Corporate Governance: US Board Diversity & Financial Performance (2010) PDF

396 Corporate Governance: An International Review, 2010, 18(5): 396–414 The Gender and Ethnic Diversity of US Boards and Board Committees and Firm Financial Performance corg_809 396..414 David A. Carter, Frank D’Souza, Betty J. Simkins, and...

396 Corporate Governance: An International Review, 2010, 18(5): 396–414 The Gender and Ethnic Diversity of US Boards and Board Committees and Firm Financial Performance corg_809 396..414 David A. Carter, Frank D’Souza, Betty J. Simkins, and W. Gary Simpson* ABSTRACT Manuscript Type: Empirical Research Question/Issue: We examine the business case for the inclusion of women and ethnic minority directors on the board. Specifically, we investigate the relationship between the number of women directors and the number of ethnic minority directors on the board and important board committees and financial performance measured as return on assets and Tobin’s Q. Research Findings/Insights: We do not find a significant relationship between the gender or ethnic diversity of the board, or important board committees, and financial performance for a sample of major US corporations. Our evidence also suggests that the gender and ethnic minority diversity of the board and firm financial performance appear to be endogenous. Theoretical/Academic Implications: Reasonable theoretical arguments drawn from resource dependence theory, human capital theory, agency theory, and social psychology suggest that gender and ethnic diversity may have either a positive, negative, or neutral effect on the financial performance of the firm. Our statistical analysis supports the theoretical position of no effect, either positive or negative. Our results are consistent with a contingency explanation because the effect of the gender and ethnic diversity of the board may be different under different circumstances at different times. Over several companies and time periods, the results could offset to produce no effect. Practitioner/Policy Implications: The results of our analysis do not support the business case for inclusion of women and ethnic minorities on corporate boards. However, we find no evidence of any negative effect either. Our evidence implies that decisions concerning the appointment of women and ethnic minorities to corporate boards should be based on criteria other than future financial performance. Keywords: Corporate Governance, Financial Performance, Board Committees, Board Composition INTRODUCTION Minow, 2004). The movement in the US for improved corpo- rate governance following governance failures and a height- E volving cultural, political, and societal views of corpo- rate board membership are partially driving interest in the demographic diversity of corporate directors. In addi- ened awareness of the importance of corporate governance produced the Sarbanes-Oxley Act of 2002, a massive piece of legislation. Other countries have passed legislation and/or tion, the global desire for better corporate governance is a guidelines regulating corporate governance as well. Rose major factor. The Cadbury Report in the United Kingdom, (2007) reports a significant interest in Scandinavian coun- the General Motors Board of Directors guidelines in the US, tries in increasing the number of women on corporate and the Dey Report in Canada illustrate an interest in boards. Norway has a law that requires 40 per cent of the improved governance in different countries (Monks & directors for a company to be women (Rose, 2007). Similar to Norway, Spain recently passed legislation requiring a quota *Address for correspondence: Spears School of Business, Oklahoma State University, for the number of female directors (Adams & Ferreira, 2009). Stillwater, OK 74078-4011, USA. E-mail: [email protected] The Higgs Report, commissioned by the British Department © 2010 Blackwell Publishing Ltd doi:10.1111/j.1467-8683.2010.00809.x BOARD DIVERSITY AND FINANCIAL PERFORMANCE 397 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License of Trade and Industry, suggests that demographic diversity and ethnic minority diversity of the board and firm financial increases board effectiveness and recommends that more performance but the empirical evidence on the link is mixed women be included on boards (Adams & Ferreira, 2009). and limited. Hillman, Cannella, and Harris (2002) contend that one of the The purpose of this empirical analysis is to explore the most important trends in US boardrooms over the past two relationship between the gender and ethnic minority diver- decades is a shift toward the inclusion of women and ethnic sity of the board and the financial performance of the firm. minorities. Our research is unique because we consider both ethnic The phenomenon of the gender and ethnic diversity of diversity and gender diversity in this analysis that is not corporate boards encompasses at least two significant, and common in the literature. We could locate only one other interrelated, propositions. The first viewpoint holds that empirical investigation that directly considered the link those competent women and ethnic minorities with the between the ethnic minority diversity of the board and human capital, external networks, information, and other financial performance of the firm. The empirical relationship characteristics of importance to the corporation deserve between the gender of corporate directors and financial per- opportunities to serve on corporate boards and in upper formance has received much more attention in the literature management. The second proposition suggests that gender than other aspects of the demographic diversity of corporate and ethnic diversity of directors results in better governance directors, possibly because of the availability of data. which causes the business to be more profitable. Karen J. However, we believe that gender diversity and ethnic diver- Curtin, a former executive vice president of Bank of sity are not the same phenomenon and will not affect the America, describes the interaction of the two propositions of firm in identical ways. We base this conjecture on the follow- board diversity in the following statement, “There is real ing: (1) previous empirical evidence on the nature of board debate between those who think we should be more diverse diversity (Hillman et al., 2002; Peterson & Philpot, 2007; because it is the right thing to do and those who think we Peterson, Philpot, & O’Shaughnessy, 2007); (2) theory, espe- should be more diverse because it actually enhances share- cially resource dependence theory and human capital holder value. Unless we get the second point across theory, which suggests significant differences between and people believe it, we’re only going to have tokenism” women directors and ethnic minority directors; and (3) the (Brancato & Patterson, 1999:7). Herman Bulls, CEO of a real evidence presented in this analysis which suggests a differ- estate advisory group and a director for Comfort Systems ence between women directors and ethnic minority direc- USA, states “When I’m sitting in that boardroom, my fidu- tors. Furthermore, our study is unique because we explore ciary responsibility is to the shareholders of that company – the relationship between the gender and ethnicity of the not social engineering. I can talk about diversity. But there members of important board committees and financial per- ought to be a business case” (Dvorak, 2008:R4). The business formance which has not been done before. We investigate case implies that competent women and ethnic minority the hypothesis of Klein (1998) that an analysis of committee directors are not substitutes for traditional corporate direc- membership and financial performance provides a different, tors with identical abilities and talents but qualified women and perhaps stronger, test of the link between board diver- and ethnic minority directors have unique characteristics sity and firm performance. Our third contribution is that we that create additional value.1 The business case for board implement econometric approaches that have not been used gender and ethnic diversity is a subset of the larger issue of extensively in previous investigations. As the empirical good international corporate governance. analysis of board diversity and firm performance has pro- A realistic understanding of the nature of any relationship gressed, more and more sophisticated analytical methods that may exist between the gender and ethnic diversity of are being applied with new data sets. Our research contrib- the board and firm financial performance has important utes to this research stream by using improved statistical implications for both public policy and the governance of tests and variables. All of the previous empirical research we business firms. If there is no difference between competent could find uses the percentage of women and/or percentage women and ethnic minority directors and other qualified of ethnic minorities on the board as the independent variable directors so that the gender and ethnic diversity of the board of interest. Our investigation uses the number of women does not influence firm governance and performance, then directors and the number of ethnic minority directors. We the desirability of gender and ethnic minority