變其企業社會責任政 策。 ${}^{15}$${}^{15}$^(15){ }^{15}

After combining nonfinancial firms with sufficient data coverage on the CRSP and Compustat databases and firms on the ESG Stats database, we obtain a sample of 1,673 nonfinancial firms for which all explanatory variables are available for the crisis period.
將 CRSP 和 Compustat 資料庫中有足夠資料覆蓋的非金融公司，以及 ESG Stats 資料庫中的公司合併後，我們得到了 1,673 家非金融公司的樣本，這些公司在危機期間的所有解釋變數都是可用的。

Table I provides descriptive statistics for our main variables. The first row of Panel A shows that our primary variable of interest, $CSR$$CSR$CSRC S R, is slightly negative with a mean value of -0.165 and a median value of -0.200 . Thus, the average and median firm has more CSR concerns than strengths, consistent with Deng, Kang, and Low (2013), Servaes and Tamayo (2013), and Borisov, Goldman, and Gupta (2016). The next row shows that Raw Crisis-Period Return is strongly negative, with a mean of $-39.1\mathrm{\%}$$-39.1\mathrm{\%}$-39.1%-39.1 \%, a median of $-40.3\mathrm{\%}$$-40.3\mathrm{\%}$-40.3%-40.3 \%, and a 25 th percentile value of $-59.5\mathrm{\%}$$-59.5\mathrm{\%}$-59.5%-59.5 \%, indicating that investors and other stakeholders were likely quite concerned about the survival prospects of many of the firms they held in their portfolios, worked for, or interacted with in business transactions. The median abnormal return is close to zero at $1.3\mathrm{\%}$$1.3\mathrm{\%}$1.3%1.3 \%, while the mean is $11.6\mathrm{\%}$$11.6\mathrm{\%}$11.6%11.6 \%. Panel A also provides definitions and descriptive statistics for firm characteristics that we use as control variables in our models; Panel B presents a correlation matrix of all the variables employed in our main analyses.
表 I 提供了主要變數的描述性統計資料。面板 A 的第一行顯示，我們關注的主要變量 $CSR$$CSR$CSRC S R 略為負值，平均值為 -0.165，中位值為 -0.200。因此，與 Deng、Kang 和 Low (2013)、Servaes 和 Tamayo (2013) 以及 Borisov、Goldman 和 Gupta (2016) 的觀點一致，平均值和中位數企業的企業社會責任關注多於優勢。下一行顯示，原始危機期間回報率為強烈負值，平均值為 $-39.1\mathrm{\%}$$-39.1\mathrm{\%}$-39.1%-39.1 \% ，中位值為 $-40.3\mathrm{\%}$$-40.3\mathrm{\%}$-40.3%-40.3 \% ，第 25 百分位數值為 $-59.5\mathrm{\%}$$-59.5\mathrm{\%}$-59.5%-59.5 \% ，顯示投資人和其他利益相關者可能相當關心他們投資組合中持有、為其工作或在商業交易中與其互動的許多公司的生存前景。異常報酬的中位數是接近零的 $1.3\mathrm{\%}$$1.3\mathrm{\%}$1.3%1.3 \% ，而平均值則是 $11.6\mathrm{\%}$$11.6\mathrm{\%}$11.6%11.6 \% 。面板 A 也提供了我們在模型中作為控制變數的公司特徵的定義和描述性統計資料；面板 B 則是我們主要分析中採用的所有變數的相關矩陣。

