Journal of Management and Governance 8: 255277, 2004.

2004 Kluwer Academic Publishers. Printed in the Netherlands.

255

An Examination of Socially Responsible Firms Board Structure1

ELIZABETH WEBB
School of Business Administration, 5998 Alcala Park, San Diego, CA 92110, USA, E-mail: webb@sandiego.edu

Abstract. This study investigates the structure of the board of directors at socially responsible (SR) rms. Using a sample of 394 SR rms and comparing these to a matched sample of rms, I nd that SR rms have characteristics associated with eective board structures. For instance, SR rms have more outsiders and women directors, and less instance of CEO/Chairman duality than non-SR rms. Results are similar when using a continuous measure of social responsibility. Also, I document that SR rms have higher Governance Index scores than the matched sample. Overall, this suggests that a reason for shareholders appeal in socially responsible rms and mutual funds may be because these rms have stronger governance mechanisms in place than do non-SR rms. In addition, it appears that eective governance structures are more likely to exist in rms that focus on a broad range of stakeholders, rather than in rms that have a strict focus on shareholder wealth maximization. Key words: agency theory, boards of directors, corporate governance, social responsibility, socially responsible investing

1. Introduction With the crash of tech stocks in the late 1990s and the proliferation of corporate accounting scandals a few years later, corporate governance has become a central issue in the business world. Since the cost of personally monitoring rms is too high for individual investors to bear, they must rely on governance mechanisms whose function it is to monitor manager activity on behalf of the shareholders. Without a strong governance system in place, managers may take advantage of their insider position at the expense of shareholders. It is precisely this activity that has caused decreases in overall stock market investments, yet one sector of the market remained strong. Investments in so-called socially responsible rms (SR) and mutual funds have increased, despite the fact that historical returns on these investments are not dierent from other rms and funds.2 In fact, in February 2003 Lipper, a rm that tracks 80,000 mutual funds

256

ELIZABETH WEBB

worldwide, reported that SR mutual funds experienced a net inow of $1.5 billion in 2002, while the non-designated SR funds in the US experienced a total outow of $10.5 billion. Interest in socially responsible rms is not a new phenomenon. Since the august 1971 launch of the Pax World Balance Fund, the rst mutual fund in the US to use broad-based social and nancial criteria for screening purposes, green mutual funds have ourished in the past three decades. SR funds generally are comprised of the equity of rms that are screened on the basis of several characteristics of stakeholder awareness including community investment, environmental issues, employment practices, and shareholder rights. According to the Social Investment Forum, nearly $1 of every $8 under professional management in the US is in a SR portfolio, and social investment grew from $1.185 trillion in 1997 to $2.16 trillion in 1999. This growth rate is roughly twice that of all assets under professional management in the US. By 2003, Lipper reported that 199 SR investment funds were in existence, up from 88 funds in 1999. Despite increases in investor interest in SR rms, there is little evidence suggesting that these rms have stronger nancial performance than their competitors. A reason for this may simply be that some investors prefer socially responsible rms because these rms do more social good. The fact that these rms are deemed SR implies that they are concerned with rm stakeholders. Since stakeholder consideration is central to Tiroles (2001) view of corporate governance, these rms may have stronger governance structures in place. Strong governance may be important to all investors (especially in the midst of corporate transgressions), regardless of personal or institutional interest in corporate social responsibility. The purpose of this study, therefore, is to investigate corporate governance structures of SR rms, since interest in strong monitoring in light of todays corporate indignities may inuence investor choice. Specically, I analyze the characteristics of the boards of directors of SR and non-SR rms. The results support the hypothesis that SR rms possess characteristics associated with stronger board structure. Johnson (1966) called for more empirical work to be done in analyzing distinctions in SR rm behavior, and 40 years later, further exploration is still needed. This study adds to the growing literature on the SR rms and corporate governance by, specically, showing how rms characterized as SR are better able to minimize agency problems through eective board structure arrangements. Additionally, the results suggest that rms with strong boards in place are more likely to be rms focused on a variety of stakeholders. This study also helps to ll a research gap that exists concerning the relationship between SR rms and governance structure. The rest of the paper is organized as follows: Section 2 reviews the literature on strong board structure characteristics, and socially responsible

AN EXAMINATION OF SOCIALLY RESPONSIBLE FIRMS

257

rms, and presents hypotheses; Section 3 describes the data and methodology; Section 4 analyzes the results; and Section 5 concludes.

2. Literature Review and Hypotheses 2.1.

DEFINING STRONG BOARD STRUCTURE

According to Jensen and Meckling (1976), the board of directors acts on shareholders behalf to monitor managers as a market solution to the contracting problems inherent in organizations. The results of over 20 years of board structure research indicate that boards that are best able to do this are those that can best alleviate agency problems between shareholders and managers through independent internal control. In a survey paper on corporate governance and board eectiveness, John and Senbet (1998) nd that empirical governance literature suggests that the degree of board independence is related to composition, and that independence fosters board eectiveness. Several examples follow. Beasley et al. (2000) nd evidence that weak governance structure is associated with agency problems and subsequent poor performance. The authors analyze the board characteristics of rms with instances of nancial statement fraud and compare these results to a benchmark sample. Results indicate that the fraudulent companies have weak (ineective) governance structures relative to the benchmark (no-fraud) rms. As compared to fraud rms, the no-fraud rms have more outsiders on the board, shorter tenure for board members, less chance that the CEO is the founder of the rm, smaller probability that the CEO is the chairman of the board, and are more likely to have a blockholder (a shareholder holding more than 5% of the outstanding stock) on the board. Howton et al. (2001) analyze board structure and IPO underpricing. The authors show that because of the monitoring function provided by board members for shareholders of the rm, a strong board (more outside ownership) will alleviate agency problems between the two parties by reducing asymmetric information, and thus reducing the extent of IPO underpricing. Core et al. (1999) also report that rms with weaker governance structures have more agency problems. They nd an inverse relationship between board strength and characteristics such as the percentage of the board composed of inside directors, board size, gray directors (those directors who are not ocially insiders, but who perform substantial services for the rm in exchange for compensation), directors over the age of 69, busy directors (directors on three or more other boards), and CEO is also the board chair. Studies on specic board characteristics support the idea that strong boards are those that possess characteristics contributing to independence. For example, Ferris et al. (2003) nd that directors with multiple board

258

ELIZABETH WEBB

appointments (busy directors) are eective monitors. This is consistent with Shivdasani (1993), who also nds that the same argument holds for outside directors who are CEOs of other companies. In a study on founding family CEOs, Jayaraman et al. (2000) nd that the relationship between stock performance and founder management (CEO is the founder) is positive for small rms yet negative for large rms. Rosenstein and Wyatt (1997) look the relationship between stock market reaction to inside director appointments and insider stock ownership. They nd that a signicant negative and zero reaction occurs when insiders own very little or a very large portion of the rms stock, respectively. The literature emphasizes characteristics that encourage active monitoring on the part of directors. In most industries and situations, having more of the good characteristics and less of the bad is preferred. An exception to this rule is in the banking industry, where larger boards are more ecient than having few directors.3 In addition, the literature focuses primarily on publicly traded US rms and therefore consideration of private rms and foreign countries regulatory environment is needed before extending results and implications internationally.