diversity is implement a five-year panel of data and associated panel primarily a public policy issue. However, if there is a posi- data methodology. We use a firm fixed effect methodology tive relationship between the gender and ethnic diversity of that has only been used once in previous research. Our sta- the board and firm performance, the economic implications tistical models tested a lagged relationship between diver- of board diversity are important. To the contrary, if the rela- sity measures and financial performance that is rare in tionship is negative, the costs of inclusion of women and previous investigations. ethnic minority directors become a factor to be considered. The paper is organized as follows. We review applicable Contingency theory (Fiedler, 1967; Lawrence & Lorsch, theory in the first section. In the second section, we discuss 1967) suggests a more complex link between the gender and some of the previous evidence on the link between board ethnic diversity of the board and firm performance in that diversity and the financial performance of the firm. We certain aspects of board diversity may be desirable in some present the hypotheses tested in the third section and organizations, and not others, and under different circum- discuss our sample, data, variables, and statistical methods stances at different times.2 Theories from economics, organi- in Section 4. We present the results of the statistical analysis zation behavior, and social psychology provide some in section five and provide concluding comments in the last understanding of the nature of the link between the gender section of the paper. The research is conducted with data © 2010 Blackwell Publishing Ltd Volume 18 Number 5 September 2010 398 CORPORATE GOVERNANCE 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License from the major US corporations listed in the Standard and female African-American directors are less likely to be busi- Poor’s 500 index. ness experts than male African-American directors and that both female and male African-American directors are less likely to be business experts than Caucasian female direc- THEORETICAL PERSPECTIVES ON THE tors. Caucasian male directors are much more likely to be DIVERSITY-PERFORMANCE business experts than either African-American or female RELATIONSHIP directors.5 Gender and ethnicity appear to be separate dimensions under resource dependence theory because The board of directors is generally believed to have at least women and ethnic minorities have different backgrounds four important functions – monitoring and controlling man- and different human capital which results in the ability to agers, providing information and counsel to managers, address different environmental dependencies.6 monitoring compliance with applicable laws and regula- Resource dependence theory provides the basis for some tions, and linking the corporation to the external environ- of the most convincing theoretical arguments for a business ment (Mallin 2004; Monks and Minow 2004). A large body of case for board diversity. Diversity holds the potential to the theory on boards addresses these functions in one way or improve the information provided by the board to managers another. One basic proposition is that the composition of the due to the unique information held by diverse directors. board affects the way the board performs these functions Differences in gender and ethnicity will very likely produce that partially determine firm performance.3 This concept unique information sets that are available to management for offers the possibility that board composition in general, with better decision making. Diverse directors provide access to the gender and ethnic minority diversity of the board a important constituencies in the external environment. The subset of board composition, is linked to firm performance. creation of this important link is crucial because over half of No single theory directly predicts the nature of the rela- the pool of human capital available to the firm is composed tionship between board diversity and financial performance of women and ethnic minorities. As a result, diverse orga- but several theories from various fields provide insight into nizations have access to more talent. Board diversity sends the issue.4 We adopt an interdisciplinary approach and draw important positive signals to the labor market and product from four important theories taken from organization market, although Caucasian women directors send a differ- theory, economics, and social psychology to provide the ent signal to these markets than ethnic minorities. Diverse theoretical basis for the hypotheses tested. directors may bring diverse perspectives and nontraditional approaches to problems as they are less likely to be insiders or business experts. The ability of an ethnically diverse Resource Dependence Theory board to provide legitimacy for the corporation with both Pfeffer and Salancik (1978) argue that boards serve to link external and internal constituencies is particularly important the corporation to other external organizations in order to in countries like the US because of increasing growth in the address environmental dependencies. Pfeffer and Salancik proportion of ethnic minority groups. (1978) suggest four primary benefits for the external link- It should be noted that the type of diversity that will be ages: (1) provision of resources such as information and important in a particular country or culture may vary widely. expertise; (2) creation of channels of communication with We observe that gender diversity is emphasized in Scandi- constituents of importance to the firm; (3) provision of com- navian countries and some other countries in Europe such as mitments of support from important organizations or Spain, possibly because of greater ethnic homogeneity. groups in the external environment; and (4) creation of However, some European countries are experiencing legitimacy for the firm in the external environment. Hillman, increasing ethnic diversity as well. Other types of demo- Cannella, and Paetzold (2000) expand these four benefits graphic diversity, including religion and age, may have into a taxonomy of director types that provide various more importance in different national and cultural settings. resources to the firm: insiders, business experts, support specialists, and community influentials. Hillman et al.’s (2000) extension of resource dependence theory suggests Human Capital Theory that different types of directors will provide different ben- Terjesen, Sealy, and Singh (2009) indicate that human capital eficial resources to the firm. As a result, a more diverse board theory is derived from the work of Becker (1964) that will provide more valuable resources, which should produce addresses the role of a person’s stock of education, experi- better firm performance. ence, and skills that can be used to the benefit of an organi- Furthermore, the type of diversity appears to be impor- zation. Furthermore, differences in gender results in tant. For example, Booth and Deli (1999) find that the pres- directors having unique human capital (Terjesen et al., 2009). ence of a commercial banker on the board is positively If human capital of corporate directors is influenced by related to the total debt of the firm and they conclude that gender, it is reasonable to hypothesize that the human commercial bankers provide expertise on, and links to, the capital of ethnic minorities would be unique relative to both bank debt market. Agrawal and Knoeber (2001) find that Caucasian men and women. Human capital theory comple- outside directors with political and legal backgrounds are ments some of the concepts associated with board diversity more likely to be on the boards of companies that sell to the derived from resource dependence theory. government or face government regulation. Similarly, One question raised by the fact that women and ethnic women directors and ethnic minority directors bring differ- minorities have unique human capital is, “the claim that ent benefits and resources. Hillman et al. (2002) find that women lack the ‘right’ human capital for directorships.” Volume 18 Number 5 September 2010 © 2010 Blackwell Publishing Ltd BOARD DIVERSITY AND FINANCIAL PERFORMANCE 399 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License (Terjesen et al., 2009:325). The evidence on the human capital (Westphal & Milton, 2000). Thus it may be that diverse direc- of women suggests that women are just as well qualified as tors will not influence the board as a result of the internal men in terms of several important qualities including level group dynamics of the board.8 of education but women are less likely to have experience as Some research has suggested that minority group business experts (Terjesen et al., 2009).7 Peterson et al. (2007) members may encourage divergent thinking in the decision- find that African-American directors assume different roles making process (Westphal & Milton, 2000). However, on the board relative to Caucasian directors which is possi- Campbell and Minguez-Vera (2008) draw from the work of bly tied to their unique human capital. This appears to be the Lau and Murnighan (1998) to argue that greater gender same for women directors as well (Hillman et al., 