0  

Does Corporate Board Diversity Affect Corporate Payout Policy?*

by

Soku Byoun Associate Professor of Finance Hankamer School of Business

Baylor University One Bear Place #98004

Waco, Texas 76798 (254) 710-7849

Soku_Byoun@baylor.edu

Kiyoung Chang Assistant Professor of Finance

College of Business University of South Florida-Sarasota/Manatee

Sarasota, FL 34243 (941) 359-4359

chang1@sar.usf.edu

Young Sang Kim Associate Professor of Finance

Haile/US Bank College of Business Northern Kentucky University Highland Heights, KY 41099

(859) 572-5160 kimy1@nku.edu

April 2011

* We appreciate the support for this project that was provided by the Hankamer School of Business at Baylor University. We also appreciate comments and suggestions by Byoung-Hyoun Hwang, Jin-Mo Kim and David Reeb.

1  

Does Corporate Board Diversity Affect Corporate Payout Policy?

Abstract

We find that firms with diverse boards are more likely to pay dividends and tend to pay larger

dividends than those with non-diverse boards. Our results suggest that board diversity has a significant impact on dividend payout policy. The impact of board diversity on dividend payout policy is particularly conspicuous for firms with potentially greater agency problems of free cash flow, suggesting that diverse board helps mitigate the free cash flow problem. Our findings are consistent with the argument that board diversity enhances the monitoring function of directors and board independence for the benefit of shareholders.

JEL Classification: G30; G34; G35 Key words: Board Diversity; Board Independence; Payout Policy; Monitoring; Free Cash Flow

2  

Does Corporate Board Diversity Affect Corporate Payout Policy?

I. Introduction

What is the effect of corporate board diversity?1 The Alliance of Board Diversity (ABD)—a

collaboration of Catalyst, the Executive Leadership Council, and the Hispanic Association of Corporate

Responsibility (HACR)—argues that diverse board directors bring independent, creative and fresh ideas

to the boardroom, enhancing corporate performance. Adams and Ferreira (2009) suggest that gender-

diverse boards are more effective monitors, while Carter et al. (2008) argue that women and minority

directors provide unique information to the board to improve strategic decision making. Thus, diverse

boards can contribute to better understanding of the complexities of the environment, reduced risk of

group think, and more astute corporate decisions. On the other hand, firms may select female and

minority directors as tokenism (Baysinger and Butler (1985), Adams and Ferreira (2009)): having a

diverse board may appear legitimate in the views of the public, the media, and the government. However,

there are potential costs of board diversity: e.g., lack of communication, pursuing distinct personal

agendas, and conflicts of interests among directors (Ferreira (2010)).

The effect of board diversity can be captured by differences in behavior when women and minority

directors are present in the board. Since it is difficult to observe directly such differences in behavior,

however, the literature almost exclusively focuses on the relation between board diversity and a measure

of firm performance in order to determine whether board diversity matters. The presumption is that if

board diversity increases board independence and brings unique information to the board, it should affect

corporate performance.

Previous studies provide mixed results. Early studies on U.S. firms generally find negative or no

impact of board diversity on firm performance. For instance, Zahra and Stanton (1988) find no significant

relation between corporate diversity and firm performance measures, whereas Shrader, Blackburn, and                                                             1 We use “diversity” in order to refer to the presence of woman or minority director in the board. We do not differentiate them because our analyses yield similar results whether the diversity is in regard to gender or ethnicity only.

3  

Iles (1997) find that the presence of woman directors has negative impact on firm performance. More

recently, however, Carter et al. (2003) and Carter et al. (2008) document a positive impact of gender and

ethnic board diversity on Tobin’s Q and market value, respectively. Erhardt et al. (2003) find evidence of

positive relation between board diversity and accounting measures of firm performance. Adams and

Ferreira (2009) show that the impact of gender diversity on firm performance is heterogeneous (i.e.,

gender diversity has beneficial effects in companies with weak shareholder rights, whereas it has

detrimental effects in companies with strong shareholder rights). Anderson et al. (2011) also show that

board diversity has beneficial effects on more complex firms and firms with greater CEO power.

Evidence from international studies is also mixed. Farrell and Hersch (2005) and Rose (2007) find

no significant results using Denmark firms, while Ahern and Dittmar (2009) find significant reduction in

their market value after Norwegian firms adjusted to mandatory female quotas on boards. In sum,

whether or not board diversity affects corporate decisions is still an unresolved issue. Besides, the

interpretations of the previous results are complicated by the endogeneity of board diversity and firm

performance.2

In this study, we investigate whether board diversity has a significant impact on corporate decisions.

To this end, we examine whether firms with diverse boards adopt different payout policies vis-à-vis those

with non-diverse boards. We focus on the dividend payout policy because one of the most important

conflicts between managers and shareholders is the free cash flow problem (Jensen, 1986). Managers can

deploy free cash flow in a way that does not maximize shareholder wealth. The literature suggests that

dividends can mitigate the free cash flow problem.3 Thus, dividend payout policy is a key component of

returning value to shareholders. If a diverse board---with a variety of skills, perspectives, backgrounds,

and resources---promotes objective monitoring and yields independent voices, it is likely to discipline

management through its impact on dividend payout policy. Consistent with this argument, Anderson et al.

(2011) suggest that shareholders derive greater monitoring benefits from a diverse board, which provides

                                                            2 See, among others, Hermalin and Weisbach (1998, 2003) and Adams et al. (2010) for discussions on this issue. 3 For example, see Grossman and Hart (1980), Easterbrook (1984), Jensen (1986), and DeAngelo, DeAngelo, and Stulz (2006).

4  

various viewpoints on executive actions. Further, Kandel and Lazear (1992) argue that heterogeneity

amongst team members enhances mutual monitoring, suggesting greater monitoring effectiveness of a

diverse board on the behalf of shareholders. If a diverse board is more independent and activistic, then it

is less likely to be a “managerial rubber-stamp” and more likely to serve as a “watchdog for shareholders.”

Accordingly, board diversity has the potential to help align the incentives of managers and shareholders

through its impact on the payout policy.

To the best of our knowledge, no other study has examined the relation between board diversity

and corporate payout policy. Examining such a relation can aid in understanding the role of board

diversity in corporate governance. Unlike firm performance which may be affected indirectly by board

diversity through its impact on corporate decisions, the dividend payout policy is a directly measurable

corporate decision that is approved by the board of directors. Thus, the major advantage of our study is to

investigate the direct impact of board diversity on a major corporate decision.

To the extent that the diverse board helps mitigate the agency problem of free cash flow through its

enhanced monitoring and independence, we expect that corporate board diversity has a greater impact on

dividend payout policy for firms with potentially greater agency problems: firms with low management

stock ownership, weak shareholder rights, and large free cash flows. This line of argument is consistent

with the findings of Adams and Ferreira (2009) and Anderson et al. (2011) in that the impact of board

diversity on firm performance is greater for companies with weak shareholder rights or strong CEOs

because those firms are likely to face greater agency problems. On the other hand, if the diverse board

suffers from more conflicts and communication breakdowns, it would be a less effective monitor and

have little impact on corporate payout policy.4

We find that firms with diverse boards are more likely to pay dividends and tend to pay larger

dividends than those with non-diverse boards. After controlling for various firm and board characteristics,

our results suggest that diverse boards have significant impacts on dividends. In particular, firms with

                                                            4 Indeed, such conflicts and communication breakdowns appear to be eminent when the board consists of foreign independent directors (Masulis et al. (2010)).

5  

diverse boards are associated with about 16% higher probability to pay dividends than firms without

diverse boards and firms with diverse boards have on average about 14% higher dividend payout ratio

(cash dividend divided by total assets) than the payout ratio of firms without diverse boards. Furthermore,

the effect of board diversity on dividend payout policy is significantly greater for firms with high free

cash flow, with more entrenched management, and with low CEO ownership. Thus, our findings suggest

that enhanced monitoring of a diverse board has greater benefits for high free cash flow firms that are

subject to managerial entrenchment and lack of managerial incentives.

We also find that firms with multiple diverse directors tend to have greater propensity to pay

dividends. Moreover, we find a positive association between the number of diverse directors in the board

and payout ratio, which suggests that there are incremental effects of additional diverse directors on the

payout level. However, the diverse board of the firm with a woman or minority CEO has little impact on

its payout policy. This finding reveals that the gender or ethnic tie between the CEO and directors clouds

the board’s independent monitoring. This finding is in line with the growing literature which suggests that

informal social ties between directors and the CEO impede objective monitoring of the CEO and board

independence (Hwang and Kim (2009, 2011) and Schmidt (2009)). Thus, it is not simply the presence of

diverse directors in the board that exerts the significant impact on the dividend payout policy but the

diversity between the board and the CEO.

In assessing the impact of board diversity on corporate payout policy, it is possible that dividend-

paying firms choose more diverse boards rather than board diversity induces more dividend payouts. We

address this endogeneity concern in several ways. First, we examine the change in payout policy when a

company adds a new diverse director to its board. This allows us to compare the change in dividend

policy for the same firm before and after it adds a diverse director. Our results show that significantly

larger portion of firms pay dividends after they added a diverse director to their boards. Also, firms pay

significantly higher dividends after adding a new diverse director for the first time. However, the change

in dividend payout level is not significant when firms add additional diverse director to their already

6  

diverse boards. Moreover, there are decreases in dividend payouts, albeit statistically insignificant, when

firms switch back to non-diverse boards.

Second, we incorporate the propensity-score matching method in order to examine the treatment

effect of board diversity after matching firms with similar conditional probabilities of adopting a diverse

board given their characteristics. We run difference-in-difference regressions based on the propensity-

score matched firms to compare the changes in dividend policy between non-diverse board firms (the

control group) and firms that add a new diverse director for the first time (the treatment group). The

results based on the matched sample confirm that there are significant and positive effects of board

diversity on dividend payout policy.

We also explore alternative explanations for the positive association between board diversity and

dividend payout policy. If firms that are more attentive to shareholders’ voices choose high payout policy

(Baker and Wurgler (2004), Li and Lie (2006)) as well as more diverse board since it is right thing to do

in the eyes of shareholders, we may observe the positive association between board diversity and payout

policy. For example, companies may have pressure from the institutional investors for greater board

diversity as well as higher dividend payouts. Firms with high information asymmetry may also pay higher

dividends to signal their prospects to shareholders (Ross (1977), Bhattacharya (1979)) while promoting

board diversity. Accordingly, we examine these alternative explanations by conditioning our tests on the

ownership structure and the extent of information asymmetry. We find no supporting evidence for these

alternative arguments.

We perform an array of robustness checks including instrumental variable probit and tobit

regressions, the two stage least square instrumental variable approach (2SLS-IV), and the generalized

method of moment (GMM) estimation method in order to address the endogeneity concern. We also

include various other board characteristics as controls in our regressions. Our results are robust to these

alternative estimations and specifications. Additionally, board diversity remains relevant for dividend

payout policy throughout various sub-periods. Admittedly, however, we cannot completely rule out the

potential for endogeneity bias.

7  

Our study contributes to the governance literature in the following ways. First, we show that a

diverse board exerts a significant impact on the corporate dividend policy. Given the findings of recent

studies on the relation between board diversity and firm performance (see, e.g., Carter et al. (2003),

Erhardt et al. (2003), Carter et al. (2008), Adams and Ferreira (2009), Anderson et al. (2011)), our

evidence is particularly revealing. Even though the previous studies conjecture that the positive impact of

board diversity on firm performance stems from effective monitoring, how such enhanced monitoring

yields benefits to shareholders remains unclear. Our evidence suggests that the positive impact of a

diverse board on firm performance can be partly attributable to its effectiveness in addressing the agency

problem of free cash flow through improved monitoring and independence of the board.

Second, we provide evidence for the effect of board diversity on a major corporate decision.