2.2.

INCORPORATING SOCIAL RESPONSIBILITY

While corporate governance and board structure are becoming mainstream topics in academic literature, the area of social responsibility has received much less attention. Milton Friedman dened corporate social responsibility in 1970 when he said that it is to conduct the business in accordance with shareholders desires, which generally will be to make as much money as possible while conforming to the basic rules of society, both those embodied in law and those embodied in ethical custom. More recently, Carroll (1991) uses a broader denition of corporate social responsibility. According to Carroll, corporate social responsibility refers to a business entitys attention to and fulllment of responsibilities to multiple stakeholders which exist at various levels: economic, legal, ethical, and philanthropic. SR rms therefore are those rms considered to be positively aecting a broad class of stakeholders. Studies on the relationship between corporate social responsibility and nancial performance report mixed results. Waddock and Graves (1997) address the question of whether CSP and nancial performance are related. Using a constructed index from the rm Kinder, Lydenberg, and Domini and nancial statement data, they nd that CSP and protability are positively related.4 They suggest that causality goes both ways in that rms with strong nancial performance have slack resources that can be spent on CSP measures, and that good social performance may be linked

AN EXAMINATION OF SOCIALLY RESPONSIBLE FIRMS

259

to good managerial practice, which in turn leads to strong nancial performance. These results are consistent with earlier evidence by McGuire et al. (1988), who nd a positive relationship between prior stock market and accounting-based measures of performance and socially responsibility. However, in their study on the nancial eects of shareholder pressure in the boycott of South Africa, Teoh et al. (1999) nd that the divestment had no eect on stock price. Avoidance of South African investments is often used as a lter for rms in SR mutual funds. Thus it appears that despite being the most visible and successful instance of social activism aimed at corporate investment policies, the boycott of South Africa had little nancial impact. Also, Hillman and Keim (2001) show that participating in social issues that are unrelated to the core business of the rm actually hurts the rms nancial performance. Given the mixed results of SR rms nancial performance, it is worthwhile to explore any distinction in other areas of rm organization that might lead to reasons for investors sustained growth and interest in these types of rms. Tirole (2001) denes corporate governance as the design of institutions that induce or force management to internalize the welfare of stakeholders. In that vein, I analyze the characteristics in socially responsible rms boards of directors since these rms publicly acknowledge their stakeholders and therefore may have stronger governance systems in place given Tiroles view of corporate governance. The null hypothesis in this paper is that there is no dierence in the characteristics of SR and non-SR rms boards of directors. The alternative hypothesis is that SR boards possess more characteristics associated with stronger boards than do non-SR boards.

3. Data and Methodology 3.1.

DATA COLLECTION AND CHARACTERISTICS

To measure and compare the corporate governance structure characteristics of SR vs non-SR rms, the 400 rms in the DSI (as of November 2001) are used along with a matched sample. The DSI is a portfolio of ethically screened stocks from publicly traded rms in the US The social investment research rm of Kinder, Lydenberg, Domini and Company constructs the index. Firms in this index must pass multiple broad-based social screens. Firms must have a positive record of shareholder activism, community investment, environmental concerns, human rights, employment, and products and services. Firms from the alcohol, tobacco, and gambling industries, as well as rms involved in weapon and nuclear power production, are excluded from the index. The use of the DSI is appropriate as a sample of SR rms as it encompasses a wide range of social and environmental screens. An important

260

ELIZABETH WEBB

advantage for using the DSI over other available socially screened portfolios is that a group of independent researchers applies the same broad set of criteria to the rms. Each rm from the DSI is matched with a non-SR rm based on industry and size (as in Beasley (1996)). Matching by industry is an important control in board structure literature, since optimal board characteristics vary by industry (see Gillan et al. (2003)). The non-SR rms in this study are then selected by locating the rm closest in market capitalization (size) to each SR rm within the same three-digit SIC code (industry) in 2001. Thus, an initial sample of 400 SR rms and 400 non-SR rms is compiled.5 Panel A in Table I shows the characteristics of both samples from 2001. It is evident that the SR rms from the DSI are larger overall than the matched sample. The SR rms have signicantly higher net income, total assets, market value, and return on assets than the non-SR sample. This is to be expected, since many of the rms listed in the DSI are large rms within their industry, and once a rm is included in the matched sample it cannot be used as a match for a subsequent SR rm. This results in the tendency of matched rms to be smaller than SR rms within the same industry. According to Gillan et al. (2003), industry matching is important in empirical board structure research such that size dierences between samples may not aect overall results. Later, this assumption is tested in the robustness checks. Price-to-book ratios, annual returns, and return on equity are not signicantly dierent between samples, supporting the motivation for this study since investment popularity in SR rms is growing despite negligible gains in nancial returns. Data on the SR sample and the matched rms taken from 1998 to 2000 show similar results (not reported). Panel B of Table I shows the industry representation for both the SR and non-SR samples (since they are matched by industry, both have the same representation). Industries are identied using the Fama and French (1997) industry assignment schedule which apportions rms into one of 48 industries based on four-digit SIC codes. To conserve space, the 48 industries are divided into eight general categories. The DSI and the subsequent matched sample are widely distributed in terms of industry. Structural characteristics of the board of directors for both samples are taken from proxy statements issued in 2001.6 In order to test the hypothesis that SR rms have stronger governance structures in place than their nonSR counterparts, sixteen board characteristics are analyzed as outlined in Section 2. Variables collected from proxy statements include the following: the number of insiders (those directors who are currently employed by the rm or have been employed by the rm in the past), the number of outsiders (directors with no aliation to the rm), the number of gray directors (directors who are have substantial business relationships with the

AN EXAMINATION OF SOCIALLY RESPONSIBLE FIRMS

261

Table I. Sample description SR rms Panel A: Descriptive statistics Net income Total assets Market value Price-to-book Annual stock return (%) Return on assets (%) Return on Equity Non-SR rms Dierence (t-value)

297.21 (1439.80) 20989.00 (71714.00) 14383.00 (35586.00) 3.21 (9.29) 15.89 (58.77) 3.08 (12.46) 10.71 (93.21) Full sample (%)

)39.97 (3225.00) 10115.00 (43412.00) 5919.60 (18945.00) 3.53 (15.49) 10.86 (68.49) )3.16 (41.47) )49.47 (771.73)

337.20 (1.77)* 10874.00 (2.36)** 8464.00 (3.83)*** 0.32 (0.32) 5.04 (1.03) 6.24 (2.67)*** 60.18(1.44) N (in each sample)

Panel B: Industries Retail Machinery Recreation Chemicals Transportation Financial Healthcare Utilities Total