2002; diversity among board members generates more diverse Peterson, Philpot, & O’Shaughnessy 2007). The net result is opinions and critical thinking that makes decision-making that human capital theory predicts that the performance of more time-consuming and less effective. Williams and the board will be affected by board diversity as a result of O’Reilly (1998) conclude that the evidence suggests that diverse and unique human capital but the effect could be diversity may produce more conflict and employee turnover either positive or negative from a financial performance per- as well as creativity and innovation. Forbes and Milliken spective. Contingency theory (Fiedler, 1967; Lawrence & (1999) conclude from a review of the evidence that board Lorsch, 1967) is relevant as well in that human capital that effectiveness probably depends significantly on social- may be useful in one organization at some point in time but psychological processes and they argue that each facet of may not be useful under different internal and external board demography is likely to have many complex and con- circumstances. flicting effects on the processes that affect board perfor- mance. Kim, Burns, and Prescott (2009) argue that board diversity is positively related to the breadth and speed of top Agency Theory management team strategic action capability in their theo- The board function of monitoring and controlling managers retical analysis of the strategic role of the board of directors. is a fundamental concept from agency theory (Jensen & In summary, the theory and evidence on group dynamics Meckling, 1976). Carter, Simkins, and Simpson (2003) suggests that board diversity may have both positive and suggest that a more diverse board may be a better monitor of negative effects on firm performance. managers because board diversity increases board indepen- dence but they go on to say that agency theory does not provide a clear prediction of the link between board diver- EMPIRICAL RESEARCH ON THE sity and financial performance. Diverse directors are less DIVERSITY-PERFORMANCE likely to be beholden to managers according to this view, for RELATIONSHIP example TIAA-CREF adopts this proposition in their policy statements (Carleton, Nelson, & Weisbach, 1998). Further- The link between the diversity of corporate boards and the more, factors such as ownership position in the firm may financial performance of the firm has attracted the attention have a more powerful influence on board monitoring than of scholars around the world. However, we located only nine independence. Jensen (1993) and Monks Minow (2004) empirical studies that specifically tested the link between argue that high equity ownership by directors is a more board diversity and the financial performance of the firm. important factor in increasing the willingness of directors to This is a rather limited amount of evidence given the amount monitor than independence. In general, agency theory does of discussion on the topic and the volume of empirical inves- not provide as strong support for the financial benefits of tigation devoted to other topics of major interest. board diversity as does a resource dependence perspective Seven of these nine investigations examined the link but agency theory does not rule out the possibility that between the gender diversity of the board and firm perfor- board diversity is beneficial. mance. Campbell and Minguez-Vera (2008) investigate the relationship between the gender diversity of the board and financial performance for a sample of companies from Spain. Social Psychological Theory They find that board gender diversity has a positive effect on Westphal and Milton (2000) address the opposing views that firm value as measured by Tobin’s Q. Carter et al. (2003) the presence of demographic minorities on boards is often examine a sample of US firms and find a positive relationship viewed favorably by corporate stakeholders but the aca- between board gender diversity and Tobin’s Q. Adams and demic literature is more pessimistic about the extent to Ferreira (2009) find that more gender diverse boards devote which demographic minority directors can successfully more effort to monitoring managers but also find a negative influence group decisions. They further suggest that a relationship between the proportion of women on the board central finding of the literature is that demographic differ- and Tobin’s Q in an analysis of US firms. Smith, Smith, and ences lower social cohesion between groups and that the Verner (2006) find a negative relationship between gender social barriers reduce the probability that minority view- diversity of the board and gross profits to sales for a sample of points will influence group decisions (Westphal & Milton, Danish firms but no statistically significant relationship 2000). Westphal and Milton (2000) indicate that this social between board gender diversity and several other accounting psychological concept of minority status is derived from measures of financial performance. Rose (2007) does not find social impact theory. This theory predicts that individuals a significant relationship between board gender diversity who have majority status have the potential to exert a and Tobin’s Q for a different sample of Danish firms. Farrell disproportionate amount of influence in group decisions and Hersch (2005) use Poisson regressions and an event © 2010 Blackwell Publishing Ltd Volume 18 Number 5 September 2010 400 CORPORATE GOVERNANCE 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License study to investigate the addition of female directors to US ports the idea that women directors and ethnic minority boards. They find no evidence that addition of a female to the directors may have different functions on the board. board affects return on assets or market returns to sharehold- Agency theory offers the possibility that diverse directors ers. Shrader, Blackburn, and Iles (1997) find no significant may be better monitors of management. While agency relationship between the percentage of female directors on theory suggests a link between board diversity and firm the board and profit margin, return on assets, or return on performance, the nature of the link is not clear. More and equity for a sample of US companies. tougher monitors may be either positive or negative as sug- The implications of the aforementioned investigations of gested by Adams and Ferreira (2009). the link between women on the board and financial perfor- Theories from social psychology suggest that diverse (out- mance are difficult to deduce. First, the results are mixed. group) directors may not have an influence on board deci- Two investigations find a positive relationship, three find no sions due to the internal group dynamics of the board. relationship, and two find a negative relationship. Second, Furthermore, more diverse members on the board may the statistical methods, data, and time periods investigated promote creative and innovative ideas but decision making vary greatly so that the results are not easily comparable. In may be slower and more conflicted with diverse directors. essence, the overall meaning of the body of research rests on In summary, an interdisciplinary set of theories provide a the efficacy of the research methodology employed in each solid indication that a link between board diversity and firm paper. The paper by Adams and Ferreira (2009) is convinc- financial performance is a realistic possibility. However, the ing but more investigation of this topic is warranted. relationship may either be positive or negative based on the Two investigations examined the link between a diversity theory. Furthermore, the limited amount of empirical evi- measure consisting of both women and ethnic minorities dence on the relationship does not provide clear support for and firm performance. Zahra and Stanton (1988) conduct a the direction of the link being either positive or negative. As canonical correlation analysis with a sample of US firms and a result, we state the hypotheses in null format and perform find no relationship between the percentage of females plus two-tailed statistical tests. The basic hypotheses are ethnic minorities on the board and return on assets, profit Hypothesis 1. All else being equal, the number of ethnic minor- margin, sales to equity, earnings per share, and dividends. ity directors on the board is not related to the financial perfor- Erhardt, Werbel, and Shrader (2003) find a significant posi- mance of the firm. tive link between the percentage of females plus ethnic minorities on the board and return on assets and return on Hypothesis 2. All else being equal, the number of women on the equity for a sample of US firms. We only located one inves- board is not related to the financial performance of the firm. tigation that examined the relationship between the ethnic- ity of directors and financial performance. Carter et al. (2003) find a significant positive relationship between the Diversity of Board Committees and percentage of ethnic minority directors on the board and Tobin’s Q. Firm Performance Bilimoria and Piderit (1994) submit that previous research indicates the delegation of corporate governance to board THEORETICAL PREDICTIONS OF THE committees facilitates effective board and corporate func- tions. They