Aforementioned studies focus on the effect of the simple presence of women or minority directors in the

board. We show that adding a diverse director to the board has a positive impact on payout policy.

However, we find that the effect of board diversity on corporate payout policy is either negative or

insignificant for firms with a woman or minority CEO. Thus, the benefits of board diversity are not

materialized when directors share the same gender or ethnic tie with the CEO. Adding directors who

share the same gender or ethnic background as the CEO may diminish the board’s independence and

hence the benefits of board diversity. This finding is consistent with the results in Hwang and Kim (2009)

and Schmidt (2009) who show that informal social ties between directors and the CEO significantly

diminish directors’ monitory and disciplinary effectiveness. Hwang and Kim (2011) also find that

increased earnings management is associated with informal social ties between an audit committee and

the CEO. Our study complements these previous studies in that gender and ethnic ties also affect the

effectiveness of board monitoring and disciplinary functions.

Interestingly, the Securities and Exchange Commission (SEC), with the intention of increasing

corporate board diversity, issued the rule in 2010 that requires public companies to disclose how they

8  

view diversity with respect to their boards.5 Our finding has an important implication for policies aiming

to increase the number of diverse directors in corporate boardrooms. What makes the significant

difference in the board’s monitory effectiveness is not the sheer number of diverse directors in the board

but the “diversity” they add to the board.

We proceed the paper as follows. Section II describes the data and provides summary statistics.

Section III reports univariate results and Section IV presents results from the multivariate regressions.

Section V explores alternative explanations and Section VI provides robustness checks. Section VII

contains summary and concluding remarks.

II. Data and Summary Statistics

A. Data

Throughout our analysis we utilize data from the following sources: the Investor Responsibility

Research Center (IRRC, now acquired by Risk Metrics), the Center for Research in Security Prices

(CRSP), the CDA/Spectrum Institutional (13f) Holdings, and Standard & Poor's Compustat for the 12-

year period from 1997 through 2008. Our initial sample includes all firms in the IRRC database with

information available on gender and ethnicity of directors.6 Following the literature, we exclude financial

(SIC Codes 6000-6999) and utility companies (SIC Codes 4900-4999).7 From our initial sample we

further exclude firms that do not have financial and price information from Compustat and CRSP. After

imposing these requirements, our sample consists of 2,234 unique firms or 13,325 firm-year observations.

B. Board Diversity Measures

We measure board diversity in several ways. First, we define board diversity dummy variable

(Diverse_dum) that takes one if at least one board member is woman or minority and zero if no woman or

                                                            5 The SEC RIN 3235-AK28, Proxy Disclosure Enhancements, effective on February 28, 2010. 6 The IRRC director database provides director information starting from 1996. Due to many missing observations in 1996, however, our sample period starts from 1997. 7 See, for example, Bates, Kahle, and Stulz (2009).

9  

minority director is present in the board. A director is considered minority if he/she is African American,

Asian, or Hispanic. Our second measure is defined as the proportion of woman and minority directors in

the board (Pct_diverse). For the purpose of addressing potential confounding effects between gender and

ethnicity, we consider separate diversity measures for woman and minority: woman dummy variable

(Woman_dum) that equals one for a board with at least one woman and zero otherwise; the proportion of

women directors in the board (Pct_woman); and the proportion of minority directors (not including

woman) in the board (Pct_minor). Additionally, we check robustness of our results using the number of

women and/or minority directors in the board. Finally, we define a minority or woman CEO dummy

variable (Diverse_CEO) that equals one if the CEO is a female or minority and zero otherwise. We

provide definitions of the variables used in our study in the appendix.

C. Dividend Payout Measures

Following the literature, we consider three measures of dividend payout. The first measure is a

dummy variable that equals one if a firm pays cash dividend and zero otherwise (DIV_dum). We also

measure payout by: (1) dividend-to-total asset ratio (DIV_TA) and (2) dividend yield defined as dividend

per share divided by fiscal year ending stock price (DIV_P). By incorporating the stock price, dividend

yield can measure a payout policy that reflects shareholders’ perspective. However, this measure can be

affected largely by the fluctuation in stock price rather than by changes in dividends. Thus, we use

DIV_TA along with the DIV_dum for our main results. Some studies utilize other measures such as

dividend divided by net income (DIV_NI) or dividend per share divided by earnings per share before

extraordinary items (DIV_E). These earnings-based measures show how much of a firm’s earnings are

returned to shareholders in the form of dividends. However, these variables are not defined and difficult

to interpret when earnings are zero or negative. Thus, we consider these alternative measures only to

check the robustness of our main results. We do not include repurchases in our measures since pre-

committed dividends are effective means of mitigating agency conflict due to governance (John and

Knyazeva (2006)). However, we also check the robustness of our results when repurchases are included.

10  

D. Summary Statistics

Panel A in Table 1 shows the distribution of our sample firms over the 1997-2008 period. The

proportion of firms with diverse boards and the proportion of diverse directors in the board each year

have been steadily increasing over the sample period except for the most recent two years which coincide

with the 2007-2008 financial crisis period. About 64% of the sample firms have diverse directors in their

boards and these diverse directors represent about 12% of all directors.

********************************Table 1 *********************

We also report the sample distribution across the Fama and French 12 industries (available at

French’s website) excluding utilities and financial industries in Panel B. Consumer non-durables and

Chemicals and allied products industries are associated with the largest proportions of firms with diverse

boards and the highest percentages of diverse directors in the board, whereas Oil, gas, and coal extraction

and products and Telephone and television transmission industries have the lowest percentages of

diverse-board firms and diverse board members. Given the substantial variation in board diversity across

industries, we incorporate industry effects as well as year effects in our multivariate regressions.

III. Univariate Analysis

Before we analyze the effect of board diversity on dividend payout policy, we compare some board

characteristics of dividend payers versus non-payers in Table 2. The proportion of firms with at least one

woman (Woman_dum) and minority (Minor_dum) director in the board is significantly higher for

dividend-paying firms (71% and 74%, respectively) than for non-paying firms (45% and 49%,

respectively). The percentage of firms with multiple female or minority members in the board

(Multi_diverse) is also significantly higher for dividend payers (43%) than for non-payers (19%).

Dividend-paying firms have significantly larger boards (Bsize) than non-dividend-paying firms. This

11  

result is not surprising given that dividend payers are larger firms than non-payers (as shown in Table 3).

We also report other board characteristics such as the proportion of independent directors, the number of

outside board positions held by directors, the proportion of directors missing more than 25% of board

meetings, average age of directors, average tenure of board members, and an indicator for a dual position

of CEO and board chairmanship. We utilize these board characteristics as controls in our multivariate

analysis.

********************************Table 2*********************

We also examine Pearson correlation coefficients among the variables (not reported in a table).

There are positive correlations between board diversity measures and board size (34 to 44 per cent) and

between board diversity measures and the percentage of independent directors relative to the total number

of directors (21 to 25 per cent). Our results hold whether the diverse directors are outsiders or not.

However, the effect of board diversity on dividend policy tends to be stronger in the presence of

independent diverse directors.

We also compare firm characteristics for four different groups in Table 3: dividend-paying and non-

dividend-paying firms with and without diverse boards. For both dividend payers and non-payers, firms

with diverse boards are significantly larger (LogTA) and less leveraged (Leverage). Diverse board firms

also have higher earned-to-total equity ratio (RE_TE) than do non-diverse board firms. Regardless of

board diversity, dividend-paying firms are associated with lower R&D, lower Q, less cash holdings

(Cash_TA), and less stock return volatility (STDRET) while having higher return on asset (ROA) and

higher earned-to-total equity ratios. Dividend payout ratios (DIV_TA and DIV_P) indicate that do firms

with diverse boards pay significantly higher dividends than firms without diverse boards.

********************************Table 3*********************

12  

In order to examine how additional diverse directors in the board affect payout policies, we

compare payout measures across the number of diverse directors in the board in Panel A of Table 4. The

result reveals a strong monotonic and positive relation between the number of diverse directors in their

boards and the proportion of firms that pay dividends, which suggests that the more diverse directors the

firm adds, the more likely it is to pay dividend. The dividend payout ratios (DIV_TA and DIV_P) tend to

show similar monotonic patterns. These results are interesting and consistent with the argument that more

diverse members have greater influence on board decisions. Anderson et al. (2011) argue that intra-group

conflicts and less communication in the diverse board can lead to isolation of members not belonging to

the majority group. Such isolation is less likely to occur when there are several minority members in the

board and their combined voices can exert more influence on dividend payout policy.

********************************Table 4*********************

If board diversity affects corporate payout policy, we expect to observe significant changes in

payout policy after firms add a new diverse director. In order to examine this effect, we report changes in

payout policy when companies add new diverse directors to their boards in Panel B of Table 4. For those

undergoing a change from non-diverse to diverse board, all dividend payout measures are significantly

higher after the change. The proportion of firms paying dividend increases from 40% to 51%, with the

difference being significant at the 1% level, after they add a diverse director to their boards. There are

also significant increases in payout ratios after firms’ adding diverse directors to their boards for the first

time (e.g., DIV_TA changes from 0.59% to 0.84%). Firms maintaining diverse boards throughout the

sample period also show significantly greater dividend payouts relative to those that remain as non-

diverse board firms. Interestingly, when firms switch from diverse to non-diverse board firms, there are

decreases in dividends, albeit statistically insignificant.

In order to further explore how adding a new diverse director affects dividend payout over time, we

select firms that add a new diverse director between 2001 and 2004 (new-diverse board firms). We then

13  

plot their average dividend payout measures over the entire sample period in Figure 1.8 For comparison,

we also include the average dividend payouts of firms whose boards remain diverse and non-diverse,

respectively, throughout the sample period. The plots show that before they add new diverse directors to

their boards, new-diverse board firms’ payouts resemble those of non-diverse board firms. However, after

they add new diverse directors, their dividend payouts diverse from those of non-diverse board firms,

approaching those of diverse firms as time passes by.

********************************Figure 1*********************

Overall, the univariate results suggest that board diversity brings a significant change in dividend

payout policy. Given the significant differences in their board and firm characteristics, however, we turn

to multivariate regression analyses in order to control for these differences in the next section.

 

IV. Multivariate Analysis

We first test whether there is a significant difference in dividend payout policy between companies

with and without diverse boards. The regression models include the following independent variables from

the dividend literature: firm size (LogTA), leverage (Leverage), research and development expenses

(R&D), market-to-book ratio (Q), return on assets (ROA), earned-to-total equity ratio (RE_TE), cash

holdings (Cash_TA), tangible assets (PPE_TA), and stock return volatility (STDRET). Lloyd et al. (1985)

and Vogt (1994) show that firm size affects firms' dividend-payout ratio. Fama and French (2001) find

that growth firms pay substantially less dividends than non-growth firms. R&D, as a proxy for growth

opportunity, is defined as the ratio of research and development expenses to the book value of total assets.

We also include market-to-book asset ratio, where the market value of total assets is total assets minus

                                                            8 There are 117 firms that added a new director during this period. As precaution for extreme values, we winsorize payout ratios at 99 percent.

14  

common equity plus market value of equity. Leverage, defined as total debt divided by book assets,

controls for its substituting effect for dividend. DeAngelo, DeAngelo, and Stulz (2006) show that earned-

to-total equity ratio (retained earnings divided by total equity) is an important variable in explaining the

dividend payout ratio. We also include stock return volatility following Chay and Suh (2009). Even

though we do not report them in the main tables for simplicity, our results are intact when we include

various board characteristics including board size and the percentage of independent directors as

additional control variables.9

A. Regressions

In Table 5, we estimate logit regressions in order to examine the effect of board diversity on the

propensity to pay dividend. The dependent variable is the dummy variable for dividend payout

(DIV_dum). We run a separate regression for each of the five alternative board diversity measures: (1)

percentage of women in the board (Pct_woman); (2) percentage of women and minority directors in the

board (Pct_diverse); (3) percentage of minority (excluding women) directors in the board (Pct_minor); (4)

dummy variable for firms with multiple female and minority directors (Multi_diverse); (5) dummy

variable for firms with woman or minority directors in the board (Diverse_dum). We use year and

industry effects to control for time trends and omitted-variable bias. We report robust t-statistics after

adjusting for firm clustering.