20.6 16.0 7.4 8.9 6.1 15.5 6.9 18.8 100.0

81 63 29 35 24 61 27 74 394

Panel A shows the 2001 characteristic averages of the SR rms and matched rms (non-SR) used in the sample. Standard deviations in parentheses. All numbers are in millions of dollars unless noted otherwise. Panel B describes the industry breakdown of the full sample (in % of total).

company, yet are not insiders, such as lawyers, investment bankers, and consultants), the age of the CEO, a dummy variable equal to 1 if the CEO is also the chairman of the board, a dummy variable equal to 1 if the directors are elected on a staggered (as opposed to annual) basis, a dummy variable equal to 1 if there is a blockholder on the board (an outsider holding more than 5% of the outstanding stock), a dummy variable equal to 1 if directors are related, a dummy variable equal to 1 if a director is an ocer of or works with non-prot organizations, the number of women on the board, a dummy variable equal to 1 if the CEO belongs to the rms founding family, the number of directors who are over the age of 69, the number of busy directors who are on three or more boards (six or more if the director is retired), and the number of outside

262

ELIZABETH WEBB

directors who are also CEOs of other rms. The relationship between each characteristic and board strength is reported in Table II. I use market value of equity, total assets, and b as control variables in the robustness checks. Firm characteristics, and industry and market capitalization data are collected from Compustat and CRSP, respectively. For some rms in the sample, either proxy statements were not available, or certain variables (such as CEO age) were missing.7 The nal sample of rms with all data available is 394 SR rms and 394 non-SR rms. In an attempt to measure the linear association between board structure variables, several observations can be made from the correlation matrices of each sample (not reported). Both Pearson and Spearman correlation coecients are computed. Although some correlations are signicantly dierent from zero, additional tests suggest that multicollinearity does not appear to be a problem.8 The results reveal that several structural characteristics that are signicantly correlated in one sample are not correlated in the other sample of rms. For instance, as the number of members on the non-SR board increase, so does the proportion of women on the board. However, there is also a direct correlation between the number of board members and the proportion of senior directors on non-SR boards. This may indicate that non-SR boards attempt to diversify the board when it becomes larger and older. The opposite appears to be true for SR boards where the proportion of senior directors and female directors are not correlated. As an alternative measure of corporate governance, I include a Governance Index (GI) measure as introduced by Gompers et al. (2003). This index gives a score from 0 to 24 based on the number of antitakeover provisions and governance rules that are included in a rms bylaws. For example, if the rm has an antigreenmail provision it will receive one point. The authors conclude that strongest governance rms (rms with low GI scores) have stronger performance records than high GI rms. High scores on the GI (representing rms with multiple takeover defenses) are assumed to be indicative of management entrenchment problems, where managers protect themselves from takeovers at the expense of the shareholders. However, in the presence of eective boards of directors, the management entrenchment undertone inherent in takeover defense provisions can be reduced. For instance, Malekzadeh et al. (1998) and McWilliams and Sen (1997) nd that the structure of the board inuences the markets reaction to antitakeover charter amendments. Specically, McWilliams and Sen (1997) show that the market reacts negatively to antitakeover announcements, and this reaction is more pronounced when the board is dominated by insider and gray directors, and where the CEO is also the chairman of the board. Further, Gillan et al. (2003) nd that board structure variables and charter provisions studied together have o-setting results.

AN EXAMINATION OF SOCIALLY RESPONSIBLE FIRMS

263

Table II. Summary of strong board characteristics from prior research

Characteristic Percentage of outsiders Relationship with board strength + Related research Core, Holthausen, and Larcker (1999) (CHL) Fama (1980) Beasley (1996) Howton, Howton, and Olson (2001) CHL (1999) Rosenstein and Wyatt (1997) CHL (1999) Yermack (1996) CHL (1999) Jensen (1993) Yermack (1996) Carter, Simkins, and Simpson (2003) CHL (1999) Shivdasani (1993) CHL (1999) Ferris, Jagannathan, and Pritchard (2003) Beasley (1996) Burkart, Panuzi, and Schleifer (2002) Jayaraman, Khorana, Nelling, and Covin (2000) Bacon, Cornett, and Davidson (1997) CHL (1999) Bacon, Cornett, and Davidson (1997) Jensen (1993) Jensen (1993) Shivdasani (1993) Gompers, Ishii, and Metrick (2003) Gillan, Hartzell, and Starks (2003) McWilliams and Sen (1997)

Percentage of insiders Percentage of gray directors Board size CEO is not chairman

) ) ) +

Board diversity Senior citizens on board Busy directors

+ ) +

Blockholder on board CEO is from founding family

+ )

Senior CEO Board meetings Annual elections Family directors CEO directors Directors with non-prots Governance Index

) ) + ) + + )

Percentage of outsiders indicates the fraction of the board that is not aliated with the company. Percentage of insiders represents the fraction of the total board members who are (or have been) also ocers of the company. Percentage of gray directors is those directors who have substantial business relationships with the company, yet are not insiders. Senior citizens include any director that is over the age of 69. Busy directors indicate the percentage of directors who are on three or more other boards (six or more if board member is retired). Blockholder indicates whether or not an outside director holds more than 5% of the outstanding stock of the rm. Senior CEO indicates that the CEO is over the age of 69. Family indicates the presence of related board members. CEO directors represent the percentage of outside directors who are also CEOs of other rms. GI represents the Governance Index antitakeover score as measured by Gompers et al. (2003).

264

ELIZABETH WEBB

Since board structure can alter the nature of antitakeover amendments, rms with stronger board characteristics in place may be able to aord more takeover provisions without contributing to management entrenchment. For this reason, the hypothesis is that SR rms should have lower GI scores than the non-SR rms, but this dierence may be small. An additional variable is collected for a portion of the total sample from the KLD Socrates Database. This database compiles a continuous scoring mechanism that rates companies on the various degrees of social responsibility including community, diversity, employee interests, environment, and shareholder interests. 3.2.
METHODOLOGY

Univariate and multivariate analyses are used to test the hypothesis that SR boards are stronger than non-SR boards. I use a two-sample paired t-test and nonparametric paired Wilcoxon rank-sum test for dierence of means in order to compare specic board structure variables between SR and non-designated SR rms in a univariate setting. I use logistic regression to examine the relationship between rm type (either SR or non-SR) and board structure in a multivariate setting. The equation is as follows exb 1 exb where y 1 if the rm is socially responsible, 0 otherwise, and X is the vector of the sixteen governance structure variables described in the Section 3.1. and b is the vector of parameters plus an intercept term. Also, by using a continuous measure of social responsibility I am able to analyze the eect of board structure. To examine whether or not board characteristics are related to social responsibility, I report a multiple regression model with the KLD score as the dependent variable and the board data as independent variables. proby 1

4. Results 4.1.
UNIVARIATE STATISTICS

The results for the two-sample paired t-tests and Wilcoxon rank sum tests are documented in Table III. Both tests provide similar results. There is a statistically signicant dierence between SR and non-SR rms for ten of the seventeen governance structure variables (including GI score). For nine of the ten signicant board structure variables, the hypothesized relationships between governance variables of SR and non-SR rms are supported.