explain that board committees provide a means DIVERSITY-PERFORMANCE and structure for effective governance by facilitating special RELATIONSHIP tasks and addressing important corporate concerns (Bilimo- ria & Piderit, 1994). Jiraporn, Singh, and Lee (2009) argue Diversity of Overall Board and Firm Performance board effectiveness is accomplished through board commit- Resource dependence theory and human capital theory do tees. Kesner (1988) argues the most important decisions of not specifically predict a link between board diversity and the board are initiated at the committee level. If the above the financial performance of the firm but they are highly arguments are correct, the possibility exists that diverse suggestive of a positive relationship. Furthermore, the type directors may have more influence through board commit- of diversity should be important based on resource depen- tees than board membership. Simply being a part of a dence theory and human capital theory. Because women and smaller group may increase the influence of diverse direc- ethnic minorities have different human capital and external tors. Kesner (1988) argues that boards would not elect connections to the environment, we predict that they will women to the board’s most powerful and influential com- not have the same effect on board functions and, ultimately, mittees merely for the sake of firm image (tokenism) without firm performance. Brammer, Millington, and Pavelin (2007) consideration for women’s potential contributions. analyze the gender and ethnic diversity of a sample of UK Evidence supports the idea that many important decisions companies and conclude that “board diversity is influenced are made in board committees and those decisions affect the by a firm’s external business environment and particularly performance of the firm. Newman and Mozes (1999) find an imperative to reflect corresponding diversity among its that the relationship between CEO compensation and per- customers.” Brammer et al. (2007) find significant cross- formance is more favorable to the CEO if the compensation sector variation in gender diversity across industries while committee has more insiders on it. Sun and Cahan (2009) variation in ethnic diversity is much less pronounced. The examine the influence of compensation committee quality empirical evidence developed by Hillman et al. (2002), on the relationship between CEO cash compensation and Peterson and Philpot (2007), and Peterson et al. (2007) sup- accounting earnings. They find that CEO cash compensation Volume 18 Number 5 September 2010 © 2010 Blackwell Publishing Ltd BOARD DIVERSITY AND FINANCIAL PERFORMANCE 401 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License is more positively associated with earnings for firms with with RiskMetrics. Wharton Research Data Services (WRDS) is higher quality compensation committees. Adams (2003) a partner with the IRRC and describes the IRRC as the finds that board committees of diversified companies devote world’s leading source of impartial and independent infor- more effort to monitoring and board committees of growing mation on corporate governance. The IRRC gathers most of firms devote more effort to strategic issues. Garcia-Meca and the data from proxy statements (Securities and Exchange Sanchez-Ballesta (2009) complete a meta-analysis of 35 Commission statement DEF 14A). Data to compute the studies on earnings management and conclude that audit natural logarithm of total assets, the return on assets, and committee independence is one of the major mechanisms to Tobin’s Q are taken from the COMPUSTAT database. Table 1 constrain earnings management and assure the credibility of provides a description of each of the variables. a firm’s financial statements. Adams, Hermalin, and Weis- IRRC data is provided for each director individually and bach (2008) contend that the evidence on audit committees approximately 5,500 directors are included in the IRRC data- supports the concept that the make-up of the committee is base each year for the S&P 500 firms. A majority of the correlated with better accounting quality. Goh (2009) finds sample firms appear each year but a few firms migrate in that the audit committee plays an important role in monitor- and out of the index over time due to mergers, spin-offs, ing the remediation of material weaknesses in internal bankruptcy, and changes in the index by S&P. The IRRC control under the regulations of the Sarbanes-Oxley Act.9 database contains data on the following number of firms There is little direct evidence that the composition of board each year of the investigation: 1998 – 474 firms, 1999 – 473 committees is more important than the composition of the firms, 2000 – 472 firms, 2001 – 472 firms, and 2002 – 487 board in terms of producing better firm performance. We firms. We extract information from proxy statements for all could find only one investigation that addressed this issue. five years for any firm that may be on the S&P 500 list at least Klein (1998) studied the Standard and Poor’s (S&P) 500 firms once during the period 1998–2002 to mitigate potential for 1992 and 1993 and came to the following conclusion, “I sample bias due to changes in the S&P index. The final data find little association between firm performance and overall set consists of an unbalanced panel of 641 unique firms and board composition. However, by going into the inner work- 2,563 firm-years. ings of the board via board committee composition, I am able The IRRC data for the gender of a director is complete. to find significant ties between firm performance and how Gender is relatively easy to determine from information boards are structured.” Klein (1998) argues that membership in the proxy statements and company annual reports. on board committees provides a more accurate picture of However, the proxy statement does not specifically identify each director’s role on the board which should lead to a more gender or ethnicity, but the staff of the IRRC observes accurate test of the relationship between board composition titles, such as he or she, and other indirect evidence to and board effectiveness. Klein (1998) considers committee determine gender and ethnicity. Data collected by the IRRC membership to be a proxy for the duties, i.e., functions, of a on ethnicity is collected from the proxy statement, annual director on the board. Directors have a stronger and more reports, and other external sources. Like gender, the eth- direct impact on executive compensation, new director selec- nicity of directors is not a part of the proxy statement or tion, and other important actions that significantly affect other SEC filings but the proxy statement may contain corporate performance if they serve on board committees information that indirectly suggests ethnicity. However, with primary responsibility for these functions. Any unique ethnicity is more difficult to determine.10 For the years of advantages or disadvantages that might exist for women and this investigation (1998–2002), the IRRC identified the eth- ethnic minorities relative to board process are hypothesized nicity of approximately 80 per cent of the directors each to have a more direct effect through committee assignments. year. As a result, we could identify the ethnicity of the total At the minimum, the composition of board committees pro- board and board committees for 314 firms and 1,040 firm vides a valuable way to corroborate the relationship between years. We only used firms that we could identify the eth- board composition and firm performance. Based on Klein’s nicity of all of the directors. This is a strict screen but we conjecture plus empirical evidence and continuing with our believe this is the appropriate means to address the ethnic- null format, we hypothesize that ity of each board. When we used lagged variables, the sample was reduced to approximately 2,300 firm years for Hypothesis 3. All else being equal, the number of ethnic minor- the gender sample and 950 firm years for the ethnic minor- ity directors on a major board committee is not related to the ity sample. The ethnic minority sample is a sub-sample of financial performance of the firm. the gender sample. Hypothesis 4. All else being equal, the number of women direc- We investigate the nature of the data with missing ethnic- tors on a major board committee is not related to the financial ity indicators by selecting a random sample of 100 directors performance of the firm. from the directors in the IRRC database for the years 1998– 2002 without an ethnicity indicator. Then we complete an exhaustive search through LinkedIn, Lexis-Nexis, RESEARCH DESIGN Bloomberg, company websites, annual reports, and phone calls to companies to find the ethnicity of the directors in the Sample and Data sample of 100 directors with unidentified ethnicity. Our Our sample includes firms in the S&P 500 index for the analysis indicates that the sample of 100 directors with five-year period 1998–2002. We obtain data on directors and missing IRRC ethnicity indicators had a distribution of eth- other corporate governance variables from the Investor nicity almost the same as those directors in the IRRC data- Responsibility Research Center (IRRC), which is now affiliated base with identified ethnicity. © 2010 Blackwell