The sample consists of 1,673 firms with CSR data available from the MSCI ESG STATS database as of year-end 2006 and returns available during the
樣本包括 1,673 家公司，這些公司截至 2006 年底的企業社會責任資料可從 MSCI ESG STATS 資料庫取得，而回報則可在 2007 年的回報期內取得。
period August 2008 to March 2009. CSR is the total net (strengths minus concerns) CSR score computed using five stakeholder-oriented categories
2008 年 8 月至 2009 年 3 月。CSR 是以五個利害關係人為導向的類別計算出的 CSR 總淨得分（優點減去顧慮）。
(environment, employee relations, human rights, community, and diversity). To compute the total net CSR measure, we first compute the net CSR
(環境、員工關係、人權、社區和多樣性）。為了計算企業社會責任淨值，我們首先計算企業社會責任淨值。
index within each of the categories. The net CSR index for each category is computed by taking the number of strengths identified for a given firm
每個類別中的指數。每個類別的企業社會責任淨指數的計算方法是將特定公司的優勢數目
and dividing this by the maximum possible strengths in that category, and then subtracting the number of concerns identified for the firm divided by
並將其除以該類別中可能的最大優勢，然後再減去為公司找出的關注事項數目，再除以
the maximum possible concerns; the net CSR score for each category thus ranges from - 1 to +1. The total net CSR measure, CSR, is computed as the
因此，每個類別的 CSR 淨得分範圍從 - 1 到 +1。CSR 總淨值計算為
sum of the net CSR indices for the five categories and ranges from -5 to +5. CSR ratings are measured at the end of 2006. Crisis-Period Raw Return is
五個類別的 CSR 淨指數總和，範圍從 -5 到 +5。CSR 評等是在 2006 年底測量。危機期間原始報酬率為
the raw return computed over the period August 2008 to March 2009. Crisis-Period Abn. Return is the market model-adjusted return over the period
2008 年 8 月至 2009 年 3 月期間計算的原始回報。危機期間 Abn.回報為該期間經市場模型調整後的回報。
August 2008 to March 2009, with market model parameters computed over the five-year period ending in July 2008 using the CRSP value-weighted
2008 年 8 月至 2009 年 3 月，在截至 2008 年 7 月的五年期間，使用 CRSP 價值加權計算市場模型參數。
index as the market proxy. Accounting data are based on the last quarter ending at or before the end of 2007. Market Capitalization is in millions
指數作為市場代表。會計資料以截至 2007 年底或之前的最後一個季度為基礎。市值以百萬計
of dollars. Long-Term Debt is computed as long-term debt divided by assets. Short-Term Debt is computed as debt in current liabilities divided by
美元。長期債務以長期債務除以資產計算。短期債務的計算方式為流動負債中的債務除以
assets. Cash Holdings is computed as cash and marketable securities divided by assets. Profitability is computed as operating income divided by
資產。現金持有量以現金及有價證券除以資產計算。獲利能力以營業收入除以
assets. Book-to-Market is computed as book value of equity divided by market value of equity. Negative B/M is a dummy variable set to one when
資產。帳面市值（Book-to-Market）的計算方式為股本帳面價值除以股本市值。負 B/M 是一個虛變數，當
the book-to-market ratio is negative and zero otherwise. Momentum is the raw return over the period August 2007 to July 2008. Idiosyncratic Risk
否則為零。動量為 2007 年 8 月至 2008 年 7 月期間的原始回報。離散風險
is computed as the residual variance from the market model estimated over the five-year period ending in July 2008, using monthly data. Financial
計算為市場模型在截至 2008 年 7 月的五年期間估算的殘差，使用每月資料。財務
firms and micro-cap firms, which we define as firms with a market capitalization below $\mathrm{\$}250$$\mathrm{\$}250$$250\$ 250 million as of year-end 2007, are removed from the
公司和微型市值公司（我們將其定義為截至 2007 年底市值低於 $\mathrm{\$}250$$\mathrm{\$}250$$250\$ 250 百萬美元的公司）從中剔除。
sample. Control variables and returns are winsorized at the 1st and 99th percentiles.
樣本。控制變數與回報以第 1 與第 99 百分位數為 winsorized。

Table I-Continued 表一-續

We estimate various regression models of stock returns during the crisis period as a function of firms’ pre-crisis CSR ratings and a number of control variables. Panel A of Table II contains our baseline regression models. The dependent variable in columns (1) and (3) is Raw Crisis-Period Return, while in columns (2) and (4), it is Abnormal Crisis-Period Return. Our variable of interest is the firm’s CSR measured at year-end 2006. In all models, we include industry dummies (defined at the two-digit SIC level) because some industries may be more likely to invest in CSR than others and may have been differentially affected by the financial crisis. We also control for the firm’s factor loadings based on the Fama-French three-factor model plus the momentum factor. ${}^{16}$${}^{16}$^(16){ }^{16}
我們估算了危機期間股票報酬率與公司危機前 CSR 評級及多個控制變數的各種迴歸模型。表 II 的面板 A 包含了我們的基線回歸模型。第(1)列和第(3)列中的因變量是原始危機期回報，而第(2)列和第(4)列中的因變量是異常危機期回報。我們所關心的變數是公司在 2006 年年底的企業社會責任。在所有模型中，我們都加入了行業虛數（定義在兩位數 SIC 層級），因為某些行業可能比其他行業更傾向於投資 CSR，也可能受到金融危機的不同影響。我們還根據 Fama-French 三因子模型加上動量因子來控制公司的因子載荷。 ${}^{16}$${}^{16}$^(16){ }^{16}

