from symbolism to integration: new evidence on female ... · the time-covention of the measure of...

30
From Symbolism to Integration: New Evidence on Female Directors and Firm Performance Colin P. Green Lancaster University Swarnodeep Homroy* Lancaster University June 15, 2015 Abstract There is an increasing focus on enhancing female represenation on corporate boards. Existing evidence nds little e/ect of board gender diversity on rm per- formance. We return to this issue using Europe-wide data on listed rms, where the key advantage is a wider variation in levels of female representation not only on boards, but also on key governance committees. There is a positive impact of gender diversity on corporate boards on rm performance. This positive association is particularly strong for rms with higher female representation on key committees. Our results suggest that performance-enhancing e/ect of integrating the female rep- resentation on boards, over and above the symbolic tokenism. Key Words: G30, G34, J16 JEL Codes: Board of Directors, Female Director, Diversity, Performance *Corresponding Author: [email protected] 1

Upload: others

Post on 29-Aug-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

From Symbolism to Integration: New Evidence on

Female Directors and Firm Performance

Colin P. Green

Lancaster University

Swarnodeep Homroy*

Lancaster University

June 15, 2015

Abstract

There is an increasing focus on enhancing female represenation on corporate

boards. Existing evidence finds little effect of board gender diversity on firm per-

formance. We return to this issue using Europe-wide data on listed firms, where

the key advantage is a wider variation in levels of female representation not only

on boards, but also on key governance committees. There is a positive impact of

gender diversity on corporate boards on firm performance. This positive association

is particularly strong for firms with higher female representation on key committees.

Our results suggest that performance-enhancing effect of integrating the female rep-

resentation on boards, over and above the symbolic tokenism.

Key Words: G30, G34, J16

JEL Codes: Board of Directors, Female Director, Diversity, Performance

*Corresponding Author: [email protected]

1

Page 2: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

1 1. Introduction

Women are underrepresented on corporate boards. For example, while women make up

over 40% of the workforce in the UK, US and EU, membership on corporate boards ranges

from 8% to 14.8% across these jurisdictions (Equal Opportunity for Women in the Work-

place agency, 2010). Moreover, while female workforce participation and employment has

grown markedly, there has been only a sluggish change in female representation on boards.

This ongoing under-representation of women on boards, relative to their male counter-

parts, calls into question the effi ciency and equity of the board recruitment process. The

lack of gender diversity in boardrooms can have domino effect on the gender-diversity

in the workforce. Board gender diversity is a salient issue for firms and policy makers.

Apart from the obvious economic ineffi ciencies associated with labour market discrim-

ination (Becker, 1957), it is now perceived as an important criterion for institutional

investment and listings by such socially responsible indices as the FTSE4Good or Domini

400. Publicly listed firms in some European countries are mandated to have at least

40% representation of women on the board. The European Union is examining the pos-

sibility and implications of similar legislations. The next wave of governance regulations

in the EU member states is likely to address the issue of mandated gender diversity in

boardrooms for all listed firms by the year 2020 (European Union, 2012).

It is unclear if female representation on boards enhance corporate governance outcomes

and firm performance.(Ahern and Dittmar, 2012; Strovik and Tiegen, 2010; Nielsen and

Huse, 2010). Apart from the standpoints of equity and moral justice the rationale of fe-

male representation of corporate boards is not well understood. Sharder et al (1997) find

no significant association between board gender diversity and firm performance. A recent

study on US firms finds a negative impact of gender diversity on firm performance, despite

better attendance records and more effective monitoring in firms with more gender equi-

2

Page 3: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

table boards (Adams and Ferreira, 2009). Gregory-Smith et al. (2014) find no evidence

in favour of the argument that board gender diversity enhances firm performance. While

the economic implications of board gender diversity are ambiguous, decisions to increase

female representation on boards may be the result of social and political pressures. The

meagre evidence on the impact on firm performance of gender-diverse boards are based on

samples of firms with ’token’female representation of either a single individual or a couple

of individual directors on the board. Not surprisingly, these papers report no significant

impact of such diversity on firm performance (O’Reilly and Main, 2012).

This paper is different in estimating the economic impact of the diversity on corporate

boards, over and the above the symbolic tokenism. We do this in two ways. First we

employ a sample of large European firms with a wider variation in the female representa-

tion on corporate boards. Second, in addition to the traditional measure of board gender

diversity, i.e., the fraction of female directors, we use the fraction of female directors on

key governance committees. This allows us to examine the impact of female directors on

firm performance, when they are in a position to do so.

The central finding of this paper is that female representation on corporate boards

enhances firm performance. This association is of a higher order of magnitude where

there is a high proportion of women appointed on key corporate governance committees.

The results add to the literature by providing evidence that higher degree of integra-

tion of female directors through committee appointments is value-enhancing. Our results

provides evidence of impact on firm performance of female representation on firm perfor-

mance, over and above the token-representation. The impact of female directors, if any, is

more likely to be manifest in enhanced firm performance when female directors are placed

in key decision-making committees. This paper also reconciles the existing evidence by

comparing the results for UK and non-UK European firms.