Publishing Ltd Volume 18 Number 5 September 2010 402 CORPORATE GOVERNANCE 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TABLE 1 Variable Definitions Financial Performance (Dependent Variables) Tobin’s Q Tobin’s Q (Chung and Pruitt approximation) Return on Assets (ROA) (%) Annual net income divided by the book value of total assets at the end of the year Diversity (Independent Variables of Primary Interest) Number of Female Directors Total number of female directors on the board Number of Females on Nom. Com. Total number of female directors on the nomination committee of the board Number of Females on Audit Com. Total number of female directors on the audit committee of the board Number of Females on Comp. Com. Total number of female directors on the compensation committee of the board Number of Minority Directors Total number of Black and Hispanic directors* Number of Min. Dir. On Nom. Com. Total number of Black and Hispanic directors on the nomination committee of the board* Number of Min. Dir. on Audit Com. Total number of Black and Hispanic directors on the audit committee of the board* Number of Min. Dir. on Comp. Com. Total number of Black and Hispanic directors on the compensation committee of the board* Control Variable Natural Log of Firm Total Assets Natural logarithm of the book value of year-end total assets Governance Control Variables Number of Directors on the Board Total number of directors on the board Percentage Ownership of the Board Total number of common shares owned by all directors divided by the total number of common shares outstanding (multiplied by 100) Average Additional Directorships Average number of directorships held by each director on the board(in addition to the sample firm directorship) calculated as: the sum of the number of additional directorships for all directors divided by the total number of directors on the board Average Age of the Directors Sum of the ages of all directors divided by the total number of directors Combined CEO-Chairman of the Bd. Dichotomous variable equals 0 if the chief executive officer and the chairperson of the board are not the same person and 1 if a single individual holds both offices Number of Independent Directors Total number of independent directors as defined by IRRC Meeting Attendance of Directors Total number of directors that attended less than 75 per cent of the board meetings *Note: Black and Hispanic women were considered ethnic minorities. The regression results provided in Table 3 also provide in their investigation. Some investigators have developed or evidence of the comparability of the ethnicity data sample acquired specialized ethnicity data sets but the process is so and gender data sample. Models 1 and 4 are estimated with time consuming that they only have one year of data (Carter the smaller ethnic minority sample and Models 2 and 5 are et al., 2003; Peterson et al., 2007, and Brammer et al., estimated with the larger, complete gender sample. The only 2007). difference between the variables in Models 1 and 4 and Models 2 and 5 is the number of ethnic minority directors is not available in the larger, complete gender sample. Most of Model the regression coefficients are similar and those coefficients The fundamental model tested is that are significant are very similar. The IRRC data for the ethnicity of directors is the only data we are aware of that is publically available to any investiga- Perform = α + β1 Diversity + β 2 Previous Perform tor through the WRDS platform and provides the basis for + β 3 Firm Size + β 4 Governance + β 5 Firm replication. Jiraporn et al. (2009) use the IRRC ethnicity data + β6 Time Period + ε (1) Volume 18 Number 5 September 2010 © 2010 Blackwell Publishing Ltd BOARD DIVERSITY AND FINANCIAL PERFORMANCE 403 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License where Perform is the financial performance of the firm mea- the board of directors and the number of women directors sured by either Tobin’s Q or return on assets, Diversity is a on the audit committee, nomination committee, and com- measure of either the gender diversity or ethnic diversity of pensation committee. Similarly, we construct four ethnic the board and the board’s major committees, Previous minority variables from the IRRC designations of ethnicity Perform is a lagged value of Tobin’s Q or return on assets, – the total number of Black and Hispanic directors on the Firm Size is the natural log of the total assets of the company, board of directors and the total number of Black and His- Governance is a corporate governance characteristic of the panic directors on the audit committee, nomination com- firm, and Firm is a unique time-invariant unobservable firm mittee, and compensation committee. Women directors that characteristic based on firm level fixed effects in the regres- were identified as Black or Hispanic were counted as ethnic sion estimation, and Time Period is the time period in the minorities not women. We use count data instead of per- panel for that observation. The coefficient of primary interest centages because two diverse directors out of 14 total direc- is b1 and H0: b1 = 0 and Ha: b1 ⫽ 0. tors on the board may have more of an effect on firm We estimate two statistical versions of this model, a performance than one diverse director out of six total single ordinary least squares (OLS) regression equation directors.12 with firm and time fixed effects and a three stage least We calculate two measures of financial performance – squares (3SLS) regression analysis with firm and time Tobin’s Q and ROA. These measures are commonly used in fixed effects. Hermalin and Weisbach (2003) argue that the governance investigations as measures of performance but relationship of most board characteristics and firm perfor- they are not interchangeable or identical. They each measure mance are jointly endogenous. Adams and Ferreira (2009) a significantly different aspect of firm performance. Tobin’s suggest that endogeneity problems arise because of omitted Q in its original formulation is the market value of the firm’s variables that affect both the selection of diverse directors assets divided by the replacement value of the firm’s assets. and firm performance. Adams and Ferreira (2009) employ Computations of Tobin’s Q seen in the literature today often firm fixed effects in their analysis and they demonstrate use Chung and Pruitt’s (1994) approximation which equals that firm fixed effects have a significant impact on the [market value of firm’s common stock + liquidating value of results. Garay and Gonzalez (2008) use a single equation preferred stock + book value of debt (adjusted for net model with lagged dependent variables to address the working capital)] divided by the book value of total assets. problem of endogeneity. We follow a similar approach with Chung and Pruitt (1994) argue that their approximation lagged dependent variables and add fixed effects for the explains 96.6 per cent of the variability in the more theoreti- firm and time period. We consider the fixed effects esti- cally correct calculation of Tobin’s Q by Lindenberg and mates very important because they help to mitigate omitted Ross (1981). We use the Chung and Pruitt (1994) calculation variables and address unobserved changes over time. The that is basically the market value of the securities issued by firm fixed effects account for differences in the industry the firm divided by the book value of the assets. This and financial leverage used by the firm, among other firm measure is an indication of the wealth position of the major specific dimensions. We use the differential intercept providers of funds to the firm: shareholders and creditors. If dummy technique for the time period and within-group Tobin’s Q is greater than one, then the market value of the estimation for the firm fixed effects in this two-way fixed shareholders and creditors investment is greater than the effects model (Gujarati & Porter, 2009). Furthermore, we amortized historical cost of the assets. Because Tobin’s Q is a use robust standard errors in all of the panel data models market based measure to some degree, the metric embodies (Gujarati & Porter, 2009).11 a forecast of future cash flows produced by the firm and a A second problem associated with endogeneity is reverse market assessment of the investment opportunity set of the causality (Adams & Ferreira, 2009). The number of diverse firm. In theory, Tobin’s Q is a more complex measure of directors may affect performance but it is also possible that performance than ROA. Bhagat and Bolton (2008) argue that financially successful firms may select diverse directors. We stock market based measures are susceptible to investor address the question of reverse causality with 3SLS estima- anticipation. They contend that if investors anticipate an tion. We use 3SLS estimation instead of 2SLS estimation effect of a governance characteristic on financial perfor- because 3SLS accounts for cross-equation correlation that mance, long-term stock returns will not be correlated with exists in our data (Pindyck & Rubinfeld, 1998). Jackling and the governance characteristic, even if a real association exists Johl (2009) argue for 3SLS regression over 2SLS because (Bhagat & Bolton, 2008). Bhagat and Bolton (2008) rely pri- 3SLS