Columns (1) and (2) show that firms with higher CSR ratings performed significantly better during the crisis. The effect of CSR on returns is economically large: a one-standard-deviation increase in $\mathrm{CSR}$$\mathrm{CSR}$CSR\operatorname{CSR} ( 0.381 ) is associated with a 2.25 percentage point increase in raw returns and a 4.15 percentage point increase in abnormal returns during the crisis.
列 (1) 和 (2) 顯示，企業社會責任評等較高的公司在危機中表現明顯較好。企業社會責任對報酬率的影響在經濟學上是很大的： $\mathrm{CSR}$$\mathrm{CSR}$CSR\operatorname{CSR} ( 0.381 ) 的一個標準差增加與危機期間原始報酬率增加 2.25 個百分點及異常報酬率增加 4.15 個百分點有關。

One concern with the specifications reported in columns (1) and (2) is that the strong performance of high-CSR firms during the crisis may be due to omitted variables that happen to be correlated with CSR, rather than due to CSR itself. To address this possibility, in columns (3) and (4), we control for a firm’s financial health in the year before the crisis and for other firm characteristics that have been found to affect stock returns. We employ several proxies to measure a firm’s financial health and, thus, its ability to withstand a severe downturn in the economy: Cash Holdings (cash and marketable securities divided by assets), Short-Term Debt (debt in current liabilities divided by assets), Long-Term Debt (long-term debt divided by assets), and Profitability (operating income divided by assets). During a crisis, profitable, cash-rich firms with low debt can continue investing, while other firms may be forced to cut investment, especially if they have short-term debt maturing during the crisis (see Duchin, Ozbas, and Sensoy (2010), Almeida et al. (2012), and Harford, Klasa, and Maxwell (2014) for empirical evidence consistent with these expectations).
第(1)列和第(2)列所報告的規格的一個顧慮是，高 CSR 公司在危機中的強勁表現可能是由於遺漏了恰好與 CSR 相關的變量，而不是由於 CSR 本身。針對這個可能性，我們在第(3)列和第(4)列中控制了公司在危機前一年的財務健康狀況，以及其他被發現會影響股票回報的公司特徵。我們採用幾個代用指標來衡量公司的財務健康狀況，從而衡量其抵禦經濟嚴重衰退的能力：現金持有量（現金和有價證券除以資產）、短期債務（流動負債中的債務除以資產）、長期債務（長期債務除以資產）和盈利能力（營業收入除以資產）。在危機期間，獲利豐厚、現金充裕且債務低的公司可以繼續投資，而其他公司可能會被迫削減投資，尤其是在危機期間有短期債務到期的情況下（請參閱 Duchin、Ozbas 和 Sensoy (2010)、Almeida 等人 (2012) 以及 Harford、Klasa 和 Maxwell (2014) 與上述預期相符的實證證據）。

Because additional firm characteristics may also be important for stock returns (see, e.g., Daniel and Titman (1997)), we also control for Size (the log of a firm’s equity market capitalization), Book-to-Market (book value of equity divided by market value of equity), and Momentum (the firm’s raw return over the period August 2007 to July 2008). We also add a dummy for firms with a Negative Book-to-Market ratio, as such firms are likely distressed and hence their returns may behave more like those of high book-to-market firms than low book-to-market firms (see Fama and French (1992)). Finally, we control
由於其他公司特徵也可能對股票回報有重要影響（請參閱 Daniel and Titman (1997)），因此我們也控制了規模（公司股票市值的對數）、市賬率（股票賬面值除以股票市值） 和動量（公司在 2007 年 8 月至 2008 年 7 月期間的原始回報）。我們也為負市賬率的公司加入一個虛值，因為這類公司很可能陷入困境，因此它們的回報可能比低市賬 率的公司表現得更像高市賬率的公司（見 Fama and French (1992)）。最後，我們控制