3

Page 4: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

The article is structured as follows: Section 2 reviews the literature on the gender

composition of corporate boards, Section 3 introduces the sample and the estimation

methods employed for the analysis; Section 4 presents the results and Section 5 concludes.

2 2. Gender Diversity on Corporate Boards

Arguments in favour of increased representation of women on corporate boards tradition-

ally stem from concerns of equity and moral justice. However, a more gender diverse board

may also lead to productivity gains through improved decision making, displacement of

less able male directors and more effi cient monitoring (see Hermalin and Weisbach, 2003;

Adams et al. 2007). The productivity gain can be manifest in the form of better at-

tendance of directors on the board, performance sensitive CEO pay and CEO turnover

and generally better corporate governance (Adams and Ferreira, 2009). In addition to

the case for more female representation on corporate boards from an equity standpoint,

it is therefore important to understand if increased female representation also leads to

productivity gains.

The apparent incongruence of the female representation on boards and female rep-

resentation in the labour force can be due to supply side factors, discrimination, or a

combination of both. Powell and Butterfield (1994) argue that discriminatory practices

hinder the career progression of women to corporate boards. In contrast Mincer and

Polachek (1974) propose that family formation decisions result in a restricted supply of

qualified female candidates for board positions. Farrell and Hersch (2005) examine the

appointment of new directors on boards and finds that the proportion of female appoint-

ments on boards is significantly higher if the immediate predeccessor was a female director.

This highlights that the board appointment processes may not be gender neutral. Hence,

4

Page 5: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

the small proportion of female directors persists over a period of time.

If lack of gender-neautrality in female board appointments is discriminatory in na-

ture, then firms engaging in such practices are likely to face a competitive disadvantage.

Empirical evidence suggests that board composition has no significant effect on firm per-

formance and that the average effect of board gender diversity on firm performance can

be negative (Larcker, et al. 2007; Adams and Ferreira, 2009; Ahern and Dittmar, 2012;

Gregory-Smith et al. 2014). However, these results are either based on boards with only

one female director or following an enforcement of mandatory female representation. Thus

these papers may capture the nature of tokenism in female board representation rather

than the causal link of the impact of female representation on firm performance.

This analysis is based on a sample of firms with much wider dispersion of female

representation on boards. In this way, we are able to better examine the impact of female

directors on firm performance. Using data on committee assignments and the network

size of directors, we are also able to examine a possible transmission mechanism for the

impact of female board representation on firm performance.

3 3. Data

3.1 3.1 Data Source

The primary database used in the analysis is BoardEx, which provides information on

board composition and director networks for listed European firms. We use a sample of

EuroTop 100 firms for the period 2006-2013. EuroTop 100 are large firms listed in any of

stock exchanges of the European Union. Firms that appear at least once in the EuroTop

100 are followed till the end of the sample period as long as they remain listed. We use

5

Page 6: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

information on individual directors on the boards of these firms. We drop observations on

individual directors whom we observe in only one period in a given firm. We augment this

database with a range of financial performance metrics using Datastream/Worldscope.

Firms where financial performance metrics were unavailable were excluded. The final

sample consists of an unbalanced panel of 118 firms with 16,647 director-year observations.

We note the directors’gender and role classifications from BoardEx. Table 1 presents

descriptive statistics for selected firm, board and individual director characteristics. Table

2 summarizes key variable for firms with at least one female director and firms with

no female directors. All monetary variables are converted to year 2003 dollars using

Harmonized Indices of Consumer Prices published by the European Central Bank.

About 35% of our sample firms are from the UK. Therefore, it is only natural to

differentiate between the sub-sample of UK-firms, and the sub-sample of non-UK firms.

This is particularly important to position our findings with respect to the evidence from

UK firms. In the following sub-sections, we present details about how the two sub-samples

of firms compare with each other. On average, UK firms are comparable in size to other

European firms, with lower profitability and lower volatility of stock prices.

3.2 3.2 Key Variables and Summary Statistics

3.2.1 Female Representation on Corporate Boards

The standard measures of board gender diversity are a binary indicator for at least one

female director, and the fraction of female directors. We use three measures of female

representation: AnyFemale, %FemaleDirectors, and %FemaleonCommittees. AnyFemale

is a binary indicator of the presence of at least one female board members in a given

firm-year. This captures the effect of having at least one female director on the board.

6

Page 7: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

To examine the impact of increasing female representation on firm performance, we use

%FemaleDirectors, which is the ratio of number of female directors to the total number

of directors expressed as a percentage. Finally, to understand if female directors on key

committees have differential impact, we use %Female On Committees i.e. the ratio of the

combined number of female directors on three key committees (Audit, Nomination and

Remuneration) to the combined number of directors across these committees, expressed

as a percentage. This is an important variable for our empirical strategy: this measures

the extent to which female directors are integrated in the governance mechanisms of the

firm. There has been a steady, albeit incremental, increase in the gender diversity on

European boards over the last decade. We present this in Figures 1&2 for our sample

period. It seems that the fraction of firms with at least 20% female representation on the

board has increased over the sample period, but female representation on key governance

committees have been very stable throughout. This can possibly explain the empirical

evidences of no significant relationship between female directors and firm performance.