addresses both potential endogeneity and cross- marily on accounting metrics although they also present correlation between equations. The technique of 3SLS Tobin’s Q and stock returns. regression is used in several recent corporate governance On the other hand, the ROA is an indication of the ability of investigations (Bebchuk, Cremers, & Peyer, 2007; Bhagat & the firm to produce accounting based revenues in excess of Bolton, 2008; Jackling & Johl, 2009; Prevost, Rao, & Hossain, actual expenses from a given portfolio of assets measured as 2002; Setia-Atmaja, 2009). amortized historical costs. ROA is an indication of the accounting income produced for the shareholders if ROA is calculated as net income divided by the book value of total Financial Performance and Diversity Variables assets.13 Tobin’s Q and ROA are shown to be related statisti- Refer to Table 1 for a complete description of each of the cally by Yermack (1996) and Carter et al. (2003). The correla- variables used in the analysis. We create four measures of tion between the two measures in our data is approximately women’s participation on the board for each firm in the 0.6. In summary, Tobin’s Q measures wealth and ROA sample from the IRRC database – the number of women on measures income. © 2010 Blackwell Publishing Ltd Volume 18 Number 5 September 2010 404 CORPORATE GOVERNANCE 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Predetermined Variables adding a dummy variable to represent if the CEO and Chair of the Board are combined. Brickley, Coles, and Jarrell (1997) The 3SLS procedure is a simultaneous equation method that find that a combined CEO-Chair leads to lower cash flows assumes that firm performance and board diversity are and market value while Goyal and Park (2002) find that CEO endogenous and the other variables in the system model are turnover after poor financial performance is lower if the predetermined (Pindyck & Rubinfeld, 1998). Unfortunately, CEO and Chair are the same person. However, Adams, many of the variables of interest in corporate governance Almeida, and Ferreira (2005) find no evidence that CEO investigations are not truly exogenous, determined com- duality is related to a firm’s stock returns. Elsayed (2007) pletely outside the model system, but endogenous (Herma- finds no relationship between CEO duality and financial lin & Weisbach, 2003). However, we can use the lagged performance except when the performance of the firm is low values of the endogenous variables as predetermined vari- and then he finds a positive relationship. ables even though they are not determined completely The effect of independent directors on the board is a major outside of the system of equations (Pindyck & Rubinfeld, area of interest in the corporate governance literature. 1998). Jackling and Johl (2009) follow the suggestion of Numerous investigators have explored this issue and reach a Bhagat and Bolton (2008) and use the lagged value of Tobin’s mix of conclusions, but some find a relationship (Baysinger Q and ROA as instruments. We include the lagged value of & Butler, 1985; Bhagat & Bolton, 2008; Jackling & Johl, 2009; Tobin’s Q and ROA in all of our estimations.14 Rosenstein & Wyatt, 1990). We include the number of inde- Size of the firm is often used as a control variable in an pendent directors as a control variable and use the IRRC analysis of financial performance and is shown to be related definition of an independent director. Independent direc- to market returns by Fama and French (1992), among others. tors are defined by the IRRC as not an executive or employee Multiple studies show that asset size is related to Tobin’s Q and not linked in some other way. The IRRC has a list of (Faleye, 2007; Prevost et al., 2002; Yermack, 1996). We conditions that create a link including being a family include the value of the natural log of total assets in the member of an executive or a former employee. Executives of regressions to control for the size of the firm. the firm compose the largest segment of the directors that We create a second set of variables that measure various are not independent in our sample. aspects of the governance mechanisms of the firm. These The ownership position of the board is expected to affect have been shown to be related to firm performance in pre- financial performance but the exact nature of this relation- vious investigations. ship is subject to some debate (Demsetz & Villalonga, 2001). Yermack (1996) finds that board size and Tobin’s Q are Monks and Minow (2004) argue that higher board owner- inversely related. However, Jackling and Johl (2009) find a ship results in better monitoring and a more involved board. strong positive relationship between board size and finan- However, higher ownership may result in an entrenched cial performance that supports evidence from Dalton, Daily, board that does not promote the interests of all stakeholders. Ellstrand, and Johnson (1998) and Pearce and Zahra (1992). We include the percentage of the total shares outstanding The argument for a positive association between board size that are owned by the board in the financial performance and financial performance is that larger boards will bring equations. better information because of greater knowledge from more Fich and Shivdasani (2006) find that firms with a majority directors to firm decision making (Jackling & Johl, 2009). of the outside directors serving on three or more boards have Nicholson and Kiel (2007) and Van den Berghe and Levrau lower market-to-book ratios, lower profitability, and lower (2004) also argue that increasing the number of directors sensitivity of CEO turnover to firm performance. Thus, a provides a larger pool of information that should translate busy board may mean that directors are over-committed and into better performance (Jackling & Johl, 2009). The positive are not good monitors for the shareholders. On the other association between board size and financial performance hand, some investigations have found a positive effect on flows from resource dependency theory while Yermack firm performance if directors have additional directorships (1996) makes an agency theory argument for a negative (Ferris, Jagannathan, & Pritchard, 2003). Directors with mul- relationship. Both theory and empirical evidence indicate tiple directorships have more networking and contacts that that we should include the number of directors on the board may produce benefits for the firm through more extensive in the financial performance equation. This variable is also capabilities to access the external environment. The number important as a control variable because we use the number of additional board memberships of current directors is used of women directors and the number of ethnic minority as a predetermined variable in the financial performance directors instead of the percentage of women and ethnic equation. minority directors. Carter et al. (2003) find that larger boards Vafeas (1999) reports evidence that board meeting fre- are more likely to have more women and ethnic minority quency and firm performance are related and concludes that directors suggesting that having a larger number of women board activity is an important dimension of board process. and/or ethnic minority directors could be related to the size Jackling and Johl (2009) find no relationship between board of the board, not the financial performance of the firm. Other meetings and financial performance in a sample of Indian investigations have all used the percentage of women and firms. We hypothesize that meeting attendance by directors ethnic minority directors that adjusts for the number on the is an indication of the quality of board process and include board within the variable. the percentage of the total number of directors that attended Past studies have found that the leadership structure of less than seventy-five per cent of board meetings in the the firm and power of the CEO will have an impact on financial performance equation. This variable captures a dif- financial performance. Therefore, we control for this by ferent dimension of board process than the number of board Volume 18 Number 5 September 2010 © 2010 Blackwell Publishing Ltd BOARD DIVERSITY AND FINANCIAL PERFORMANCE 405 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License meetings and is considered a better measure of director regression estimates with the one-year lag. We did not lag involvement than simply the number of board meetings the diversity variables in the 3SLS regressions because they held. were hypothesized to be simultaneous and endogenous. The predetermined variables were lagged values of endogenous variables in the 3SLS regressions. Lagged Variables We use lagged variables in the fixed effect single equations models because we hypothesize that the effect of board EMPIRICAL RESULTS diversity on financial performance will occur over time. Theory does not predict the length of time required for an Descriptive Statistics effect. In addition, only one other previous investigation Table 2 provides descriptive statistics for our sample. The uses lags and those are one period (Farrell & Hersch, 2005). measures of financial performance indicate that the firms in The number of lags becomes an empirical question without the sample were financially