********************************Table 5*********************

The results show that all proxies for board diversity (woman only, ethnic minority only, and both

combined) carry highly significant and positive coefficient estimates, which suggests that diverse boards

are significantly more likely to induce firms to pay dividends than do non-diverse boards. The coefficient

                                                            9 We report many additional results in our working  paper version of the paper available at 

http://ssrn.com/author=107811.  

15  

estimates on control variables are consistent with previous results and suggest that the decision to pay

dividends is positively associated with firm size, ROA, earned-to-total equity ratio, and asset tangibility,

but negatively associated with leverage, R&D, and stock return volatility. In order to assess the effects on

probability (rather than on the odds ratio) of board diversity, we estimate the marginal probability holding

all other variables at their respective means. The result (not reported in the table) suggests that the

probability to pay dividend is about 16% higher when the firm has a diverse board.

Table 6 reports estimation results for the ordinary least square (OLS) regressions with dividend

payout ratio (DIV_TA) as the dependent variable. Again, the coefficient estimates on all the board

diversity measures are positive and significant at the 1% level. When we use Diverse_dum in regression

(5), the coefficient estimate on DIV_TA is 0.3%. Given the average payout ratio of about 2.1% for the

sample firms (in Table 3), the 0.3% estimate implies on average about 14% higher dividend payout ratio

by firms with diverse boards relative to firms without diverse boards. Thus, board diversity affects not

only the payout decision, but also the payout level of dividends. Interestingly, the coefficient estimates on

market-to-book ratio (Q) that are not significant in logit regressions are significant and positive in

regressions on payout ratio. Thus, firm’s Q does not affect the propensity to pay dividends, but high Q

firms tend to pay higher dividends.

Given that the dividend payout ratio is bounded by zero, we also run tobit regressions. Additionally,

we apply the Fama-MacBeth methodology to gauge the statistical significance of the coefficients from a

time series of yearly fitted coefficients. The results are similar to those in Table 6 and not reported in

tables. Overall, the results in Tables 5 and 6 suggest that diverse boards have positive effects on corporate

dividend payout policy. Since the results are similar when we consider gender and ethnicity separately,

we do not report them.

********************************Table 6*********************

16  

B. The effect of adding a diverse director to the board

The univariate results in Table 4 suggest that firms pay higher dividends after adding a diverse

director to their boards. We further verify this evidence in Table 7 while controlling for some firm

characteristics. The regressions include a dummy variable (Diverse_Add) that is equal to one for firms

that add a new diverse director to their boards at time t and zero otherwise. The coefficient estimate on

Diverse_Add measures the effect of adding a diverse board member to the board. For payout dummy

(DIV_dum) and payout ratio (DIV_TA), we run logit and OLS regressions, respectively, for all sample

firms (regressions (1) and (4)) and also separately for firms that do not have any diverse director in their

boards at time t-1 (regressions (2) and (5)) and for firms that already have diverse directors in their boards

at time t-1 (regressions (3) and (6)). The sample size drops to 9,832 firm-year observations as we require

information on prior board diversity. The coefficient estimates on Diverse_Add are significant only for

firms that do not have any diverse directors in their boards as of the previous year. The results suggest

that there is a significant incremental impact of an additional director on the payout policy only when the

firm adds a new diverse director to its board for the first time. Our unreported results further reveal that

there is little effect of board diversity when the majority of the board consists of diverse directors.

********************************Table 7*********************

C. The effect of a woman/ minority CEO

If the main benefits of a diverse board are its effective monitoring and independence, these benefits

are less likely when the board members and the CEO belong to the same diverse group. Besides, a

woman/minority CEO may have brought in directors with the same gender or ethnic tie, which can cloud

directors’ objective monitoring. Thus, we test if the diverse board has the same impact on the dividend

payout policy for firms whose CEO is also an ethnic minority or woman. The results are shown in Table 8.

We run logit regressions ((1) to (3)) for payout propensity (DIV_dum) and OLS regressions ((4) to (6)) for

payout ratio (DIV_TA). Models (1) and (4) include a dummy variable that is equal to one for firms with a

17  

diverse CEO (Diverse_CEO) along with the diverse board dummy variable (Diverse_dum). The

coefficient estimates on Diverse_dum are positive and significant. The coefficient estimates on

Diverse_CEO are negative but not significant in both regressions. When we divide the sample into firms

with and without a diverse CEO, however, the effect of board diversity on the payout propensity is

negative and marginally significant for firms with a diverse CEO, while the effect on the payout ratio is

insignificant for these firms. The effect of board diversity on the payout policy for firms with a non-

diverse CEO remains positive and highly significant. We observe similar results when we use the

proportion of diverse directors in the board instead of the diverse board dummy variables (not reported).

These results suggest that the benefits of board diversity are limited for firms in which the CEO and other

directors have different gender or ethnic backgrounds. Thus, our findings suggest that the differential

payout policy for the diverse-board firm is not driven simply by the presence of diverse directors in its

board but by the diversity that they bring to the boardroom.

********************************Table 8*********************

D. The effect of board diversity and free cash flow

If the positive impact of the board diversity on the payout policy is resulting from its enhanced

monitoring, then the impact of board diversity on the payout policy is likely to vary systematically with

firms’ susceptibility to the free cash flow problem. Thus, we refine our tests by conditioning the effect of

board diversity on the following three proxies for the degree of free cash flow problem: free cash flow

(FCF), management entrenchment index (GINDEX as measured by Gompers, Ishii, and Metrick (2003)),

and CEO ownership. We define: 1) High-FCF that equals one for firms with FCF greater than the sample

median and zero otherwise; 2) high GINDEX if GINDEX is greater than 9 and low GINDEX otherwise;

and 3) low CEO ownership if the proportion of CEO ownership is less than the sample mean and high

CEO ownership otherwise.

18  

The logit regression results for the dividend payout propensity are presented in Table 9. Regression (1)

includes an indicator for the diverse board (Diverse_dum), an indicator for high free cash flow (High-FCF)

and the interaction between these two variables (FCF*Diverse). The positive and significant coefficient

estimate on the Diverse_dum suggests that firms with diverse boards, on average, are more likely to pay

dividends than firms without diverse boards. The coefficient estimate on High-FCF is negative and

marginally significant, suggesting that high free cash flow is associated with low dividend payout

propensity. The positive and significant coefficient estimate on the interaction variable (FCF*Diverse)

suggests that there is a significant differential effect of the board diversity on dividend payout propensity

between high and low FCF firms.

********************************Table 9*********************

We further compare the effects of board diversity on payout policy between low and high GINDEX

firms and between low and high CEO ownership firms in regressions (2)-(5) in order to examine whether

the effect of board diversity on dividend policy depends on managerial entrenchment and CEO ownership

as well as free cash flow. The significant and positive coefficient estimates on the interaction variable

(FCF*Diverse) in regressions (2) and (4) suggest that board diversity has the most pronounced effects on

the propensity to pay dividends for firms with high GINDEX and low CEO ownership that yield large

free cash flows. Thus, the results are consistent with our conjecture that enhanced monitoring of a diverse

board exerts a greater influence on the dividend payout policy of a high FCF firm that is also subject to

managerial entrenchment and lack of managerial incentives.

We also run the OLS regressions with payout ratio (DIV_TA) as the dependent variable in Table 10.

The coefficient estimates on the interaction variable (FCF*Diverse) are all positive and significant. The

results suggest that firms with diverse boards pay significantly higher dividends when they produce

higher free cash flows regardless of managerial entrenchment or incentives.

19  

********************************Table 10*********************

We also estimate regressions with the proportion of diverse directors (Pct_diverse) instead of

diversity dummy variable (Diverse_dum) as an alternative board diversity measure, with DIV_P instead

of DIV_TA as the dependent variable, and with various board characteristics included as additional

control variables. The results (not reported) suggest that the effect of board diversity is the most

significant for high FCF firms with high GINDEX and that the differential effects between high and low

GINDEX/CEO ownership firms are less pronounced when Pct_diverse is used instead of Diverse_dum.

The results are otherwise similar to those in Table 10.

E. Propensity Score Matching and Difference-in-Difference Approach

In this section we incorporate the propensity-score matching method in order to match firms with

similar conditional probabilities of adopting a diverse board given their characteristics. Our main interest

here is to match firms with similar propensities to have a diverse board, as predicted by their current

board and firm characteristics and then to examine the effect of board diversity on the dividend payout

policy between treatment (diverse board) and control (non-diverse board) firms. Accordingly, we select

firms that add diverse directors to their boards for the first time during the sample period as the treatment

group and firms that never have diverse directors in their boards during the sample period as the control

group. We first run a logit regression with independent variables related to firm and board characteristics

and with the dependent variable defined as a binary variable indicating whether a firm adds a diverse

director or not. We follow Adams and Ferreira (2009) to include board size (Bsize), the proportion of

independent directors (Pct_indep), the logarithm of total assets (LogTA), and the number of business

segments (Segments) in our logit model. Additionally we include return on assets (ROA), Tobin’s q (Q),

and the proportion of director ownership (Dir_own).

From the logit regression for each year, we obtain the propensity score, the predicted probability of

being a diverse board. We then use propensity scores to identify initial matched candidates by imposing a

20  

caliper (the maximum propensity score distance) within a quarter of standard deviation of the logit of the

propensity score. In this process, the observations are ordered randomly since nearest neighbor matching

without replacement depends on the observation order. We then choose a matching firm from the initially

matched firms for each firm in the treatment (diverse board) group that has the closest Mahalanobis

distance calculated based on the propensity score, firm size, board size, and the proportion of independent

directors. From this process, we generate 342 matched pairs with similar propensities and characteristics.

Using Mahalanobis metric matching within propensity score caliper produces the best balance for the

covariates between groups and is considered superior to the method relying only on the propensity score

(Dagostino (1998)).

In Table 11, we use the matched sample to test if there are significant differential treatment effects

(difference-in-difference) between firms with and without treatment by examining the changes from prior

to post treatment (adding a diverse director). In Panel A of Table 11, we report the difference-in-

difference payouts. Pre (post)-treatment is the period prior to (after) the matching. The changes in

dividends after the treatment suggest that there are significant and positive treatment effects for diverse

board firms but no significant treatment effects for matched non-diverse board firms based on all three

dividend payout measures. The differential changes (difference-in-difference) are also positive and

significant, suggesting that firms have significantly greater propensity to pay dividends and pay

significantly higher dividends after adding a diverse director relative to matched control firms. It is

notable that, unlike the overall sample of firms with diverse boards, firms that add a diverse director to

their boards for the first time tend to pay lower dividends than matched control firms. In fact, similar

patterns are observed in Figure 1. Thus, firms that newly adopt diverse boards are not necessarily high

dividend-paying firms before they adopt diverse boards.

In Panel B of Table 11, we run difference-in-difference regressions which, along with other control

variables, include the following two dummy variables and their interaction: Treatment that is equal to one

for a firm that adds a diverse director during the 1997-2008 sample period and zero otherwise and

Post_Treat that is equal to one for the period after the firm added a new diverse director to its board. The

21  

interaction of these two dummy variables (Treat_effect) measures the treatment effect of board diversity.

The negative coefficient estimates on Treatment suggest that treatment firms tend to pay less and lower

dividends than control firms. The coefficients on Post_Treat tend to be negative but insignificant when

year and industry effects are included, reflecting the diminishing dividends in more recent years (Fama

and French(2001) and DeAngelo et al. (2006)). The coefficient estimates on treatment effect (Treat_effect)

are significant and positive for all regressions, reinforcing the result in Table 4 and Figure 1.