AN EXAMINATION OF SOCIALLY RESPONSIBLE FIRMS

265

However, the GI score and board size have the opposite sign as to what was expected, supporting evidence suggesting that takeover defenses and board structure are substitute methods of eective governance.9 Perhaps the most important governance variables analyzed here are the percent of inside and outside directors on the board. With a greater proportion of outsiders, a board is more independent and has more eective monitoring power (Fama (1980)). Since the hypothesis stated that socially responsible rms should have stronger, more eective governance structures than the matched sample, then more outsiders and fewer insiders on SR boards would be supportive of this proposition. As indicated in Table III, the dierence between the percentages of insiders and outsiders on SR and nonSR boards of directors is statistically signicant. SR rms tend to have boards with fewer insiders (23%) and more outsiders (71%) than the matched sample of nondesignated SR rms (31% and 61%, respectively). This provides support for the hypothesis that SR rms have stronger boards than non-SR rms. In addition, gray directors are seen as a hindrance to board independence. The results show that SR rms are less likely to have gray directors on the board than non-SR rms (5% and 7% respectively). Another indication of board independence and eectiveness as a monitor is the level of diversity. Carter et al. (2003) nd that diversity increases board eectiveness, and subsequently increases shareholder value. The measure for board diversity in the present study is the percentage of women on the board of directors. The results indicate that socially responsible rms have a signicantly larger percentage of women on the board (13%) than do the boards of non-SR rms (8%). While the majority of CEOs are also the board chairman for both samples, it appears that SR rms are less likely than non-SR rms to have CEO/Chair duality (72% vs 78%). As indicated in Jensen (1993), having separation between the CEO and chairman of the board creates independence and increases eectiveness of the board, which reduces agency problems between shareholders and managers. As anticipated, socially responsible rms tend to be managed by professional managers rather than by founding family members. CEOs of non-SR rms are almost twice as likely to be founding family members as CEOs of SR rms (15% vs 8%, respectively). Burkart et al. (2002) nd that family-managed rms tend to have lower returns on sales and assets than professionallymanaged rms, and family CEOs are promoted to the post an average of 9 years earlier than professional managers (which is likely to be detrimental to rm performance). Their results suggest that rms run by founding family members tend to perform worse than rms managed otherwise. The results for the fraction of busy directors and outside directors who are also CEOs of other corporations provide noteworthy insight. SR rms have more directors who are also directors on three or more other boards

266
Table III. Dierence in means tests and descriptive statistics Variable SR Mean (Std Error) Non-SR Mean (Std Error)

ELIZABETH WEBB

Dierence Paired Wilcoxon Full sample (SRNon-SR) t-statistic rank sum mean (median) Z 4.41*** 0.61 )2.07** )1.58 0.00 0.10 )0.42 0.38 )3.69*** )5.46*** 8.72*** )2.53** 0.95 2.35** 1.92* 8.39*** 2.44** 4.62*** 1.21 )1.98** )1.70 )0.03 )0.06 0.03 0.36 )3.50*** )4.76*** 8.26*** )1.49* 0.44 2.32** 1.52* 7.65*** 10.00 7.41 0.75 0.60 0.05 0.17 55.73 0.31 0.12 0.27 0.66 0.06 0.10 0.17 0.17 0.10 9.45 (10.00) (7.00) (1.00) (1.00) (0.00) (0.00) (55.50) (0.00) (0.00) (0.22) (0.69) (0.00) (0.08) (0.13) (0.15) (0.10) (10.00)

Members 10.43 (3.02) 9.56 (3.30) 0.87 Meetings 7.47 (3.16) 7.34 (3.31) 0.13 Duality 0.72 (0.45) 0.78 (0.41) )0.06 Terms 0.57 (0.49) 0.63 (0.48) )0.06 Block 0.05 (0.22) 0.05 (0.22) 0.00 Family 0.17 (0.37) 0.17 (0.37) 0.00 Age 55.63 (7.48) 55.83 (8.08) )0.20 NP 0.32 (0.47) 0.31 (0.46) 0.01 FF 0.08 (0.26) 0.15 (0.36) )0.07 Pinside 0.23 (0.14) 0.31 (0.16) )0.08 Poutside 0.71 (0.16) 0.61 (0.17) 0.10 Pgray 0.05 (0.08) 0.07 (0.11) )0.02 Psenior 0.10 (0.13) 0.10 (0.12) 0.00 Pbusy 0.18 (0.17) 0.16 (0.16) 0.02 Pceos 0.18 (0.15) 0.16 (0.14) 0.02 Pfemale 0.13 (0.09) 0.08 (0.08) 0.05 GI 9.68 (2.60) 9.40 (2.76) 0.28

***, **, * indicates signicance at the .01, .05, and .10 level, respectively. Sample consists of 394 rms listed on the Domini 400 Index of SR rms in the year 20002001 and 394 rms matched on industry and size (Non-SR). The number of board meetings is represented by Meetings. Duality is a dummy variable taking the value of 1 if the chairman of the board is also the CEO of the company. Term is a binary variable equal to 1 if there is staggered election of board members, or 0 if members are elected annually. Block is a dummy variable equal to 1 if an outside director holds more than 5% of the outstanding stock of the rm. Family is a dummy variable equal to 1 if there are related directors on the board. Age indicates the age of the CEO. NP is a dummy variable equal to 1 if a director is an ocer of or works with a non-prot organization. FF is a dummy variable equal to 1 if the CEO belongs to the rms founding family. Pinside represents the fraction of the total board members who are (or have been) also ocers of the company. Poutside indicates the fraction of the board that is not aliated with the company. Pgray are the percentage of directors who have substantial business relationships with the company, yet are not insiders. Psenior includes any director that is over the age of 69. Pbusy indicates the percentage of directors who are on three or more other boards (six or more if board member is retired). Pceos represents the percentage of outside directors who are also CEOs of other rms. Pfemales represents the proportion of women on the board. GI is the Governance Index score as measured by Gompers et al. (2003). A negative t-statistic (or Wilcoxon Z-statistic) indicates that the non-SR variable is less than the SR variable, and a positive t-statistic indicates SR is less than non-SR.