successful on average over the direction from theory. We estimate the equations with a one- five-year period investigated but there was wide variation in year lag and a two-year lag. The results are essentially the the performance variables. The mean Tobin’s Q was 1.19, same but we lose data with the two-year lag. So we report which is above one and suggests the market value of the firm TABLE 2 Descriptive Statistics Variables Firm Years Mean Std. Dev. Min. Max. Panel A: Women Directors Sample Tobin’s Q 2,563 1.19 1.17.14 4.54 Return on Assets (%) 2,563 3.90 5.22 -7.88 14.97 Number of Female Directors 2,563 1.30.93 0 6 Number of Females on Nom. Com. 2,563.48.65 0 4 Number of Females on Audit Com. 2,563.65.69 0 4 Number of Females on Comp. Com. 2,563.43.60 0 4 Natural Log of Firm Total Assets 2,563 8.35 2.19 2.03 13.87 Number of Directors on the Board 2,563 11.21 3.15 4 32 Percentage Ownership of the Board 2,563 5.85 12.84 0 100 Average Additional Directorships 2,563 1.36.73 0 4.62 Average Age of the Directors 2,563 59.35 3.24 43.33 70.38 Combined CEO-Chairman of the Bd. 2,563.71.45 0 1 Number of Independent Directors 2,563 7.77 2.80 0 29 Meeting Attendance of Directors 2,563.26.57 0 4.3 Panel B: Minority Directors Sample Tobin’s Q 1,040 1.25 1.18.14 4.54 Return on Assets (%) 1,040 4.34 5.33 -7.88 14.97 Number of Minority Directors 1,040 1.09.89 0 6 Number of Min. Dir. on Nom. Com. 1,040.39.58 0 3 Number of Min. Dir. on Audit Com. 1,040.58.67 0 3 Number of Min. Dir. on Comp. Com. 1,040.35.56 0 3 Natural Log of Firm Total Assets 1,040 8.82 2.04 2.08 13.87 Number of Directors on the Board 1,040 11.83 3.02 5 27 Percentage Ownership of the Board 1,040 3.83 10.75 0 91.50 Average Additional Directorships 1,040 1.56.74.05 4.62 Average Age of the Directors 1,040 59.96 2.59 48.6 70.38 Combined CEO-Chairman of the Bd. 1,040.76.42 0 1 Number of Independent Directors 1,040 8.45 2.63 1 26 Meeting Attendance of Directors 1,040.27.60 0 4.3 Number of Female Directors 1,040 1.58.95 0 6 Note: Panel A is the sample of firms with all of the directors identified as male or female. Panel B is the sample of firms with all of the directors identified as Black, Hispanic, or Caucasian. © 2010 Blackwell Publishing Ltd Volume 18 Number 5 September 2010 406 CORPORATE GOVERNANCE 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TABLE 3 Fixed Effects Regression Estimates of the Relationship Between Firm Performance and the Number of Ethnic Minority and Women Directors on the Board Independent Variables Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Tobin’s Q Tobin’s Q Tobin’s Q ROA ROA ROA Dependent Dependent Dependent Dependent Dependent Dependent Variable Variable Variable Variable Variable Variable Number of minority directors.02 (.66).33† (1.78) Number of female directors.03 (1.10).01 (.98).41* (2.07).57** (4.68) Number of directors.01 (.58) -.00 (-.55) -.00 (-.37).11 (1.24).06 (1.11).11* (1.97) CEO-Chair duality.04 (.82) -.02 (-.58) -.02 (-.55) -.29 (-.52) -.35 (-1.58) -.32 (-1.46) Additional directorships.00 (.16).00 (.09).00 (.22).18 (.04).09 (.48).17 (.96) Meeting attendance.03 (1.3).02 (1.44).02 (1.42) -.05 (-.20).04 (.22).01 (.06) Board ownership.00 (1.25).00 (1.14).00 (1.12).02 (1.30).01 (1.21).01 (1.08) Independent directors -.01 (-1.48) -.01 (-.89) -.00 (-.74) -.11 (-1.31) -.03 (-.54).00 (.07) Average age of directors -.01 (-1.12) -.00 (-.97) -.00 (-1.05) -.06 (-.79) -.03 (-.82) -.05 (-1.12) Log total assets -.01 (-.88).00 (0.24).02 (.41) -.06 (-.86) -.04 (-.92) -.01 (-.34) Tobin’s Q.78** (38.43).79** (61.85).79** (62.18) ROA.59** (14.74).58**(19.87).59** (20.26) Intercept.74 (1.55).49* (2.02).49* (2..0) 4.84 (.99) 3.38 (1.42) 3.65 (1.50) Firm Fixed Effects Yes Yes Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Yes Yes Firm-Year Observations 958 2,347 2,347 908 2,224 2,224 Firms in Sample 339 577 577 315 533 500 Adj. R-square.81.79.79.37.37.36 F-Statistic 173.98 387.07 409.05 31.21 60.19 62.45 Notes: All independent variables are lagged one period. Robust standard errors are used to calculate t-statistics. Probability values are based on a t-statistic for a two-tailed test of significance. The first number in each cell is the regression coefficient and the value in parentheses is the associated t-value. **indicates p <.01, *indicates p <.05, and †indicates p <.10. is greater than the book value of the assets. However, the samples with the gender sample mean of 11.21 directors variation in the sample is significant with the minimum compared to the ethnic minority director sample mean of Tobin’s Q 0.14 and the maximum 4.54. The ROA reveals 11.83. The mean number of women on the important board similar high variation. The mean ROA is 3.90 per cent, but committees indicates that women have slightly higher rates the minimum is -7.88 per cent and the maximum is 14.97 per of participation than ethnic minorities on board committees, cent. The financial performance of the gender diversity but this is partially explained by the slightly lower number sample and the ethnic diversity sample suggest that the of ethnic minority directors compared to women. firms in both samples have similar levels of performance. In 71 per cent of the firm years, the positions of CEO and This should not be surprising because the ethnic minority chairperson of the board were combined. The average board sample is a sub-sample of the gender sample. Mean Tobin’s has approximately 11 directors, of which about eight are Q for the gender sample is 1.19 compared to 1.25 for the independent. The average director held 1.5 directorships in ethnic minority sample and mean ROA is 3.90 per cent for addition to the one in the sample firm. The average age of a the gender sample and 4.34 per cent for the ethnic minority director was about 59 years and about 25 per cent of the sample. The firms in the two samples are very similar in size directors in the sample did not attend 75 per cent of the with the mean natural log of total assets 8.35 for the gender board meetings. These statistics are quite similar to those sample and 8.82 for the ethnic minority sample. reported in other analyses of major U. S. corporations, for The average number of women directors on a board is 1.30 example Carter et al. (2003). and the average number of ethnic minority directors on a board is 1.09 over the five-year period of the sample. However, the average number of women on a board was 1.58 in the ethnic minority sample, which indicates that the Fixed Effect Single Equation Analysis two samples are similar but not identical. The average Table 3 reports the fixed effect regression tests of Hypoth- number of directors on a board was very similar for the two eses 1 and 2. Volume 18 Number 5 September 2010 © 2010 Blackwell Publishing Ltd BOARD DIVERSITY AND FINANCIAL PERFORMANCE 407 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TABLE 4 Fixed Effects Regression Estimates of the Relationship Between Firm Performance and the Number of Ethnic Minority and Women Directors on Board Committees: Dependent Variable is Tobin’s Q Independent Variables Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Audit Nomin. Comp. Audit Nomin. Comp. Minority Minority Minority Women Women Women Number of diverse (minority.02 (.75).01 (.62).01 (.19).01 (.58).01 (.61).02 (.84) or women) directors on committee Number of female directors.03 (1.14).03 (1.19).03 (1.22) Number of directors.01 (.71).01 (.71).01 (.74) -.00 (-.40) -.00 (-.35) -.00 (-.39) CEO-Chair duality.04 (.84).04 (.85).04 (.85) -.02 (-.57) -.01 (-.55) -.01 (-.52) Additional directorships.00 (.18).01 (.21).01 (.27).00 (.16).00 (.17).00 (.14) Meeting attendance.03 (1.27).03 (1.27).03 (1.29).02 (1.44).02 (1.45).02 (1.39) Board ownership.00 (1.24).00 (1.21).00 (1.20).00 (1.17).00 (1.10).00 (1.16) Independent directors -.01 (-1.54) -.01 (-1.48) -.01 (-1.54) -.00 (-.78) -.00 (-.84) -.00 (-.75) Average age of directors -.01 (-1.12) -.01 (-1.19) -.01 (-1.21) -.00 (-1.04) -.00 (-1.04) -.00 (-1.06) Log total assets -.01 (-.85) -.01 (-.87) -.01 (-.79).00 (.35).00 (.32).01 (.35) Tobin’s Q.79** (38.35).79** (38.35).79** (38.36).79** (62.17).79** (62.20).79** (62.73) Intercept.73 (1.52).75 (1.53).77 (1.60).49* (2.04).49* (2.05).50* (2.05) Firm Fixed Effects Yes Yes Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Yes Yes Firm-Year Observations 958 958 958 2,347 2,347 2,347 Firms in Sample 339 339 339 577 577 577 Adj. R-square.81.81.81.79.79.79 F-Statistic 177.67 175.31 174.36 384.72 383.33 385.10 Notes: All independent variables are lagged one period. Robust standard errors are used to calculate t-statistics. Probability values are based on a t-statistic for a two-tailed test of significance. The first number in each cell is the regression coefficient and the value in parentheses is the associated t-value. **indicates p <.01, *indicates p <.05, and † indicates p <.10. Models 1 and 2 reveal that the coefficients for the number and the number of ethnic minority directors on boards. The of ethnic minority directors and the number of female direc- interaction term was not significant in any of the equations tors are not different from zero which means there is no and the other coefficients were very similar to the results evidence of a significant link between Tobin’s Q and the reported in Table 3. So we did not report the full results number of women directors or