********************************Table 11*********************

It is also possible that firms that expect to pay high dividends adopt diverse boards. In order to

examine this possibility, we also match firms based on the propensity to pay dividends rather than on the

propensity to add a diverse director. Not surprisingly, the pre-treatment dividend measures are similar for

these matched groups (not reported), but the treatment effects of board diversity remain positive and

highly significant (Table 12). Thus, it is less likely that firms’ expected dividend payouts are confounded

with the diversity treatment effects. Regression results with free cash flow problem proxies for the

matched samples are similar to those in Tables 9 and 10 and not reported.

********************************Table 12*********************

V. Alternative explanations

A. The effect of Institutional Investors

As an alternative to the monitoring and independence argument of the diverse board, we explore the

potential influence of institutional investors on board diversity and dividend policy. Our interest here is

whether the positive association between board diversity and dividend policy is driven by the demand of

institutional investors. In order to check the validity of this alternative explanation, we first run

regressions where the diversity indicator is the dependent variable and lagged institutional ownership (the

22  

proportion of shares owned by institutions relative to total outstanding shares) is an independent variable,

along with other control variables. Our results (not reported) indicate that the coefficient estimate on

lagged institutional ownership is positive and significant, suggesting that institutional investors have a

strong impact on a firm’s board diversity. 10 Given this result, we further examine the impact of

institutional ownership on dividend payout policy in Table 13. For the overall sample, the estimation

results for regressions (1) and (4) show that the percentage of institutional ownership (P_institution) is

negatively associated with the propensity to pay dividend and the payout ratio. Furthermore, the impacts

of board diversity on dividend payouts remain positive and significant for both low- and high-institutional

ownership firms. It is possible that institutional investors exert influence on board diversity for the

purpose of strengthening board independence (e.g., see Dobbin and Jung (2010) for this line of argument).

Nonetheless, our evidence is difficult to reconcile with the argument that institutional investors directly

affect both board diversity and dividend payout policy in such a way that they are positively associated.

********************************Table 13*********************

B. Signaling

A common alternative argument for the dividend payout policy is that dividends are used as a

signaling device (Ross (1977), Bhattacharya (1979)). If firms with high information asymmetry pay

higher dividends while promoting board diversity to signal their prospects to shareholders, the positive

effect of board diversity on the payout policy can be attributed to signaling rather than monitoring.

Accordingly, we examine this alternative explanation by conditioning our tests on the extent of

information asymmetry facing the firm’s shareholders. If signaling is the underlying motivation, we

would expect high information asymmetry to reinforce the increased use of dividends by diverse-board

firms, ceteris paribus.

                                                            10 Investigating board diversity in this regard would be an interesting endeavor. Yet, further investigation of this issue is beyond the scope of our study. Anderson et al. (2011) examine the determinants of diverse board, but without considering the role of institutional investors.  

23  

Following the literature, we choose proxies for firm’s information asymmetry as follows: i) limited

analyst following (Chang, Dasgupta, and Hillary (2006)), disagreement among analysts (Krishnaswami

and Subramaniam (1999)), earnings surprise (Barclay and Smith (1995)), a low proportion of tangible

assets (Barth and Kasznik (1999)), and market response to earnings announcements (John, Knyazeva, and

Knyazeva (2011)).

Our results (not reported) suggest that whether a firm faces low or high information asymmetry does

not yield any significant difference on the relation between board diversity and dividend payout policy.

Thus, we do not find supporting evidence for the signaling explanation.

VI. Robustness Checks

This section summarizes various robustness checks we perform throughout our analyses. Even though

we suppress many results from reporting for the sake of brevity, the working paper version contains many

additional tables.

A. Instrumental Variables Approach

Despite our various efforts to address the endogeneity issue, it is still possible that a firm’s board

diversity captures unobserved firm characteristics that also determine its dividend payout policy. Our

preliminary tests indicate that there is significant correlation between our main diverse proxy variables

and the error terms in the dividend regressions. In order to address this endogeneity issue, we first apply

the 2SLS-IV approach (Table 13). For instrumental variables, we consider the following variables

obtained from KLD Research & Analytics: (1) an indicator variable for firms with notable progress in the

promotion of women and minorities (div_str_b); (2) an indicator for firms that show strong record of

purchasing or contracting with women and minority-owned businesses (div_str_e); and (3) institutional

ownership (Inst_Own). We choose these indicator variables since they are likely to be related to board

diversity but unrelated to the dividend policy. We also rely on the GMM approach (not reported) that is

known to be less prone to the endogeneity bias. Our results are robust to these alternative estimation

24  

approaches. However, to the extent that our instruments are imperfect and the endogeneity-robust GMM

approach still contains bias, our results are not fully free from the endogeneity bias.

********************************Table 14*********************

B. Alternative Specifications

We also produce results using alternative payout ratios (DIV_NI and DIV_E), including repurchases

in payout ratio measures, and alternative diversity measures (Pct_diverse, Woman_dum, Pct_woman, and

Pct_minor). Given that dividend payout ratio is bound by zero, we also run a tobit regression for each of

dividen payout ratios. We also control for the impact of other governance quality on the dividend payout

ratio as used in John and Knyazeva (2006), Jiraporn et al. (2010), and Masulis and Mobbs (2011): multi-

segment dummy, geographic-segment dummy, outside directorship, and other board characteristics. We

also examine whether our results are consistent over various sub-periods. From the sub-period analysis

we confirm that our results are not driven by Sarbanes-Oxley (SOX) Act. Furthermore, we check our

results without the 2007-2008 financial crisis period. Our results are robust to these alternative measures,

specifications, estimation methods, and time periods.

VII. Conclusion

We find that firms with diverse boards have greater propensity to pay dividends and pay larger

dividends, especially when firms generate large free cash flows and when firms are prone to agency

problems. Furthermore, adding a new diverse director to the board is accompanied by a significant

increase in dividend payouts in the following years. Our findings are surprising since we did not presume

such clear effects of board diversity on dividend payouts, let alone any prior inkling from the literature.

Yet, our evidence is consistent with the argument that board diversity enhances the monitoring function

and independence of directors for the benefit of shareholders. More interestingly, when diverse members

25  

are the majority of the board or when the CEO and directors share the same diverse background, there is

little evidence of diversity effect on the dividend policy. Thus, what makes real difference is not the

number of diverse directors in the board but board diversity that they add to the boardroom.

26  

References Adams, Renée B., and Daniel Ferreira. 2009. “Women in the Boardroom and Their Impact of Governance

and Performance.” Journal of Financial Economics 94:2, 291-309. Adams, Renée B., Benjamin E. Hermalin and Michael S. Weisbach, 2010. “The Role of Boards of

Directors in Corporate Governance: A Conceptual Framework & Survey.” Journal of Economic Literature 48, 58-107.

Ahern, Kenneth R., and Amy K. Dittmar. 2009. “The Changing of The Boards: The Value Effect of a

Massive Exogenous Shock.” Working Paper, University of Michigan. Anderson, Ronald C., David M. Reeb, Arund Upadhyay, and Wanli Zhao. 2011. “The Economics of

Director Heterogeneity.” Financial Management, forthcoming. Baker Malcolm and Jeffery Wurgler, 2004. “A Catering Theory of Dividends.” Journal of Finance 59,

1125-1165. Barclay, Michael, and Clifford W. Smith, 1995. “The Maturity Structure of Debt.” Journal of Finance

50(2), 609-631. Barth, Mary, and Ron Kasznik, 1999. “Share Repurchases and Intangible Assets.” Journal of Accounting

and Economics 28(2), 211-241. Bates, Thomas W., Kathleen M. Kahle, and Rene Stulz, 2009. “Why Do U.S. Firms Hold so Much More

Cash than They Used to?" Journal of Finance 64(5), 1985-2021. Baysinger, Barry D., and Henry N. Butler. 1985. “Corporate Governance and the Board of Directors:

Performance Effects of Changes in Board Composition.” Journal of Law, Economics, and Organization 1:1, 101-124.

Bhattacharya, S. 1979. “Imperfect Information, Dividend Policy and the ‘Bird in the Hand’ Fallacy.” Bell

Journal of Economics 10, 259– 270. Carter, David A., Betty J. Simkins, and W. Gary Simpson. 2003. “Corporate Governance, Board

Diversity, and Firm Value.” Financial Review 38:3, 33-53. Carter, David A., Frank D’Souza, Betty J. Simkins, and W. Gary Simpson. 2008. “The Diversity of

Corporate Board Committees and Financial Performance.” Working Paper, Oklahoma State University.

Chang, Xin, Sudipto Dasgupta, and Gilles Hilary. 2006. “Analyst Coverage and Financing Decisions.”

Journal of Finance 61(6), 3009-3048. Chay, J.B. and Jungwon Suh. 2009. “Payout Policy and Cash-flow Uncertainty.” Journal of Financial

Economics 93, 88-107. Dagostino, Ralph, 1998. “Propensity Score Methods for Bias Reduction in the Comparison of a

Treatment to a Non-Randomized Control Group.” Statistics in Medicine 17, 2265-2281.

27  

DeAngelo, Harry, Linda DeAngelo, and Rene M. Stulz. 2006. “Dividend Policy and the Earned/Contributed Capital Mix: A Test of the Lifecycle Theory.” Journal of Financial Economics 81, 227-254.

Dobbin, Frank and Jiwook Jung. 2010. “Corporate Board Gender Diversity and Stock Performance: The

Competence Gap or Institutional Investor Bias?” Working Paper, Harvard University. Erhardt, Niclas L., James D. Werbel, and Charles B. Shrader. 2003. “Board of Director Diversity and

Firm Financial Performance.” Corporate Governance: An International Review 11:2, 102-111. Easterbrook, F. H., 1984. “Two Agency-Cost Explanations of Dividends.” American Economic Review

74, 650-659. Fama, E. F., and H. Babiak. 1968. “Dividend Policy: An Empirical Analysis.” Journal of the American

Statistical Association 63, 1132-1161. Fama, E., French, K., 2001. “Disappearing Dividends: Changing Firm Characteristics or Lower

Propensity to Pay?” Journal of Financial Economics 60, 3–43. Fama, E. F. and MacBeth, J. D., 1973. “Risk, Return, and Equilibrium: Empirical Tests.” Journal of

Political Economy 81, 607-636. Farrell, Kathleen A., and Philip L. Hersch. 2005. “Additions to Corporate Boards: The Effect of Gender.”

Journal of Corporate Finance 11:1, 85-106. Ferreira, Daniel, Chapter 12: Board Diversity, 2010. “Corporate Governance: A Synthesis of Theory,

Research, and Practice,” Wiley, 225-241. Edited by H. Kent Baker, Ronald Anderson. Gompers, P., J. Ishii, and A. Metrick, 2003. “Corporate Governance and equity prices,” Quarterly

Journal of Economics 118, 107-155. Grossman, S.J. and O.D. Hart, 1980. “Takeover Bids, the Free-Rider Problem, and the Theory of the

Corporation.” Bell Journal of Economics 11, 42-54. Hermalin, Benjamin E., and Michael S. Weisbach. 1998. "Endogenously Chosen Boards of Directors and

Their Monitoring of the CEO." American Economic Review 88:1, 96-118. Hermalin, Benjamin E., and Michael S. Weisbach. 2003. “Boards of Directors as an Endogenously

Determined Institution: A Survey of the Economic Literature” FRBNY Economic Policy Review 9, 7-26.

Hwang, B.-H., and S. Kim. 2009. “It Pays to Have Friends.” Journal of Financial Economics 93, 138-158. Hwang, B.-H., and S. Kim. 2011. “The Effect of Social Ties between the CEO and Audit-Committee

Members on Earnings Management.” Working Paper, Purdue University. Jensen, M. C., 1986. “Agency Costs of Free Cash Flow, Corporate Finance Takeovers.” American

Economic Review 76, 323-339 Jiraporn, Pornsit, Jang-Chul Kim, and Young Sang Kim, 2010. “Dividend Payouts and Corporate

Governance Quality: An Empirical Investigation.” The Financial Review, forthcoming.