AN EXAMINATION OF SOCIALLY RESPONSIBLE FIRMS

267

(busy) and they have more directors who are CEOs than do non-SR rms. Eighteen percent of SR directors are classied as busy, compared to 16% of non-SR directors. CEOs of other rms make up 18% of the directors on SR boards and 16% of non-SR directors. The expertise of these directors may be valuable and sought after by numerous rms, and thus they become busier than other directors. In addition, there is a signicant dierence between the numbers of board members on SR vs non-SR boards. Yermack (1996) nds that a more eective board (as measured by Tobins Q) is comprised of a smaller number of directors. Here, it appears that SR rms have, on average, one more director on the board than do the matched rms (10.43 and 9.56, respectively). However, Hermalin and Weisbach (2003) report that small boards may not be optimal for all rms. It is interesting to note that the presence of board members that work with nonprot organizations is not signicantly dierent between the two samples, especially since it is highly correlated with the percentage of women on the board, and SR boards have signicantly more women than do non-SR boards. Also, there does not appear to be a dierence between the percentage of senior board members on SR and non-SR boards. But since the age of the CEO on both boards does not dier, this may be an explanation since the two variables (proportion of seniors and age) are positively correlated. In addition, the presence of a blockholder on the board is not dependent on rm type. Since the sign on this variable was ambiguous given the prior literature, it may not be related to the reduction of agency problems found with other governance structure variables. In the presence of strong board characteristics, takeover defenses can be associated less with management entrenchment since shareholders are protected by other governance mechanisms. Therefore, SR rms can afford to adopt such provisions without hurting shareholder wealth. SR rms have an average GI score of 9.68 while non-SR rms have an average score of 9.10. The small numerical dierence between the two samples GI scores represents the fact that, on average, non-SR rms have a slightly smaller number of antitakeover amendments than their SR counterparts. The small dierence is consistent with the theory that board structure and takeover defense mechanisms are substitute methods of eective governance. Taken together, the results of the dierence in means tests support the hypothesis that SR rms have more eective governance structures in place than their non-SR counterparts. In the next section, a probability model is used to specify the exact functional relationship between board characteristics and the likelihood that the rm is SR while controlling for rm size.

268 4.2.
MULTIVARIATE STATISTICS

ELIZABETH WEBB

4.2.1. Logistic regression I report the results of the logistic regression equation examining board structure and rm type in Table IV. The dependent variable, rm type, indicates the probability that the rm is a SR company. The independent variables are the sixteen governance structure variables. The coecients indicate the signs of the partial eects of each independent variable on the response probability (that the rm is SR). The logistic regression results indicate that eight of the sixteen governance structure variables are signicant in the model. Many of the results are in agreement with the univariate t-tests. Model 1 in Table IV shows the logistic regression with all of the governance characteristics as independent variables, excluding the percent of outside directors. Here the ndings indicate that when more women are present on the board, it is signicantly more likely that the board is from a SR rm. Also, rms classied as non-SR are more likely to have a high percentage of insiders and grays on the board. In addition, SR boards are more likely to have a CEO who is not also the chairman of the board, as indicated by the negative coecient on the duality variable. Surprisingly, board members involved with non-prot organizations are more likely to represent a rm from the matched sample. However, Weber (2002) notes that a conict of interest exists when outside directors are associated with non-prot organizations that are substantially funded by the company. This ambivalence may alter the predicted relationship between non-prot directors and corporate social responsibility. In addition, boards elected on an annual basis rather than by staggered elections are more likely to be the boards of SR rms than of the matched sample. The presence of family members on the board of directors indicates that the rm is SR. But for CEOs who are part of the founding family, there is less of a chance that these CEOs come from SR rms. As a check for robustness, additional variables are added to the model. As mentioned earlier, the SR rms in the sample are larger on average than the matched rms. Since rm size may be a confounding factor in the logistic regression analysis, three control variables are included in Model 3: b, the natural logarithm of market value of equity, and the natural log of total assets. b is used to capture any risk-induced bias that may confound the results. Market value of equity and total assets are proxies for size. Model 1 shows that the addition of these variables has little eect on the signicant predictors of SR rms. The log of rm market value is positive and signicant in the model, which is consistent with the univariate results in Table I.

AN EXAMINATION OF SOCIALLY RESPONSIBLE FIRMS

269

Table IV. Logistic regressions

Independent variable Intercept Members Meetings Duality Terms Block Family Age NP FF Pinside Poutside Pgray Psenior Pbusy Pceos Pfemale Log market value Beta Log total assets GI Present concordant Likelihood ratio N Expected sign ) ) ) ) + ) ) + ) ) + ) ) + + + Model 1 )1.34 (2.48) )0.015 (0.18) )0.01 (0.02) )0.41 (3.42)* )0.39 (4.61)** )0.02 (0.01) 0.29 (1.29)* )0.00 (0.04) )0.45 (4.84)** )0.63 (4.54)** )1.92 (8.06)*** )1.73 (3.13)* 0.87 (1.42) 0.03 (0.00) )0.48 (0.51) 5.91 (28.90)*** 0.32 (11.92)*** )0.07 (0.11) 0.02 (0.05) 73.5 123.79*** 656 Model 2 )2.24 (29.78)*** 0.03 (1.27) )0.54 (8.69)*** )0.46 (8.16)*** Model 3 )1.80 (29.78)* 0.01 (0.07) 0.00 (0.02) )0.55 (5.04)** )0.68 (9.77)*** )0.08 (0.03) 0.29 (0.96) )0.00 (0.02) )0.40 (3.51)* 0.01 (0.00) 2.59 (14.60)*** 0.15 (0.04) 0.60 (0.82) )1.07 (2.16) 5.49 (20.15)***

)0.36 (1.92) 3.30 (38.05)***

)0.11 (0.05) )0.46 (0.61) 5.83 (36.67)***

76.0 130.03*** 785

0.09 (5.14)** 70.9 70.92*** 535

***, **, * indicates signicance at the .01, .05, and .10 level, respectively. Sample consists of 394 rms listed on the Domini 400 Index of SR rms in the year 20002001 and 394 rms matched on industry and size (NSR). Dependent variable is rm type, a binary variable equal to 1 if rm is from socially responsible sample, or 0 otherwise. Equation analyzed is: prob(y = 1) = p = exb =1 eX b , where X is the vector of sixteen governance structure variables as follows. The number of board meetings is represented by Meetings. Duality is a dummy variable taking the value of 1 if the chairman of the board is also the CEO of the company. Term is a binary variable equal to 1 if there is staggered election of board members, or 0 if members are elected annually. Block is a dummy variable equal to 1 if an outside director holds more than 5% of the outstanding stock of the rm. Family is a dummy variable equal to 1 if there are related directors on the board. Age indicates the age of the CEO. NP is a dummy variable equal to 1 if a director is an ocer of or works with a non-prot organization. FF is a dummy variable equal to 1 if the CEO belongs to the rms founding family. Number in parentheses is chi-square test statistic. Pinside represents the fraction of the total board members who are (or have been) also ocers of the company. Poutside indicates the fraction of the board that is not aliated with the company. Pgray are the percentage of directors who have substantial business relationships with the company, yet are not insiders. Psenior includes any director that is over the age of 69. Pbusy indicates the percentage of directors who are on three or more other boards (six or more if board member is retired). Pceos represents the percentage of outside directors who are also CEOs of other rms. Pfemales represents the proportion of women on the board. GI is the Governance Index score as measured by Gompers et al. (2003).