Tobin’s Q and the number of which are available from the authors. ethnic minority directors. Model 3 contains only the control Table 4 contains the fixed effect regression estimates for variables and the adjusted R2 exhibits almost no change for the board committee equations with Tobin’s Q as the depen- the larger gender sample when the number of women direc- dent variable. None of the regression models indicate a sig- tors is omitted. The only explanatory variable that is signifi- nificant relationship between Tobin’s Q and the number of cant in Models 1, 2, and 3 is lagged Tobin’s Q. The correlation women directors on a board committee or the number of between Tobin’s Q and lagged Tobin’s Q is high with a ethnic minority directors on a board committee. The only Pearson correlation coefficient of 0.9.15 The regression esti- significant variable in any of the equations is lagged Tobin’s mates in Table 3 for Model 4 reveal a significant (0.10 level) Q and the coefficients for lagged Tobin’s Q are very similar positive link between ROA and the number of ethnic minor- to the equations in Table 3. ity directors on the board. Model 5 in Table 3 exhibits a Table 5 contains the fixed effect regression estimates for positive significant (0.01 level) coefficient for the number of the board committee equations with ROA as the dependent women directors on the board. When the diversity variables variable. Models 1, 2, and 3 indicate that the number of are omitted in Model 6 of Table 3, the R2 exhibits almost no ethnic minority directors on any of the major board commit- change. The lagged ROA variable is highly significant in tees is not associated with ROA. However, Models 4, 5, and Models 4, 5, and 6 but t-statistics are lower than the t-statistics 6 indicate that the coefficients for the number of women on for lagged Tobin’s Q in Models 1, 2, and 3. The Pearson all of the major board committees are positively related to correlation coefficient between ROA and lagged ROA is 0.5. ROA. The regression coefficients for the number of women We estimate Models 1, 2, 4, and 5 in Table 3 with an added on the audit committee and nomination committee reported interaction term between the number of women on boards in Models 4 and 5 of Table 5 are positive and significant at © 2010 Blackwell Publishing Ltd Volume 18 Number 5 September 2010 408 CORPORATE GOVERNANCE 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TABLE 5 Fixed Effects Regression Estimates of the Relationship Between Firm Performance and the Number of Ethnic Minority and Women Directors on Board Committees: Dependent Variable is Return on Assets Independent Variables Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Audit Nomin. Comp. Audit Nomin. Comp. Minority Minority Minority Women Women Women Number of diverse (minority.37 (1.46) -.04 (-.16).16 (.59).53** (3.17).53** (3.20).46* (2.19) or women) directors on committee Number of female directors.43* (2.24).47* (2.41).46* (2.35) Number of directors.14 (1.55).14 (1.63).14 (1.60).10† (1.80).11* (1.99).10† (1.87) CEO-Chair duality.04 (.11).07 (.19).05 (.13) -.35 (-1.58) -.32 (-1.43) -.29 (-1.33) Additional directorships.21 (.83).27 (1.08).25 (.98).11 (.65).13 (.74).14 (.77) Meeting attendance -.07 (-.28) -.06 (-.25) -.06 (-.23).03 (.18).04 (.27).00 (.03) Board ownership.02 (1.31).02 (1.26).02 (1.22).01 (1.36).01 (1.02).01 (1.18) Independent directors -.12 (-1.47) -.12 (-1.50) -.12 (-1.46) -.01 (-.24) -.02 (-.40).00 (.06) Average age of directors -.06 (-.81) -.07 (-.95) -.07 (-.96) -.04 (-1.07) -.04 (-1.06) -.05 (-.81) Log total assets -.05 (-.78) -.04 (-.65) -.05 (-.71) -.03 (-.64) -.03 (-.74) -.02 (-.52) Return on Assets.59** (14.52).59** (14.81).60** (14.91).58** (20.03).58** (20.25).58** (20.16) Intercept 4.70 (.96) 5.28 (1.07) 5.33 (1.09) 3.64 (1.50) 3.61 (1.50) 3.66 (1.52) Firm Fixed Effects Yes Yes Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Yes Yes Firm-Year Observations 908 908 908 2,224 2,224 2,224 Firms in Sample 315 315 315 533 533 533 Adj. R-square.37.37.37.36.36.36 F-Statistic 31.23 31.23 31.29 59.22 59.43 59.29 Notes: All independent variables are lagged one period. Robust standard errors are used to calculate t-statistics. Probability values are based on a t-statistic for a two-tailed test of significance. The first number in each cell is the regression coefficient and the value in parentheses is the associated t-value. **indicates p <.01, *indicates p <.05, and †indicates p <.10. the.01 level. The regression coefficient for the number of cannot be rejected for any of the board committees with women on the compensation committee is positive and sig- either Tobin’s Q or ROA used as a measure of financial nificant at the.05 level. The only other coefficients in Models performance. The tests of Hypothesis 3 provide no evidence 4, 5, and 6 of Table 5 that are significant are for the number of a significant relationship between the numbers of ethnic of directors on a board committee and lagged ROA. The minority directors on any of the board committees investi- coefficients for the number of directors are significant at the gated and financial performance as measured by either ROA.10 and.05 levels and positive. or Tobin’s Q. Tests of Hypothesis 4 support a positive, sig- The fixed effect single equation analysis rejects null nificant relationship between the number of women direc- hypothesis 1 at the.10 probability level when ROA is used to tors on all of the board committees investigated and measure firm financial performance. Null Hypothesis 1 financial performance as measured by ROA. However, there cannot be rejected when Tobin’s Q is used to measure finan- was no relationship between the number of women on board cial performance. The hypothesis tests support the idea that committees and Tobin’s Q. a positive relationship exists between the number of ethnic The results of the fixed effect single regression analysis minority directors on the board and ROA but there is no suggest a different relationship between the number of support for a relationship between the number of ethnic women directors on the board and financial performance minority directors on the board and Tobin’s Q. Null Hypoth- and the relationship between the number of ethnic minority esis 2 can be rejected at the.01 level when ROA is used to directors on the board and financial performance. The evi- measure financial performance but cannot be rejected when dence suggests a positive link in both cases between the Tobin’s Q is used to measure financial performance. Tests of number of diverse directors on the board and ROA. Hypothesis 2 indicate that a positive relationship exists However, the evidence is stronger for female directors. Fur- between the number of women on the board and ROA but thermore, the evidence for board committees supports a do not support a relationship between the number of positive link between the number of female directors and women on the board and Tobin’s Q. Null Hypothesis 3 ROA but does not support a link between ROA and ethnic Volume 18 Number 5 September 2010 © 2010 Blackwell Publishing Ltd BOARD DIVERSITY AND FINANCIAL PERFORMANCE 409 14678683, 2010, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-8683.2010.00809.x by University Of Thessaly, Wiley Online Library on [09/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TABLE 6 3 SLS Regression Estimates of the Relationship Between Firm Performance and the Number of Ethnic Minority and Women Directors on the Board Independent Variables Model 1 Model 2 Model 3 Model 4 Tobin’s Q Tobin’s Q ROA ROA Dependent Dependent Dependent Dependent Endogenous Endogenous Endogenous Endogenous Variable Variable Variable Variable Number of minority directors.15 (.64) 1.72 (.81) Number of female directors.01 (.19) -.16 (-.74).19 (.46) 1.21 (.76) Number of directors.00 (.20).01 (.44) -.02 (-.15).01 (.09) CEO-Chair duality.01 (.31) -.01 (-.18).00 (.01) -.22 (-.95) Additional directorships.01 (.08).05 (1.63) -.36 (-.61) -.07 (-.30) Meeting attendance.01 (.28).00 (.04).08 (.28).01 (.07) Board ownership.00† (1.67).00 (.79).01 (.80) -.06 (-.61) Independent directors -.01 (-1.22) -.00 (-.73) -.04 (-.03) -.20 (-1.11) Average age of directors -.01 (-.62) -.00 (-.86) -.03 (-.42).15 (.98) Log total assets -.01 (-1.17).01 (1.21) -.14 (-1.50) -.04 (-.49) Tobin’s Q.81** (60.39).80** (42.93) ROA.56** (14.54).53** (12.83) Intercept.17 (1.13).00 (.02) 3.07* (2.16) 1.69† (1.69) Firm Fixed Effects Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Chi Square Statistic 4,738.96 7,332.11 408.91 863.77 Notes: In Models 1 and 3 the number of minority directors is an independent endogenous variable that is not lagged and all of the other independent variables are lagged one time period. In Models 1 and 3 the number of female directors is not endogenous and is lagged one time period. In Models 2 and 4 the number of female directors is an independent endogenous variable that is not lagged and all of the other independent variables are lagged one time period. Only the results of the 3SLS procedure equations with Tobin’s Q and ROA as the dependent variable are reported. Probability values are based on a z-statistic for a two-tailed test of significance. The first number in each cell is the regression coefficient and the value in parentheses is the associated z-value. **indicates p <.01,