28  

John, Kose and Anzhela Knyazeva. 2006. “Payout Policy, Agency Conflicts, and Corporate Governance.” Working Paper, New York University.

John, Kose, Anzhela Knyazeva, and Diana Knyazeva, 2011. “ Does Geography Matter? Firm Location

and corporate payout policy?” Journal of Financial Economics, forthcoming. Kandel, E., E. P. Lazear, 1992, “Peer Pressures and Partnerships.” Journal of Political Economy 100,

801-817. Knyazeva, Anzhela, Diana Knyazeva, and Charu Raheja. 2009. “The Benefits of Focus vs. Heterogeneity:

An Analysis of Corporate Boards.” Working Paper, University of Rochester and Wake Forest University.

Krishnaswami, Sudha, and Venkat Subramaniam, 1999, Information asymmetry, valuation, and the corporate spin-off decision, Journal of Financial Economics 53, 73-112.

Kroszner, Randall S., and Philip E. Strahan. 2001. “Bankers on Boards: Monitoring, Conflicts of Interest,

and Lender Liability.” Journal of Financial Economics 62:3, 415−452. Li, W. and E. Lie, 2006. “Dividend Changes and Catering Incentives.” Journal of Financial Economics

80, 293-308. Lloyd, P., and J. Jahera, and D. Page. 1985. “Agency Costs and Dividend-Payout Ratios.” Quarterly

Journal of Economics 24, 19-29. Masulis, Ronald W., and Shawn Mobbs. 2011. “Are All Insider Directors the Same? Evidence from the

external directorship market” Journal of Finance, forthcoming. Masulis, Ronald W, Cong Wang, and Fei Xie. 2010. “Globalizing the Boardroom – The Effects of

Foreign Directors on Corporate Governance and Firm Performance.” Working Paper, Vanderbilt University and Australian School of Business.

Rose, Casper. 2007. “Does Female Board Representation Influence Firm Performance? The Danish

Evidence.” Corporate Governance: An International Review 15:2, 404-413. Ross, S. 1977. “The Determination of Financial Structure: The Incentive-Signaling Approach,” Bell

Journal of Economics 1, 23– 40. Schmidt, B. 2009. “Costs and Benefits of “Friendly” Boards during Mergers and Acquisitions.” Working

Paper, Emory University. Shrader Charles, Virginia B. Blackburn, Paul Iles, 1997. “Women in Management and Firm Financial

Performance: An Exploratory Study.” Journal of Managerial Issues 9, 355-372. Tuhus-Dubrow, Rebecca. 2009. “The Female Advantage.” Boston Globe, May 3. Ideas, Section, 1. Vogt, C. 1994. “The Cash Flow/Investment Relationship: Evidence from U.S. Manufacturing Firms.”

Financial Management 23, 3-20. Zahra, S. A., and Stanton, W. W., 1988. “The Implications of Board of Directors Composition for

Corporate Strategy and Performance.” International Journal of Management 5, pp.229-236.

29  

Appendix. Variable Definitions

Variables Description

Firm characteristics Data source: Compustat, CRSP

LogTA Natural log of total assets

Leverage Total debt / total assets

R&D R&D expenses / total assets; if missing, recorded as zero Q (Book value of total assets + market value of equity- book value of equity)/ total assetsROA Net Income / total assets

RE_TE Retained earnings / total common equity

Cash_TA Cash and marketable securities / total assets

PPE_TA Net Property plant & equipment / total assets

STDRET Standard deviation of previous two years' daily stock returns

DIV_TA Cash dividends / total assets

DIV_NI Cash dividends / net income

DIV_P Cash dividends / stock price (fiscal year end) High-FCF One for firms with free cash flow greater than the sample median, zero otherwise P_institution The percentage of institutional investor holdings Segments Number of business segments Board and CEO related Data source: IRRC and Risk Metrics

Woman_dum One if at least one board member is female, zero otherwise

Minor_dum One if at least one board member is minority, zero otherwise

Num_woman Number of female board members

Pct_woman Percentage of female board members in the board relative to total directors Diverse_dum

One if at least one board member is woman or minority, zero otherwise. Minority includes African American, Asian, and Hispanic

Num_diverse Total number of female and minority directors in the board

Pct_diverse Percentage of female and minority directors relative to total directors

Num_minor Number of minority directors, not including woman

Pct_minor Percentage of minority directors, not including woman, relative to total directors

Diverse_CEO One if the CEO is female or minority, zero otherwise

Multi_diverse One if there are more than one female or minority members in the board, zero otherwise

Pct_indiverse Percentage of independent female and minority board members relative to total directors

Bsize Board size (number of directors) Pct_indep Percentage of independent board members

Outside_board Average number of outside board positions held by board members Pct_attend Average proportion of board members who attend meeting less than 75% Avgage Average age of board members Tenure Average tenure years of board members Chair_CEO One if the CEO is also the chairperson of the board, zero otherwise GINDEX Gompers, Ishii, and Metrics (2003) management entrenchment index CEO ownership Ratio of shares held by the CEO to common shares outstanding Dir_own Percentage of director ownership Diverse_Add One if the firm adds a diverse director to its board, zero otherwise

30  

Table 1. Sample Distribution

The sample consists of 13,325 firm-year observations that are available from Risk Metrics (former IRRC database) and Compustat/CRSP from 1997 to 2008. Panel A shows the number (#) of sample firms, the number of firms with diverse boards, the percentage of firms with diverse boards, and the percentage of woman and minority directors in board each year. Panel B shows the sample distribution across Fama and French 12 industries excluding utility and financial industries. Panel A. Sample Distribution by Year

Year # of Sample Firms

# of Firms with Diverse Board

% of Firms with Diverse Boards

% of Diverse Directors in the Board

1997 1,115 584 52.38 6.86

1998 1,243 626 50.36 8.02

1999 1,224 714 58.33 10.26

2000 1,239 743 59.97 10.83

2001 1,294 784 60.59 11.17

2002 1,071 703 65.64 12.49

2003 1,083 734 67.77 13.36

2004 1,087 757 69.64 13.80

2005 1,072 745 69.50 13.79

2006 1,035 760 73.43 14.79

2007 906 642 70.86 14.47

2008 956 696 72.80 15.12

Average 1,110 707 63.70 12.08

Total 13,325 8,488

31  

Panel B. Sample Distribution across Fama and French Industry Classification

F-F Industry

# of Sample Firms

# of Firms with Diverse Boards

% of Firms with Diverse

Boards

% of Diverse Directors in

the Board

Consumer Non Durables 1,103 907 82.23 17.88

Consumer Durables 433 271 62.58 10.99

Manufacturing 2,256 1,430 63.39 10.81

Oil, Gas, and Coal Extraction and Products

666 310 46.55 7.69

Chemicals and Allied Products

569 458 80.49 16.60

Business Equipment 2,843 1,420 49.95 9.13

Telephone and Television Transmission

362 263 72.65 14.84

Wholesale, Retail, and Some Service

1,945 1,431 73.57 14.68

Healthcare, Medical Equipment, and Drugs

1,301 850 65.33 11.44

Other 1,847 1,148 62.15 11.00

Total 13,325 8,488

32  

Table 2. Diverse Board Characteristics by Dividend Payers

The sample consists of 13,325 firm-year observations that are available from Risk Metrics (former IRRC database) and Compustat/CRSP from 1997 to 2008. The table shows the board characteristics and governance information of the sample firms divided into dividend payers and non-payers. Dividend Payers are firms with positive cash dividends and Dividend Non-Payers are firms without any cash dividend payouts. The mean and median difference test statistics are reported as t-stat and z-stat, respectively. Variable definitions are provided in the appendix.

Dividend Non-Payers Dividend Payers Difference Variable N Mean Median N Mean Median t-stat z-stat

Woman_dum 6218 0.4490 0.0000 7107 0.7116 1.0000 3.88*** 30.73***

Num_woman 6218 0.5917 0.0000 7107 1.0962 1.0000 33.44*** 33.38***

Pct_woman 6218 0.0687 0.0000 7107 0.1064 0.1000 24.44*** 24.56***

Diverse_dum 6218 0.5042 1.0000 7107 0.7532 1.0000 30.87*** 29.82***

Num_diverse 6218 0.7683 1.0000 7107 1.5115 1.0000 36.40*** 35.32***

Pct_diverse 6218 0.0893 0.0769 7107 0.1450 0.1250 28.00*** 27.70***

Minor_dum 6218 0.4868 0.0000 7107 0.7363 1.0000 30.62*** 29.60***

Pct_minor 6218 0.0206 0.0000 7107 0.0385 0.0000 16.78*** 21.10***

Diverse_CEO 6218 0.0376 0.0000 7107 0.0269 0.0000 3.53*** 3.53***

Multi_diverse 6218 0.1903 0.0000 7107 0.4317 0.0000 30.88*** 29.83***

Pct_indiverse 6190 0.0716 0.0000 7103 0.1265 0.1111 30.46*** 29.17***

Bsize 6218 7.9791 8.0000 7107 9.8548 10.000 48.26*** 46.33***

Pct_indep 6190 0.6443 0.6667 7103 0.6767 0.7059 10.63*** 11.44***

Outside_board 5808 1.7225 1.6364 6402 1.9651 1.8750 22.12*** 12.12***

Pct_attend 6218 0.0186 0.0000 7107 0.0172 0.0000 1.64 1.45

Avgage 6218 57.8988 58.200 7107 60.1438 60.2308 30.61*** 28.06***

Tenure 6218 10.5272 8.0000 7107 11.2478 8.0000 4.54*** 1.86*

Chair_CEO 6218 0.5849 1.0000 7107 0.6740 1.0000 10.68*** 10.64***

*** p<0.01, ** p<0.05, * p<0.1

33  

Table 3. Firm Characteristics by Board Diversity and Dividend Payout

The sample consists of 13,325 firm-year observations that are available from Risk Metrics (former IRRC database) and Compustat/CRSP from 1997 to 2008. Firms are classified as Diverse Board if they have at least one woman or minority director in their boards and Non-Diverse Board otherwise. Dividend Payers are firms with positive cash dividends and Dividend Non-Payers are firms without any cash dividend payouts. The mean and median difference test statistics are reported as t-stat and z-stat, respectively. Variable definitions are provided in the appendix.