270

ELIZABETH WEBB

Model 2 in Table IV includes only those independent variables that were signicant in the univariate tests to explain dierences in SR and non-SR boards. Here, only the proportion of outsiders is used to avoid any multicollinearity issues with the logistic regression analysis. The results are again identical to Model 1, except that the founding family variable is no longer signicant in the model. In Model 3 of Table IV, I include the GI variable in the logistic regressions. Since sample size is signicantly reduced when using the GI score (325 SR rms and 210 non-SR rms), I do not include it in the previous models. The results of this regression are consistent with the prior results. CEO/Chair duality, election terms, non-prot aliation, proportion of outsiders and proportion of women on the board are signicant in the model. In addition, the GI score is positive and signicant, indicating that SR rms are have higher GI scores than non-SR rms, which is consistent with results from univariate tests reported in Table III. I also run regressions using the control variables and the signicant variables from Model 3 (results not reported). Doing so does not change the results. 4.2.2. Continuous measure of corporate social responsibility Next, I analyze the eect of board structure on social responsibility. Identical independent variables are used in Table V as in Model 2 in Table IV. In this case, I use the KLD Socrates social responsibility score as a continuous measure of social responsibility. A higher-score indicates a higher-degree of corporate social responsibility. Three separate regressions are run for the full sample, SR rms, and matched rms, respectively. As reported in Table V, several board characteristics for the full sample signicantly explain variation in the KLD score. The number of board members is directly related to the social responsibility score, which supports the original hypothesis according to Hermalin and Weisbach (2003) and results found in the univariate tests in Table III. Also, when boards are elected on a staggered basis, the KLD score decreases. This supports the initial hypothesis that boards elected annually should be linked with stronger board structures and subsequently with SR rms boards. The proportions of outsiders and women on the board are directly related to the KLD score, which also corroborates initial hypotheses. Control variables include log of market value of equity, beta, and log of total assets.10 Total assets, interestingly, is negatively related to the KLD score. This informally supports the idea that size is not an overall determinant of whether or not the rm is from the SR sample. This issue will be addressed further in the next section. Next, separate regressions are run using the SR and non-SR samples. Interestingly, the matched rm regressions show no statistically signicant

AN EXAMINATION OF SOCIALLY RESPONSIBLE FIRMS

271

Table V. Regressions of KLD score on board characteristics Independent variable Sample All rms Intercept Members Duality Terms FF Poutside Pbusy Pceos Pfemale Log market value Beta Log total assets Adj. R2 F N )0.016 ()0.02) 0.091 (1.86)* )0.358 ()1.28) )0.424 ()1.73)* )0.019 ()0.04) 1.549 (1.82)* )0.093 ()0.12) )1.270 ()1.48) 7.742 (6.07)*** 0.148 (1.22) 0.112 (0.41) )0.351 ()2.90)*** 9.94% 5.06*** 405 SR rms )0.137 ()0.16) 0.022 (0.42) )0.231 ()0.84) )0.328 ()1.31) )0.269 ()0.58) 1.720 (1.92)* )0.771 ()1.01) )1.193 ()1.37) 6.427 (5.10)*** 0.220 (1.76)* 0.003 (0.01) )0.262 ()2.01)** 7.89% 3.52*** 324 Non-SR rms Dierence )1.469 ()0.54) 0.156 (1.42) )0.384 ()0.42) 0.411 (0.60) 0.774 (0.74) )2.215 ()1.01) 1.373 (0.61) )0.233 ()0.09) 5.502 (1.29) 0.194 (0.58) 0.012 (0.02) )0.273 ()0.98) )2.81% 0.80 80 )0.134 ()1.23) 0.153 (0.18) )0.740 ()1.16) )1.043 ()1.03) 3.935 (1.89)* )2.143 ()1.04) )0.960 ()0.41) 0.925 (0.24) 0.026 (0.08) )0.009 ()0.01) 0.011 (0.04) 18.19% 4.92*** 405

***, **, * indicates signicance at the .01, .05, and .10 level, respectively. Sample consists of rms listed on the Domini 400 Index of SR rms in the year 20002001 and rms matched on industry and size (Non-SR). Dependent variable is the total social responsibility score given by KLD. Members indicate the number of board members. Duality is a dummy variable taking the value of 1 if the chairman of the board is also the CEO of the company. Term is a binary variable equal to 1 if there is staggered election of board members, or 0 if members are elected annually. FF is a dummy variable equal to 1 if the CEO belongs to the rms founding family. Poutside indicates the fraction of the board that is not aliated with the company. Pbusy indicates the percentage of directors who are on three or more other boards (six or more if board member is retired). Pceos represents the percentage of outside directors who are also CEOs of other rms. Pfemales represents the proportion of women on the board. T-value in parentheses.

relationships between board characteristics and the KLD score. Regressions for the SR sample show some signicant characteristics, including proportion of outsiders and women on the board, which are positively related to the KLD score. However, dierences between coecients from the SR and non-SR samples are not statistically signicant (except in the case of the proportion of outsiders on the board, where the dierence is signicant at the 10% level). I also include the GI score as an independent variable in the regression (results not reported). Using all the rms in the sample with GI scores, results are consistent with prior models. Coecients on board size and proportion of women on the board are positive and signicant at the 0.05 and 0.01 levels, respectively. The coecient on GI is not signicant, however. This indicates

272

ELIZABETH WEBB

that the GI score is not related to the KLD score after controlling for board and size factors.

4.3.

ROBUSTNESS CHECKS

4.3.1. Reduced samples To test whether dierences in board characteristics are driving the results, I repeat the paired t-tests from Table III using samples that are split between the above-median and below-median values for board size, proportion of insiders, proportion of outsiders, proportion of busy directors, proportion of CEOs on the board, and proportion of women on the board. I evaluate these variables and their inuence on the other board structure variables further since they are found to be signicantly dierent between SR and non-SR rms in the original tests. This process will help to identify the most important structural dierences between SR and non-SR boards by controlling for dierences in certain governance structure variables. For both high and low values for each structural variable, there continues to be a statistically signicant relationship for both the proportion of outsiders and proportion of women on the board. Board size is signicantly dierent in all models except when using samples divided on abovemedian proportion of insiders and above-median proportion of women on the board. Duality, founding family CEO, proportion of insiders, and proportion of grays are also signicantly dierent between the two samples for most of the subdivided rms. Additionally, process of director elections, proportion of senior directors and proportion of busy directors are signicantly dierent in several samples, but not the majority of the reduced samples. In conclusion, the univariate dierences between SR and non-SR boards are robust to sample revisions based on high and low values of certain board characteristics. Particularly, the proportion of outsiders and women on the board, board size, duality, founding family CEOs, proportion of insiders and proportion of grays are signicantly dierent between SR and non-SR boards even after controlling for dierences in specic board characteristics. 4.3.2. Potential sample bias One criticism of using the DSI as the sample of socially responsible rms in this study is that proxy voting guidelines screen for good governance characteristics. In order to arm that these results are not simply due to a sample bias, I reduce the sample to those DSI rms that are cross-listed on the Calvert Social Index, which is another broad-based SR mutual fund