This table presents regression estimates of crisis-period returns on CSR and control variables. Crisis-period returns are measured as both raw buy and hold returns and abnormal returns over the period August 2008 to March 2009. In Panel A, we use a linear measure of CSR, CSR, which is the net (strengths minus concerns) CSR score computed using five stakeholder-oriented categories, measured at the end of 2006. In Panel B, we use dummy variables for CSR quartiles such that CSR2 takes the value of one if the firm is in the second CSR quartile and zero otherwise, CSR3 takes the value of one if the firm is in the third CSR quartile and zero otherwise, and CSR4 takes the value of one if the firm is in the fourth CSR quartile and zero otherwise. In Panel C, we employ the linear measure of CSR and add measures of corporate governance. ESG Stats Governance Index is the net CSR index for the governance category and is computed by taking the number of strengths, divided by the possible maximum, and subtracting the number of concerns, divided by the possible maximum; this measure thus ranges from -1 to +1 . The $E$$E$EE-Index is the sum of six dummies reflecting the following antitakeover provisions: (i) a staggered board, (ii) limits to amend the charter, (iii) limits to amend bylaws, (iv) supermajority voting requirements, (v) golden parachutes for executives, and (vi) the ability to adopt a poison pill (see Bebchuk, Cohen, and Ferrell (2009)), obtained from MSCI Governance Metrics. Board Independence (fraction of board consisting of outside directors), Board Size, a dummy if the CEO Is Not the Chairman, and Board Ownership (fraction of outstanding shares owned by board members) are obtained from the MSCI Directors database. When the governance metrics are not available on the MSCI databases, we set them to zero and code a missing variable dummy that we set to one if that governance item is missing. These dummies are included in all models, but their coefficients are not reported. The control variables are as defined in Table I. Industry dummies are defined at the two-digit SIC code level. Financial firms and micro-cap firms with a market capitalization below $\mathrm{\$}250$$\mathrm{\$}250$$250\$ 250 million are removed from the sample. The control variables and returns are winsorized at the 1st and 99th percentiles. Heteroskedasticity-consistent standard errors are presented in parentheses. ${}^{\ast \ast \ast },{}^{\ast \ast }$${}^{\ast \ast \ast },{}^{\ast \ast }$^(******),^(****){ }^{* * *},{ }^{* *}, and * indicate that the parameter estimate is significantly different from zero at the $1\mathrm{\%},5\mathrm{\%}$$1\mathrm{\%},5\mathrm{\%}$1%,5%1 \%, 5 \%, and $10\mathrm{\%}$$10\mathrm{\%}$10%10 \% level, respectively.
本表顯示危機期間回報對企業社會責任和控制變數的回歸估計值。危機期間的報酬率是以 2008 年 8 月至 2009 年 3 月的原始買入與持有報酬率以及異常報酬率來衡量。在面板 A 中，我們使用企業社會責任的線性測量，即 CSR，它是使用五個利害關係人導向的類別計算出的企業社會責任淨得分（優點減去顧慮），測量時間為 2006 年底。在面板 B 中，我們使用虛變數來表示 CSR 四分位數，如果公司位於 CSR 第二 四分位數，則 CSR2 的值為 1，否則為 0；如果公司位於 CSR 第三 四分位數，則 CSR3 的值為 1，否則為 0；如果公司位於 CSR 第四 四分位數，則 CSR4 的值為 1，否則為 0。在面板 C 中，我們採用企業社會責任的線性衡量方法，並加入公司治理的衡量方法。ESG Stats Governance Index 是治理類別的企業社會責任淨指數，其計算方法是取優勢的數目除以可能的最大值，再減去顧慮的數目除以可能的最大值；因此，這個量度的範圍是 -1 到 +1 。 $E$$E$EE 指數是反映以下反收購規定的六個假指標之和：(i) 交錯董事會、(ii) 修改憲章的限制、(iii) 修改附例的限制、(iv) 超過大多數投票規定、(v) 高管的黃金降落傘，以及 (vi) 採用毒丸的能力（請參閱 Bebchuk、Cohen 和 Ferrell (2009)），這些假指標來自 MSCI Governance Metrics。董事會獨立性 (由外部董事組成的董事會比例)、董事會規模、CEO 非董事長的假指標，以及董事會所有權 (董事會成員擁有的已發行股份比例) 均取自 MSCI Directors 資料庫。 當 MSCI 資料庫中沒有管治指標時，我們將其設置為 0，並編碼一個遺漏變數假設，如果該管治項目遺漏，則將其設置為 1。這些虛數包含在所有模型中，但其係數並未報告。控制變數的定義如表 I 所示。行业虚拟变量定义在两位数 SIC 代码级别。市值低於 $\mathrm{\$}250$$\mathrm{\$}250$$250\$ 250 百萬的金融公司和微型公司從樣本中剔除。控制變數和收益率在第 1 個和第 99 個百分位數進行勝數化處理。括號中為異方差一致的標準誤 差。 ${}^{\ast \ast \ast },{}^{\ast \ast }$${}^{\ast \ast \ast },{}^{\ast \ast }$^(******),^(****){ }^{* * *},{ }^{* *} 和*分別表示參數估計值在 $1\mathrm{\%},5\mathrm{\%}$$1\mathrm{\%},5\mathrm{\%}$1%,5%1 \%, 5 \% 和 $10\mathrm{\%}$$10\mathrm{\%}$10%10 \% 水平上與零有顯著差異。

(Continued) (續)

Table II-Continued 表二-續