Non-Diverse Board (Diverse_dum= 0)

Diverse Board (Diverse_dum= 1)

Difference

Variable N Mean Median N Mean Median t-stat z-stat

Dividend Payers

DIV_TA 1753 0.0179 0.0104 5353 0.0218 0.0153 4.52*** 13.12***

DIV_P 1753 0.0182 0.0109 5349 0.0219 0.0155 2.89*** 11.33***

DIV_NI 1754 0.1410 0.1473 5353 0.4774 0.2390 2.00** 12.43***

LogTA 1753 6.7784 6.6665 5353 8.0787 7.9295 33.58*** 32.08***

Leverage 1749 0.2142 0.2017 5344 0.2260 0.2186 2.75*** 4.39*** R&D 1753 0.0176 0.0000 5353 0.0184 0.0009 0.84 4.69***

Q 1752 1.8888 1.5224 5348 1.9662 1.5965 2.38** 4.69***

ROA 1753 0.0487 0.0575 5353 0.0562 0.0572 2.99*** 0.518

RE_TE 1754 0.5826 0.7072 5353 0.7785 0.8143 7.52*** 10.66*** Cash_TA 1753 0.1165 0.0452 5353 0.0838 0.0439 -10.05*** -2.55**

PPE_TA 1753 0.5518 0.4666 5353 0.5649 0.5095 1.36 2.69***

STDRET 1749 0.1125 0.1020 5327 0.0921 0.0848 -17.60*** -17.44***

Dividend Non-Payers

LogTA 3083 6.4139 6.3973 3135 7.1434 7.0290 24.23*** 22.30***

Leverage 3067 0.1854 0.1424 3124 0.1956 0.1547 2.10** 3.32***

R&D 3083 0.0574 0.0208 3135 0.0453 0.0060 -6.17*** -6.13***

Q 3076 2.2199 1.7061 3124 2.2263 1.6725 0.16 -0.36

ROA 3072 -0.0200 0.0393 3128 0.0191 0.0467 5.45*** 4.64***

RE_TE 3083 0.1450 0.3492 3135 0.2755 0.4282 4.05*** 6.55***

Cash_TA 3079 0.2122 0.1460 3132 0.1901 0.1306 -4.48*** -2.31**

PPE_TA 3083 0.3829 0.2917 3135 0.3961 0.3053 1.55 2.74***

STDRET 3079 0.1680 0.1498 3123 0.1436 0.1279 -11.82*** -13.17***

*** p<0.01, ** p<0.05, * p<0.1

34  

Table 4. Board Diversity and Dividend Payout

The sample consists of 13,325 firm-year observations that are available from Risk Metrics (former IRRC database) and Compustat/CRSP from 1997 to 2008. There are 4,837 firm-year observations with non-diverse boards and 8,488 firm-year observations with diverse boards. Panel A shows the association between the number of diverse directors in the firm’s board and average dividend payout measures. Panel B shows: the changes in dividend payouts for the same firms when their boards switch from non-diverse to diverse boards; the changes in dividend payouts for the same firms when their boards switch from diverse to non-diverse boards; and the differences in dividend payouts between firms whose boards remain diverse and non-diverse (No Change). Variable definitions are provided in the appendix.

Panel A. The Number of Diverse Directors and Payout Policy

# of Diverse Directors in the Board

# of firm-year observations

# of Dividend Payer

% of Dividend Payer

DIV/ TA (%)

DIV/ Price (%)

0 4,837 1,754 36.26 0.65 0.66

1 4,237 2,285 53.93 1.13 1.16

2 2,528 1,693 66.97 1.33 1.50

3 1,055 787 74.60 1.71 1.52

4 399 345 86.46 2.60 2.06

5 or more 269 243 90.33 2.58 2.18

Average 53.34 1.11 1.12

Panel B. Change in Board Diversity and Dividend Payout Policy

Non-Diverse to Diverse

Diverse to Non-Diverse

No Change

N Mean N Mean N Mean

DIV_dum Non-Diverse 1,296 39.81 193 41.97 3,158 35.24

(%) Diverse 1,847 50.57 262 45.80 6,240 68.17

Change/Difference 10.75*** 3.83 32.93***

DIV_TA Non-Diverse 1,296 0.59 193 0.65 3,150 0.70

(%) Diverse 1,846 0.84 262 0.98 6,234 1.56

Change/Difference 0.25*** 0.33 0.86***

DIV_P Non-Diverse 1,296 0.56 193 0.82 3,149 0.70

(%) Diverse 1,846 1.04 257 0.93 6,231 1.45

Change/Difference 0.48*** 0.11 0.75***

*** p<0.01, ** p<0.05, * p<0.1

35  

Table 5. The Likelihood of Dividend Payout with Logit Regression

The sample consists of 13,202 firm-year observations that are available from Risk Metrics (former IRRC database) and Compustat/CRSP with non-missing variables from 1997 to 2008. The dependent variable is dividend dummy (DIV_dum), which equals to 1 if a firm pays cash dividend, zero otherwise. Variable definitions are provided in the appendix. In the parentheses are the robust standard errors adjusted for firm clustering. (1) (2) (3) (4) (5) VARIABLES DIV_dum DIV_dum DIV_dum DIV_dum DIV_dum Pct_woman 2.092*** (0.548) Pct_diverse 2.033*** (0.429) Pct_minor 2.003*** (0.675) Multi_diverse 0.507*** (0.102) Diverse_dum 0.501*** (0.099) LogTA 0.356*** 0.333*** 0.366*** 0.326*** 0.335*** (0.042) (0.042) (0.042) (0.043) (0.042) Leverage -1.060*** -1.062*** -1.080*** -1.067*** -1.046*** (0.304) (0.303) (0.300) (0.304) (0.303) R&D -3.360** -3.561** -3.463** -3.493** -3.574** (1.498) (1.507) (1.513) (1.494) (1.515) Q -0.010 -0.009 -0.007 -0.010 -0.011 (0.039) (0.039) (0.038) (0.039) (0.039) ROA 0.261 0.258 0.236 0.248 0.281 (0.358) (0.357) (0.353) (0.357) (0.359) RE_TE 0.172*** 0.170*** 0.175*** 0.169*** 0.172*** (0.043) (0.042) (0.042) (0.042) (0.042) Cash_TA -0.617 -0.647 -0.614 -0.631 -0.571 (0.400) (0.398) (0.392) (0.398) (0.398) PPE_TA 0.778*** 0.767*** 0.789*** 0.778*** 0.764*** (0.163) (0.163) (0.162) (0.163) (0.163) STDRET -16.165*** -16.104*** -16.402*** -16.075*** -15.931*** (1.119) (1.114) (1.113) (1.115) (1.114)

Year YES YES YES YES YES

Industry YES YES YES YES YES Constant -0.080 0.073 0.061 0.183 -0.052 (0.376) (0.376) (0.379) (0.382) (0.373) Pseudo R2 0.2906 0.2920 0.2885 0.2921 0.2927 Observations 13,202 13,202 13,202 13,202 13,202

*** p<0.01, ** p<0.05, * p<0.1

36  

Table 6. Dividend Payout Ratio Regression

The sample consists of 13,201 firm-year observations that are available from Risk Metrics (former IRRC database) and Compustat/CRSP with non-missing variables from 1997 to 2008. This table shows estimation results of the ordinary least square (OLS) regression model. The dependent variable is the ratio of cash dividend to total assets (DIV_TA). Variable definitions are provided in the appendix. In the parentheses are the robust standard errors adjusted for firm clustering.

(1) (2) (3) (4) (5) VARIABLES DIV_TA DIV_TA DIV_TA DIV_TA DIV_TA

Pct_woman 0.013*** (0.004) Pct_diverse 0.014*** (0.003) Pct_minor 0.017*** (0.005) Multi_diverse 0.003*** (0.001) Diverse_dum 0.003*** (0.001) LogTA 0.001** 0.001 0.001** 0.001* 0.001* (0.000) (0.000) (0.000) (0.000) (0.000) Leverage -0.007** -0.007** -0.007** -0.007** -0.007** (0.003) (0.003) (0.003) (0.003) (0.003) R&D -0.017** -0.018** -0.018** -0.018** -0.017** (0.008) (0.008) (0.008) (0.008) (0.008) Q 0.003*** 0.003*** 0.003*** 0.003*** 0.003*** (0.001) (0.001) (0.001) (0.001) (0.001) ROA -0.002 -0.002 -0.002 -0.002 -0.002 (0.001) (0.001) (0.001) (0.001) (0.001) RE_TE 0.001** 0.001* 0.001** 0.001* 0.001** (0.000) (0.000) (0.000) (0.000) (0.000) Cash_TA 0.006 0.006 0.006 0.006 0.006* (0.004) (0.004) (0.004) (0.004) (0.004) PPE_TA 0.007*** 0.006*** 0.007*** 0.007*** 0.007*** (0.001) (0.001) (0.001) (0.001) (0.001) STDRET -0.058*** -0.057*** -0.059*** -0.058*** -0.057*** (0.012) (0.012) (0.012) (0.012) (0.012)

Year YES YES YES YES YES

Industry YES YES YES YES YES

Constant 0.012*** 0.013*** 0.014*** 0.014*** 0.012*** (0.004) (0.004) (0.004) (0.004) (0.004) Observations 13,201 13,201 13,201 13,201 13,201 R-squared 0.135 0.136 0.134 0.135 0.135

*** p<0.01, ** p<0.05, * p<0.1

37  

Table 7. The Effect of Additional Diverse Director on Dividend Payout Policy

The sample consists of 9,832 firm-year observations that are available from Risk Metrics (former IRRC database) and Compustat/CRSP with non-missing variables and information on the change in board diversity from 1997 to 2008. This table shows the effect of adding diverse board members on propensity to pay dividends and dividend payout. Models (1) to (3) show logit regression results and Models (4) to (6) show OLS regression results. Diverse_dum equals to one if at least one board member is woman or minority, zero otherwise. Dividend payout ratio is measured by DIV_TA (Dividends / Total assets). Diverse_add takes 1 if diverse board member is added on the board from Year (t-1) to Year (t), else 0. Variable definitions are provided in the appendix. In the parentheses are the robust standard errors adjusted for firm clustering.

(1) (2) (3) (4) (5) (6)

All sample Diverse_dum

(t-1)=0 Diverse_dum

(t-1)>=1 All sample

Diverse_dum (t-1)=0

Diverse_dum (t-1)>=1

DIV_dum DIV_dum DIV_dum DIV_TA DIV_TA DIV_TA

Diverse_dum (t-1) 0.534*** 0.007***

(0.110) (0.001)

Diverse_add 0.064 0.240** -0.047 0.001 0.003** -0.001 (0.075) (0.119) (0.098) (0.001) (0.002) (0.001)

LogTA 0.307*** 0.151** 0.367*** 0.001*** -0.001 0.002*** (0.045) (0.072) (0.055) (0.001) (0.001) (0.001)

Leverage -0.790** -0.537 -0.870* -0.004 -0.010 -0.001 (0.335) (0.449) (0.448) (0.006) (0.007) (0.007)

R&D -3.381* -5.739** -2.532 -0.044* -0.075* -0.041 (1.726) (2.794) (2.047) (0.024) (0.040) (0.029)

Q -0.031 -0.022 -0.049 0.004*** 0.002* 0.005*** (0.044) (0.064) (0.054) (0.001) (0.001) (0.001)

ROA 0.614 0.280 1.176* 0.019** 0.005 0.031** (0.453) (0.575) (0.642) (0.009) (0.010) (0.013)

RE_TE 0.169*** 0.118 0.184*** 0.002*** 0.001 0.002*** (0.046) (0.075) (0.056) (0.001) (0.001) (0.001)

Cash_TA -0.446 0.118 -1.061* 0.005 0.016* -0.008 (0.438) (0.582) (0.553) (0.007) (0.010) (0.008)

PPE_TA 0.767*** 0.627** 0.878*** 0.012*** 0.014*** 0.011*** (0.181) (0.253) (0.234) (0.002) (0.005) (0.002)

STDRET -16.186*** -13.629*** -17.571*** -0.241*** -0.239*** -0.228*** (1.212) (1.613) (1.628) (0.022) (0.037) (0.023)

Year YES YES YES YES YES YES Industry YES YES YES YES YES YES

N 9832 3795 6037 9831 3795 6036 *** p<0.01, ** p<0.05, * p<0.1

38  

Table 8. The Effect of Woman/ Minority CEO

The sample consists of 13,201 firm-year observations that are available from Risk Metrics (former IRRC database) and Compustat/CRSP with non-missing variables from 1997 to 2008. This table shows the Logit and OLS regression model. For Models (1) to (3), the dependent variable is dividend dummy (DIV_dum). For Models (4) to (6), the dependent variable is payout ratio (DIV_TA). Diverse_CEO equals to one if the CEO is female or minority, zero otherwise. Diverse_dum equals to one if a firm has at least one board member is woman or minority, zero otherwise. Minority includes African American, Asian, and Hispanic. Models (2) and (5) show the regression results of Diverse_CEO = 1 while Models (3) and (6) show the regression results of Diverse_CEO = 0. Variable definitions are provided in the appendix. In the parentheses are the robust standard errors adjusted for firm clustering.