AN EXAMINATION OF SOCIALLY RESPONSIBLE FIRMS

273

that does not list governance characteristics in proxy voting guidelines. This leaves 181 SR rms and 181 matched rms. The rst column of Table VI shows the paired t-test results between these reduced samples. The Calvert rms exhibit similar board characteristics as the Domini rms indicating that these boards are more independent than the matched sample boards. Specically, these rms have more outside directors (and subsequently less insiders and grays), more women on the board, less instances that the CEO is also the chairman of the board, more frequent use of annual elections of directors, and more CEOs on the board than the matched sample. Dierences from the Domini sample include that meetings and proportion of senior directors have dierent signs (but in both tests, t-values are not signicant), and that proportion of busy directors is no longer signicant in the model. However, the strongest board characteristics that lead to independent and strong boards of directors are identical in both SR samples, therefore eliminating the possibility of a sample bias. This supports the previous verication of the univariate tests, which showed that certain board characteristic dierences between the two samples are robust to reducing the sample by separating between small and large values of signicant characteristics from Table III. Particularly, the proportion of outsiders and women on the board, board size, duality, founding family CEOs, proportion of insiders and proportion of grays are signicantly dierent between SR and non-SR boards, which conrm the results in Table VI. In addition, to arm that the results are based on a SR eect rather than a size eect, I reduce the sample to include only those SR rms whose matched rms are within at least 20% of the equity market value of the DSI rm. This generates a sample size of 101 SR rms and 101 matched rms. The univariate results using this sample are similar to those reported in Table III and are shown in the second column of Table VI. Again, the SR rms have more outsiders, less insiders, more women, lower instance of the CEO being from the founding family, and a greater instance of the CEO not also holding the position of chairman of the board. Less inuential characteristics dier somewhat with the full sample: the number of board members and the proportion of gray directors are not signicant, and the director election process (terms) is not signicant in the model (although this was only marginally signicant in the full sample). I also test for dierences in the GI score with the reduced samples. This reduces the samples further, and results are not reported in the table. The paired dierence t-tests between GI scores for SR and non-SR rms using both the Calvert cross-listed reduced sample and the size dierence restriction sample are consistent with Table III. Both are signicantly dierent at the 0.05 level. In summary of the reduced sample tests, characteristics that are the strongest indicators of board strength and independence are consistently

274
Table VI. Tests for Dierence in means with reduced samples Independent variable Sample Calvert cross-listed rms Members Meetings Duality Terms Block Family Age NP FF Pinside Poutside Pgray Psenior Pbusy Pceos Pfemale N
a

ELIZABETH WEBB

MV within 20% )0.960 (0.339) 0.140 (0.889)a )2.387 (0.019)a 0.705 (0.482) 0.962 (0.338)a 0.847 (0.400)a )0.960 (0.338)a )0.847 (0.399) )2.594 (0.011)a )2.196 (0.030)a 2.560 (0.011)a )1.231 (0.221) 1.282 (0.203)a )1.113 (0.268) 0.601 (0.549) 4.543 (0.000)a 202

4.195 (0.000) )1.614 (0.108) )1.760 (0.080)a )2.977 (0.030)a )0.229 (0.819) 0.744 (0.458)a )0.279 (0.781)a 0.639 (0.524)a )2.372 (0.190) )3.922 (0.000)a 7.436 (0.000)a )3.270 (0.001)a )0.413 (0.681) 2.150 (0.330) 2.965 (0.030)a 6.858 (0.000)a 362

indicates that the t-test has the same sign and signicance as the full sample results from Table III. The full sample consists of 394 rms listed on the Domini 400 Index of SR rms in the year 20002001 and 394 rms matched on industry and size (NSR). This is then reduced to rms cross-listed on the Calvert Social Index Fund (181 pairs) and matched rms within 20% of each others market value (101 pairs). The number of board meetings is represented by Meetings. Duality is a dummy variable taking the value of 1 if the chairman of the board is also the CEO of the company. Term is a binary variable equal to 1 if there is staggered election of board members, or 0 if members are elected annually. Block is a dummy variable equal to 1 if an outside director holds more than 5% of the outstanding stock of the rm. Family is a dummy variable equal to 1 if there are related directors on the board. Age indicates the age of the CEO. NP is a dummy variable equal to 1 if a director is an ocer of or works with a non-prot organization. FF is a dummy variable equal to 1 if the CEO belongs to the rms founding family. Pinside represents the fraction of the total board members who are (or have been) also ocers of the company. Poutside indicates the fraction of the board that is not aliated with the company. Pgray are the percentage of directors who have substantial business relationships with the company, yet are not insiders. Psenior includes any director that is over the age of 69. Pbusy indicates the percentage of directors who are on three or more other boards (six or more if board member is retired). Pceos represents the percentage of outside directors who are also CEOs of other rms. Pfemales represents the proportion of women on the board. T-values are reported with p-values in parentheses below.

AN EXAMINATION OF SOCIALLY RESPONSIBLE FIRMS

275

signicant in the SR sample, while some of the more ambiguous board strength characteristics (such as proportion of busy and gray directors) may dier when using dierent samples. This result gives further insight as to the central characteristics of an eective board of directors. 5. Conclusion The crisis in corporate governance has created a sense of distrust among investors in the US. In growing numbers, investors are apparently considering more than the rms nancials when buying stock. A companys commitment to social responsibility and monitoring on behalf of the stakeholders are becoming important issues in investment. This study analyzes distinctions in the board structures of SR rms. Results support the hypothesis that SR rms have more eective boards of directors than their non-SR counterparts. The characteristics of SR boards may contribute to the continued investor interest in these types of rms, for it appears that stronger boards exist in rms that emphasize stakeholder utility maximization rather than a strict adherence to shareholder wealth maximization strategies. These results add to the growing body of empirical literature on corporate governance and highlight several important characteristics of eective board of directors. An important indication of board of directors independence and eectiveness is the composition of the board. SR boards have a signicantly higher percentage of outside directors than the matched sample. Likewise, non-SR rms have more insiders and more gray directors on the board than SR rms. There are typically more women on SR boards, indicating that SR boards are more diverse than non-SR boards. Directors on SR boards tend to be busier, in that they serve on other boards and are often CEOs of other rms, but have fewer meetings than non-SR directors. SR boards have more directors on average than non-SR boards. The CEO is less likely to be from the founding family on an SR board, and is less likely to be the rms chairman of the board of directors than CEOs from non-SR boards. It is not clear from these results, however, that being SR is a precursor to having a strong corporate governance structure. If stakeholder consideration is rewarded by increased investor interest, and Tiroles denition of corporate governance as a monitor of managers on behalf of stakeholders is accepted, it seems reasonable to hypothesize that SR rms should have strong governance structures in place. Therefore, future research on the causality between corporate social responsibility and governance structure using a time series dataset would be a worthwhile contribution to the governance literature.