(1) (2) (3) (4) (5) (6) VARIABLES DIV_dum DIV_dum DIV_dum DIV_TA DIV_TA DIV_TA All sample Diverse_CEO

= 1 Diverse_CEO

= 0 All sample Diverse_CEO

= 1 Diverse_CEO

= 0 Diverse_CEO -0.143 -0.001 (0.261) (0.002) Diverse_dum 0.383*** -1.399* 0.407*** 0.003*** 0.002 0.003*** (0.111) (0.727) (0.111) (0.001) (0.003) (0.001) LogTA 0.351*** 0.758*** 0.347*** 0.001* 0.002** 0.001 (0.042) (0.244) (0.043) (0.000) (0.001) (0.000) Leverage -1.084*** 2.578 -1.132*** -0.007** 0.006 -0.007** (0.300) (1.648) (0.304) (0.003) (0.008) (0.003) R&D -3.584** 6.258 -4.070*** -0.019** 0.024 -0.020** (1.521) (6.287) (1.526) (0.008) (0.018) (0.008) Q -0.008 0.011 -0.009 0.003*** 0.002* 0.003*** (0.038) (0.133) (0.039) (0.001) (0.001) (0.001) ROA 0.244 3.754* 0.149 -0.002 0.002 -0.002 (0.354) (2.044) (0.348) (0.001) (0.006) (0.001) RE_TE 0.175*** 0.252 0.176*** 0.001** -0.000 0.001** (0.042) (0.236) (0.043) (0.000) (0.001) (0.000) Cash_TA -0.596 -0.292 -0.558 0.006 0.017 0.006 (0.393) (2.233) (0.397) (0.004) (0.012) (0.004) PPE_TA 0.782*** 2.393** 0.778*** 0.007*** 0.010 0.007*** (0.162) (1.063) (0.162) (0.001) (0.006) (0.001) STDRET -16.273*** -17.090*** -16.458*** -0.058*** -0.076*** -0.058*** (1.112) (6.066) (1.128) (0.012) (0.021) (0.012)

Year YES YES YES YES YES YES

Industry YES YES YES YES YES YES

Constant 0.156 -1.149 0.194 0.015*** 0.010 0.015*** (0.383) (2.213) (0.386) (0.004) (0.009) (0.004) Observations 13,202 269 12,906 13,201 269 12,905 R-squared 0.135 0.475 0.133

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

39  

Table 9. Board Diversity, Free Cash Flow and the Likelihood of Dividend Payout

The sample consists of 13,202 firm-year observations that are available from Risk Metrics (former IRRC database) and Compustat/CRSP with non-missing variables from 1997 to 2008. This table shows the logit regression model with agency interaction variables. The dependent variable is Dividend dummy (DIV_dum), which equals to 1 if a firm paid common cash dividend, zero otherwise. Diverse_dum equals to one if at least one board member is woman or minority, zero otherwise. Minority includes African American, Asian, and Hispanic. High-FCF equals to 1 if a firm’s free cash flow is above the sample mean, zero otherwise. FCF*Diverse is the interaction variable of High-FCF multiplied by Diverse_dum. Model (1) includes all sample firms. In Model (2) and (3), firms are divided into high GINDEX if GINDEX is greater than 9 and low GINDEX otherwise (Gompers, Ishii, and Metrick, 2003). In Model (4) and (5), firms are divided into low CEO ownership if the proportion of CEO ownership is less than the sample mean and high CEO ownership otherwise. Variable definitions are provided in the appendix. In the parentheses are the robust standard errors adjusted for firm clustering.

(1) (2) (3) (4) (5) VARIABLES DIV_dum DIV_dum DIV_dum DIV_dum DIV_dum All Sample High GINDEX Low GINDEX Low CEO

Ownership High CEO Ownership

Diverse_dum 0.281** 0.310 0.172 0.241* 0.393* (0.127) (0.222) (0.143) (0.145) (0.215) High-FCF -0.208* -0.576** -0.051 -0.291** -0.023 (0.126) (0.235) (0.136) (0.143) (0.208) FCF*Diverse 0.363** 0.703*** 0.232 0.497*** 0.047 (0.144) (0.254) (0.163) (0.166) (0.243) LogTA 0.348*** 0.410*** 0.311*** 0.424*** 0.087 (0.042) (0.073) (0.048) (0.047) (0.082) Leverage -0.983*** 0.144 -1.487*** -0.700* -1.441*** (0.306) (0.583) (0.339) (0.359) (0.486) R&D -3.465** -0.840 -4.912*** -3.436** -4.228* (1.489) (2.745) (1.667) (1.727) (2.517) Q -0.073* -0.027 -0.077* -0.064 -0.109 (0.042) (0.086) (0.047) (0.048) (0.075) ROA -0.247 0.139 -0.383 -0.048 -0.804** (0.285) (0.731) (0.265) (0.367) (0.396) RE_TE 0.163*** 0.179*** 0.145*** 0.182*** 0.098 (0.042) (0.064) (0.053) (0.047) (0.086) Cash_TA -0.600 -1.186* -0.238 -0.714 -0.307 (0.400) (0.683) (0.437) (0.495) (0.562) PPE_TA 0.786*** 0.729*** 0.816*** 0.768*** 0.839*** (0.164) (0.273) (0.190) (0.183) (0.287) STDRET -15.694*** -14.650*** -15.446*** -15.182*** -16.255*** (1.113) (1.889) (1.281) (1.279) (1.971) Year YES YES YES YES YES

Industry YES YES YES YES YES Constant -0.087 -1.305* 0.298 -0.533 1.407** (0.382) (0.696) (0.433) (0.451) (0.658) Observations 13,202 4,680 8,522 10,360 2,842 *** p<0.01, ** p<0.05, * p<0.1

40  

Table 10. Board Diversity, Free Cash Flow and Payout Ratio

The sample consists of 13,201 firm-year observations that are available from Risk Metrics (former IRRC database) and Compustat/CRSP with non-missing variables from 1997 to 2008. This table shows the ordinary least square (OLS) regression model with agency interaction variables. The dependent variable is the ratio of dividend to total assets (DIV_TA). Diverse_dum equals to one if at least one board member is woman or minority, zero otherwise. Minority includes African American, Asian, and Hispanic. High-FCF equals to 1 if a firm’s free cash flow is above the sample mean, zero otherwise. FCF*Diverse is the interaction variable of High-FCF multiplied by Diverse_dum. Model (1) includes all sample firms. In Model (2) and (3), firms are divided into high GINDEX if GINDEX is greater than 9 and low GINDEX otherwise (Gompers, Ishii, and Metrick, 2003). In Model (4) and (5), firms are divided into low CEO ownership if the proportion of CEO ownership is less than the sample mean and high CEO ownership otherwise. Variable definitions are provided in the appendix. In the parentheses are the robust standard errors adjusted for firm clustering.

(1) (2) (3) (4) (5) VARIABLES DIV_TA DIV_TA DIV_TA DIV_TA DIV_TA All Sample High

GINDEX Low GINDEX

Low CEO Ownership

High CEO Ownership

Diverse_dum -0.001* 0.000 -0.002** -0.001 -0.002 (0.001) (0.001) (0.001) (0.001) (0.002) High-FCF -0.002*** -0.005** -0.002* -0.002** -0.002 (0.001) (0.002) (0.001) (0.001) (0.002) FCF*Diverse 0.007*** 0.006*** 0.006*** 0.006*** 0.007*** (0.001) (0.002) (0.001) (0.001) (0.002) LogTA 0.001* 0.000 0.001 0.001 -0.000 (0.000) (0.001) (0.000) (0.000) (0.000) Leverage -0.007** 0.010 -0.012*** -0.003 -0.017*** (0.003) (0.008) (0.002) (0.004) (0.003) R&D -0.018** 0.002 -0.024** -0.014* -0.044* (0.008) (0.017) (0.009) (0.007) (0.026) Q 0.003*** 0.006*** 0.002*** 0.004*** 0.000 (0.001) (0.001) (0.000) (0.001) (0.001) ROA -0.002 0.023** -0.004*** -0.001 -0.002 (0.001) (0.009) (0.001) (0.002) (0.002) RE_TE 0.001* 0.001*** 0.001 0.001*** -0.001 (0.000) (0.000) (0.001) (0.000) (0.002) Cash_TA 0.007* 0.002 0.009** 0.001 0.020** (0.004) (0.005) (0.005) (0.003) (0.010) PPE_TA 0.006*** 0.007*** 0.006*** 0.005*** 0.008*** (0.001) (0.002) (0.002) (0.001) (0.002) STDRET -0.056*** -0.070*** -0.048*** -0.065*** -0.038* (0.012) (0.010) (0.013) (0.007) (0.020) Year YES YES YES YES YES

Industry YES YES YES YES YES Constant 0.003 -0.002 0.007 0.003 0.013** (0.003) (0.005) (0.004) (0.004) (0.005)

Observations 13,201 4,679 8,522 10,359 2,842

R-squared 0.140 0.224 0.118 0.188 0.086 *** p<0.01, ** p<0.05, * p<0.1

41  

Table 11. Differential Impacts of Board Diversity on Dividend Policy for Matched Firms

The sample includes propensity score-Mahalanobis distance matched 342 pairs of firms or 5,476 firm-year observations in Risk Metrics (former IRRC database) and Compustat/CRSP from 1997 to 2008. Pre-treatment is the period prior to the matching and post-treatment is the period after the matching. The treatment group consists of firms that added a new diverse director to their boards for the first time. The control group consists of matched firms whose boards remain non-diverse. In Panel B, Treatment is equal to one for a firm that adds a diverse director for the first time during the 1997-2008 sample period and zero otherwise. Post_Treat is equal to one for the period after the firm added a new diverse director to its board. The interaction (Treat_effect) measures the treatment effect of board diversity. In Models (1) and (2), the dependent variable is Dividend dummy(DIV_dum). In Models (3) and (4), the dependent variable is DIV_TA. Diverse_dum equals to one if a firm has at least one board member is woman or minority, zero otherwise. Minority includes African American, Asian, and Hispanic. Variable definitions are provided in the appendix. In the parentheses are the robust standard errors adjusted for firm clustering. Panel A. The Treatment Effect (Adding a Diverse Director) on Dividend Payout Policy

Observations

DIV_dum (%)

DIV_TA (%)

DIV_P (%)

Control Group

Treatment Group

Control Group

Treatment Group

Control Group

Treatment Group

Control Group

Treatment Group

Pre-Treatment

1143 1124 58.97 40.21 1.43 0.60 1.14 0.58

Post-Treatment

1583 1165 60.33 50.99 1.33 0.86 1.24 1.08

Changes 1.36 10.78*** 0.10 0.25*** 0.10 0.50***

Difference-in-Difference

9.42*** 0.16*** 0.40***

*** p<0.01, ** p<0.05, * p<0.1

42  

Panel B. Difference-in-Difference Regressions

(1) (2) (3) (4) VARIABLES DIV_dum DIV_dum DIV_TA DIV_TA Treatment -0.574*** -0.601*** -0.006*** -0.007*** (0.104) (0.099) (0.001) (0.001) Post_Treat -0.062 -0.282*** -0.002* -0.001 (0.106) (0.094) (0.001) (0.001) Treat_effect 0.287** 0.295** 0.002** 0.003** (0.134) (0.129) (0.001) (0.001) LogTA 0.355*** 0.315*** 0.001*** 0.001*** (0.028) (0.027) (0.000) (0.000) Leverage -1.409*** -1.073*** -0.003 -0.001 (0.243) (0.236) (0.003) (0.003) R&D -4.548*** -6.120*** -0.021*** -0.029*** (1.168) (0.924) (0.005) (0.005) Q -0.058* -0.082** 0.003*** 0.003*** (0.035) (0.033) (0.000) (0.000) ROA 0.075 0.446 -0.001 -0.002 (0.540) (0.539) (0.002) (0.002) RE_TE 0.187*** 0.229*** 0.001*** 0.002*** (0.047) (0.048) (0.000) (0.000) Cash_TA -0.379 -1.006*** 0.006** 0.003 (0.283) (0.261) (0.003) (0.003) PPE_TA 0.808*** 0.765*** 0.005*** 0.004*** (0.114) (0.108) (0.001) (0.001) STDRET -17.402*** -14.593*** -0.041*** -0.056*** (1.108) (0.842) (0.011) (0.011) YEAR YES NO YES NO INDUSTRY YES NO YES NO Constant 0.163 0.144 0.000 0.005** (0.292) (0.248) (0.002) (0.002) Observations 5,476 5,476 5,475 5,475 R-squared 0.188 0.130

*** p<0.01, ** p<0.05, * p<0.1

43  

Table 12. The Effect of Board Diversity for Firms Matched by Propensity to Pay Dividends

The sample consists of propensity score-Mahalanobis distance matched 7,968 firm-year observations in Risk Metrics (former IRRC database) and Compustat/CRSP from 1997 to 2008. We use dividend propensity score with Caliper method to find the matching sample by firm and year. Model 1, 2, and 3 the dependent variable is Dividend dummy(DIV_dum). Model 4-6 the dependent variable is DIV_TA. Diverse_dum equals to one if a firm has at least one board member is woman or minority, zero otherwise. Pct_diverse is the percentage of diverse board members. Minority includes African American, Asian, and Hispanic. High-FCF equals to 1 if a firm’s free cash flow is above the sample mean, zero otherwise. FCF*Diverse is the interaction variable of High-FCF multiplied by Diverse_dum. Variable definitions are provided the appendix. In the parentheses are the robust standard errors adjusted for firm clustering.