276 Notes

ELIZABETH WEBB

I would like to thank Jacqueline Garner, Michael Gombola, Thomas McWilliams, Gerard Olson, and participants at the 2003 Eastern Finance Association conference for helpful comments and assistance. I am especially grateful to Edward Nelling for his guidance and support. All remaining errors are my own. 2 Bernhut (2002) describes how there is no conclusive evidence on a correlation between corporate social responsibility and stock price, but that subtle advantages, including greater loyalty and commitment from stakeholders, are the main benets of being socially responsible. 3 This may be due to increased regulation in the banking industry. See, for example, Gillan et al. (2002), and Adams and Mehran (2003). 4 Kinder, Lydenberg, Domini & Co. is an agency that reports company proles based on dierent aspects of social responsibility including charitable giving, community involvement, diversity, employee welfare, and the natural environment. 5 The full sample of the socially responsible rms and their respective matched rms is available upon request. 6 In some cases, proxy statements from 1999, 2000, or 2002 were used depending on availability. Proxy statements from 2001 were used for 90% of the SR rms and 96.6% of non-SR rms. 7 Most often, a lack of a proxy statement was due to mergers or bankruptcies. 8 VIF statistics are less than 10. The variance ination factor (VIF) is equal to 1=1 R2 , i where R2 is the coecient of multiple determination between variables. i 9 Eliminating the largest rms (top 1%) and smallest rms (bottom 1%) from the sample yields a slight change in coecients but identical signicance levels and coecient signs as reported in Table III. Thus, outliers are not inuencing the univariate results. 10 In the event that total assets and market value are correlated, I also run the regressions omitting market value from the set of independent variables. Results are consistent with those reported in Table VII, although the p-value for the coecient on the proportion of outsiders variable using the full sample increases to 0.101 making this variable only marginally signicant.

References
Adams, R. and H. Mehran: 2003, Is Corporate Governance Dierent for Bank Holding Companies?, Economic Policy Review 9: 123142. Bacon, C., M. Cornett and W. Davidson: 1997, The Board of Directors and Dual-Class Recapitalizations, Financial Management 26: 523. Beasley, M.: 1996, Board of Director Composition and Financial Statement Fraud, The Accounting Review 71: 443466. Beasley, M., J. Carcello, D. Hermanson and P. Lapides: 2000, Fraudulent Financial Reporting: consideration of Industry Traits and Corporate Governance Mechanisms, Accounting Horizons 14: 441454. Bernhut, S.: 2002, Corporate Social Responsibility, Ivey Business Journal, March/April, 18 20. Burkart, M., F. Panunzi and A. Shleifer: 2002, Family Firms, Working Paper, NBER series. Carroll, A.: 1991, CSP measurement: A Commentary on Methods for Evaluating an Illusive Construct, Research in Corporate Social Performance and Policy 12. Carter, D., B. Simkins and W. Simpson: 2003, Corporate governance, board diversity, and rm value, Financial Review 38: 3353. Core, J., R. Holthausen and D. Larcker: 1999, Corporate governance, chief executive ocer compensation, and rm performance, Journal of Financial Economics 51: 371406. Fama, E.: 1980, Agency problem and the theory of the rm, Journal of Political Economy 88: 288.

AN EXAMINATION OF SOCIALLY RESPONSIBLE FIRMS

277

Fama, E. and K. French: 1997, Industry Costs of Equity, Journal of Financial Economics 43: 153194. Fama, E. and M. Jensen: 1983, Separation of Ownership and Control, Journal of Law and Economics 26: 301325. Ferris, S., M. Jagannathan and A. Pritchard: 2003, Too busy to mind the business? Monitoring by directors with multiple board appointments, Journal of Finance, forthcoming. Friedman, M.: 1970, The Social Responsibility of Business is to Increase its Prots, The New York Times Magazine, September 13. Gillan, S., J. Hartzell and L. Starks: 2003, Industries, investment opportunities, and corporate governance structures, Working Paper, SSRN series, University of Delaware. Gompers, P., J. Ishii and A. Metrick: 2003, Corporate Governance and Equity Prices, Quarterly Journal of Economics 118: 107155. Hermalin, B. and M. Weisbach: 2003, Board of Directors as an Endogenously Determined Institution: A Survey of the Economic Literature, Economic Policy Review, forthcoming. Hillman, A. and G. Keim: 2001, Shareholder value, Stakeholder Management, and Social Issues: Whats the bottom line? Strategic Management Journal 22: 125139. Howton, S., S. Howton and G. Olson: 2001, Board Ownership and IPO Returns, Journal of Economics and Finance 25: 100112. Jayaraman, N., A. Khorana, E. Nelling and J. Covin: 2000, CEO Founder Status and Firm Financial Performance, Strategic Management Journal 21: 12151224. Jensen, M.: 1993, The Modern Industrial Revolution, Exit and the Failure of Internal Control Systems, Journal of Finance 48: 831880. Jensen, M. and W. Meckling: 1976, Theory of the rm: Managerial Behavior, agency costs and ownership structure, Journal of Financial Economics 34: 305360. John, K. and L. Senbet: 1998, Corporate Governance and Board Eectiveness, Journal of Banking and Finance 22: 371403. Johnson, O.: 1966, Corporate Philanthropy: An Analysis of Corporate Contributions, Journal of Business 39: 489504. Malekzadeh, A., V. McWilliams and N. Sen: 1998, Implications of CEO Structural and Ownership Powers, Board Ownership and Composition on the Markets Reaction to Antitakeover Charter Amendments, Journal of Applied Business Research 14: 5362. McGuire, J., A. Sundgren and T. Schneeweis: 1988, Corporate Social Responsibility and Firm Financial Performance, Academy of Management Journal 31: 854872. McWilliams, V. and N. Sen: 1997, Board Monitoring and Antitakeover Amendments, Journal of Financial and Quantitative Analysis 32: 491505. Rosenstein, S. and J. Wyatt: 1997, Inside Directors, Board Eectiveness, and Shareholder Wealth, Journal of Financial Economics 44: 229250. Shivdasani, A.: 1993, Board Composition, Ownership Structure, and Hostile Takeovers, Journal of Accounting and Economics 16: 167199. Teoh, S., I. Welch and C. Wazzan: 1999, The Eect of Socially Activist Investment Policies on the Financial Markets: Evidence From the South African boycott, Journal of Business 72: 3589. Tirole, J.: 2001, Corporate Governance, Econometrica 69: 135. Waddock, S. and S. Graves: 1997, The Corporate Social Performance-Financial Performance Link, Strategic Management Journal 18: 303319. Weber, J.: 2002, Boardroom Charity: Reforms Dont go far Enough, Business Week, June 10: 28. Yermack, D.: 1996, Higher Market Valuation of Companies with a Small Board of Directors, Journal of Financial Economics 40: 185211.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.