(1) (2) (3) (4) (5) (6) VARIABLES DIV_dum DIV_dum DIV_dum DIV_TA DIV_TA DIV_TA

Diverse_dum 0.482*** 0.268* 0.002** -0.002 (0.109) (0.143) (0.001) (0.001) Pct_Diverse 2.051*** 0.012*** (0.457) (0.003) FCF*Diverse 0.361** 0.006*** (0.159) (0.001) FCF -0.189 -0.002** (0.139) (0.001) LogTA 0.196*** 0.190*** 0.198*** 0.000 -0.000 0.000 (0.044) (0.044) (0.044) (0.000) (0.000) (0.000) Leverage -0.416 -0.429 -0.435 -0.001 -0.001 -0.002 (0.324) (0.323) (0.324) (0.003) (0.003) (0.003) R&D -2.682 -2.654 -2.801* -0.019 -0.019 -0.021 (1.638) (1.626) (1.642) (0.015) (0.015) (0.015) Q 0.056 0.058 0.048 0.003*** 0.003*** 0.003*** (0.043) (0.043) (0.045) (0.001) (0.001) (0.001) ROA -0.219 -0.262 -0.253 0.004 0.004 0.003 (0.373) (0.372) (0.379) (0.003) (0.003) (0.003) RE_TE 0.152*** 0.150*** 0.149*** 0.000 0.000 0.000 (0.043) (0.043) (0.043) (0.001) (0.001) (0.001) Cash_TA 0.288 0.216 0.275 0.017** 0.017** 0.017** (0.438) (0.438) (0.436) (0.007) (0.007) (0.007) PPE_TA 0.406** 0.401** 0.391** 0.005*** 0.005*** 0.004*** (0.168) (0.168) (0.170) (0.001) (0.001) (0.002) STDRET -5.489*** -5.570*** -5.389*** -0.079*** -0.079*** -0.077*** (1.189) (1.190) (1.186) (0.011) (0.010) (0.010)

Year YES YES YES YES YES YES Industry YES YES YES YES YES YES Constant -0.822** -0.682* -0.701* 0.014*** 0.015*** 0.016*** (0.408) (0.409) (0.414) (0.004) (0.004) (0.004) Observations 7,968 7,968 7,968 7,967 7,967 7,967 R-squared 0.099 0.101 0.103

*** p<0.01, ** p<0.05, * p<0.1

44  

Table 13. Dividend Payout Policy with Institutional Holdings The sample consists of 13,201 firm-year observations that are available from Risk Metrics (former IRRC database) and Compustat/CRSP with non-missing variables from 1997 to 2008. This table shows the Logit and OLS regression results. In Models (1) to (3), the dependent variable is Dividend dummy (DIV_dum). In Models (4) to (6), the dependent variable is DIV_TA. Diverse_dum equals to one if a firm has at least one board member is woman or minority, zero otherwise. Minority includes African American, Asian, and Hispanic. P_institution is the percentage of institutional investor holdings. High_insti is a dummy variable, which equals to 1 if institutional investor holding is higher than the sample average, and zero otherwise. Models (2) and (5) show the regression results for High_insti= 1 while Models (3) and (6) show the regression results for High_insti = 0. Variable definitions are provided in the appendix. In the parentheses are the robust standard errors adjusted for firm clustering.

(1) (2) (3) (4) (5) (6) VARIABLES DIV_dum DIV_dum DIV_dum DIV_TA DIV_TA DIV_TA All sample High_insti = 1 High_insti = 0 All sample High_insti = 1 High_insti = 0

Diverse_dum 0.492*** 0.588*** 0.433*** 0.003*** 0.003*** 0.002*** (0.102) (0.131) (0.116) (0.001) (0.001) (0.001) P_institution -1.800*** -0.005*** (0.267) (0.001) LogTA 0.349*** 0.467*** 0.263*** 0.001* 0.000 0.001 (0.045) (0.061) (0.046) (0.000) (0.000) (0.000) Leverage -0.934*** -0.401 -1.299*** -0.006** 0.005 -0.013*** (0.328) (0.414) (0.347) (0.003) (0.006) (0.002) R&D -3.925** -3.770* -3.485** -0.018** -0.010 -0.021** (1.701) (2.142) (1.738) (0.008) (0.010) (0.009) Q -0.013 -0.054 0.036 0.003*** 0.002*** 0.004*** (0.042) (0.054) (0.047) (0.001) (0.001) (0.001) ROA 0.783 0.995 0.176 -0.001 0.002 -0.002* (0.512) (0.729) (0.383) (0.001) (0.003) (0.001) RE_TE 0.159*** 0.195*** 0.146*** 0.001** 0.001*** 0.001 (0.046) (0.068) (0.045) (0.000) (0.000) (0.001) Cash_TA -0.568 0.245 -1.255*** 0.007* 0.010** 0.004 (0.419) (0.528) (0.443) (0.004) (0.004) (0.005) PPE_TA 0.706*** 0.667*** 0.788*** 0.006*** 0.007*** 0.006*** (0.170) (0.216) (0.186) (0.001) (0.002) (0.002) STDRET -16.394*** -15.006*** -16.782*** -0.059*** -0.063*** -0.055*** (1.175) (1.671) (1.284) (0.012) (0.009) (0.016) Year YES YES YES YES YES YES Industry YES YES YES YES YES YES Constant 0.252 -2.269*** 0.006 0.002 -0.003 0.003 (0.407) (0.549) (0.407) (0.004) (0.004) (0.004) Observations 11,441 5,823 7,362 13,184 5,822 7,362 R-squared 0.142 0.111 0.163

*** p<0.01, ** p<0.05, * p<0.1

45  

Table 14. Robustness Test: 2SLS –IV Regression

The sample consists of firm-year observations that are available from Risk Metrics (former IRRC database) and Compustat/CRSP with non-missing instrument variables from 1997 to 2008. This table shows the result for two stage instrumental variable method (2SLS-IV). The dependent variable is DIV_TA. Diverse_dum equals to one if at least one board member is woman or minority, zero otherwise. Minority includes African American, Asian, and Hispanic. In first stage, we use instrumental variables: BSIZE, Pct_indep, Promotion, and Contracting. Promotion (DIV-str-B) is an indicator variable that equals one for a company that has made notable progress in the promotion of women and minorities, particularly to line positions with profit-and-loss responsibilities in the corporation, zero otherwise. Contracting (DIV-str-E) is in indicator variable that equals one for a company that does at least 5% of its subcontracting, or otherwise has a demonstrably strong record on purchasing or contracting, with women and/or minority-owned businesses, zero otherwise. Model (2) and (3) include only dividend payers. Variable definitions are provided in the appendix. In the parentheses are the robust standard errors adjusted for firm clustering.

(1) (2) (3) VARIABLES First

Stage DIV_TA First

Stage DIV_TA First

Stage DIV_TA

Diverse_dum 0.019*** 0.025*** (0.004) (0.007) Pct_diverse 0.050** (0.019) LogTA 0.041*** -0.001* 0.034*** -0.003*** 0.024*** -0.003*** (0.004) (0.001) (0.005) (0.001) (0.001) (0.001) Leverage -0.007 -0.002 -0.026 0.010 0.023* 0.008 (0.036) (0.004) (0.044) (0.008) (0.014) (0.008) R&D 0.516*** -0.023* 1.003*** -0.034 0.238*** -0.019 (0.131) (0.013) (0.210) (0.035) (0.065) (0.032) Q -0.002 0.003*** -0.008 0.006*** -0.000 0.006*** (0.004) (0.001) (0.005) (0.001) (0.002) (0.001) ROA 0.022 -0.004 -0.058 0.028** 0.025 0.025** (0.035) (0.002) (0.097) (0.013) (0.030) (0.013) RE_TE 0.009* 0.001** -0.000 0.001 0.004* 0.000 (0.005) (0.001) (0.007) (0.001) (0.002) (0.001) Cash_TA -0.027 0.008* 0.132** 0.032*** 0.023 0.024*** (0.041) (0.004) (0.056) (0.009) (0.017) (0.008) PPETA 0.030* 0.007*** 0.024 0.008*** 0.016*** 0.008*** (0.016) (0.002) (0.017) (0.003) (0.005) (0.002) STDRET -0.913*** -0.068*** -0.820*** -0.096*** -0.247*** -0.108*** (0.113) (0.011) (0.179) (0.021) (0.055) (0.020) Bsize 0.050***

(0.002) 0.037***

(0.003) -0.000

(0.000)

Pct_indep 0.348*** 0.372*** 0.126*** (0.032) (0.035) (0.012) Promotion 0.094*** 0.061*** 0.051*** (0.010) (0.012) (0.003) Contracting -0.036* -0.027*** 0.042*** (0.021) (0.020) (0.006) Year YES YES YES YES YES YES

46  

Industry YES YES YES YES YES YES Constant -0.288*** 0.006 -0.070*** 0.009 -0.208*** 0.025*** (0.050) (0.004) (0.024) (0.006) (0.016) (0.006) Observations 6606 6,606 3,931 3,931 3,921 3,928 R-squared 0.2675 0.079 0.2378 0.095 0.3355 0.147

*** p<0.01, ** p<0.05, * p<0.1

47  

Figure 1. Board Diversity and Dividend Payouts from 1997 to 2008 New-Diverse indicate firms that add a new diverse director between 2001 and 2004. Diverse and Non-Diverse are firms whose boards remain diverse and non-diverse, respectively, throughout the sample period.

 

 

 

           

0.0%

20.0%

40.0%

60.0%

80.0%

100.0%

97 98 99 00 01 02 03 04 05 06 07 08

Panel A. Proportion of Firms Paying Dividend

Non‐Diverse Diverse New‐Diverse

0.00%

0.20%

0.40%

0.60%

0.80%

1.00%

1.20%

1.40%

1.60%

1.80%

97 98 99 00 01 02 03 04 05 06 07 08

Panel B. Dividend/Total Asset

Non‐Diverse Diverse New‐Diverse

0.00%

0.20%

0.40%

0.60%

0.80%

1.00%

1.20%

1.40%

1.60%

97 98 99 00 01 02 03 04 05 06 07 08

Panel C. Dividend/Price

Non‐Diverse Diverse New‐Diverse