The time-covention of the measure of female representation is not immediately ap-

parent. Some studies use contemporaneous female representation (Adams and Ferreira,

2009), whereas some other uses lagged measures (Gregory-Smith, et al. 2014). All mea-

sures of female representation used in this paper is lagged by one period.1 We control

for a range of board characteristics, viz. board size2, board independence (percentage of

independent directors on the board),3 and an indicator for a female CEO.

It is critical to understand the comparability of the measures of board gender diversity

for the two sub-groups. In Figures 3 and 4, we present the comparison of the proportion

1We check the robustness of our sample using contemporaneous measures of gender diversity. Theresults are very similar to the baseline estimates and are available on request.

2In the case of two-tier boards, board size is the linear summation of the number of directors on boththe management and the supervisory board.

3The definition of an independent director varies marginally across countries. However, the basicremains that for a director to be considered independent, she will not be a current (or a former) employee,a relative of a current executive, or has business relations with the firm.

7

Page 8: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

of female on the board and the proportion of females on committees, for the UK and

non-UK firms. A larger proportion of UK firms have at least 10% and 50% female direc-

tors compared to non-UK firms. However, a larger fraction of non-UK firms have more

than 20%, and more than 50% female representation on governance committees. This

highlights a possible difference in the extent to which female directors are integrated in

the governance mechanisms.

3.2.2 Firm Characteristics

The association between board gender diversity and performance often varies with the

choice of firm performance measure (Erhardt, et al. 2003; Smith et al. 2006). The

primary measure of firm performance for this analysis is Return on Assets (ROA). To test

for the robustness of the results, we also use other measures of performance: Tobin’s Q

approximated by market-to-book value ratio (MTBV) and Returns on Equity (ROE).We

control for the risk in a firm’s information and operation environment by the standard

deviation of monthly stock returns over the previous 12 month period. Natural logarithm

of annual sales is used as a proxy for firm size.

3.2.3 Director Characteristics

Annual compensation of directors in our sample comprises of an annual retainer fee,

fees for attending board meetings and some equity compensation. The compensation

schedule is similar for all the directors. A priori, it is not clear if nominal pay differences

may impact upon the association between female representation and firm performance.

Therefore we d not use the pay information in our empirical models. Summary statistics

are presented in Tables 1 and 2. The selection of individual directors on committees and

their impact on firm performance may be driven by the skills and experience. We include

8

Page 9: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

the age of the directors, the time in current role and the time on the board to mitigate

the unobserved heterogeneity in director ability.

4 4. Empirical Methods

The central question is whether female participation is associated firm performance. To

do that we estimate the following

FirmP erf ormance = βFemaleDirectorsonBoard+ γZ (1)

where β captures the effect of female directors on firm performance and Z is a vector

of all firm and director characteristics. The primary measure of firm performance used for

this analysis is Return on Assets (ROA) and the primary measure of female participation

in boards is the % of female on the board in a given firm-year. We test for the robustness

of our results using alternative measures of firm performance and female representation.

Next we aim to understand the transmission mechanism of female director represen-

tation on firm performance. We analyse the probability of individual female directors’

appointment on key governance committees- viz. audit, nomination and remuneration.

We examine how the network of female directors impacts upon their appointment in these

committees and through these appointments on firm performance. We estimate the fol-

lowing probit model to estimate the likelihood of female directors being appointed on

committees:

CommitteeAppointment = αFemale+ φZ (2)

9

Page 10: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

The dependent variable is a binary indicator of a female director being appointed

on one of the three committees in a given firm-year. α is the linear probability of an

individual female director being appointed on committees and Z is a vector of all firm

and director characteristics.

Adams and Ferreira (2009) argue that the positive association of firm performance and

female representation on boards may be driven by potential endogeneity in female appoint-

ments. Endogeneity concerns arise from omitted unobservable characteristics which may

simultaneously affect firm performance and the appointment of female directors. We ad-

dress this concern by employing a range of econometric techniques. We use firm fixed

effects to control for any time-invariant omitted variables. It may be argued that the

direction of causality is actually reverse: more successful firms hire more female directors.

Although it is not immediately apparent why that might be true, we address potential

reverse-causality concerns using instrumental variables. We use the network size4 of the

female directors and the square of the network size as instruments. Whilst the network

size may be expected to drive female appointment on boards, it is unlikely to indepen-

dently impact upon firm performance, except through the control variables included in

the regression.

Further, the firm performance indicators are likely to be serially correlated. Therefore

we estimate the above specifications with lagged dependent variables using GMM. All

indicators of female representation on boards and firm level characteristics are also lagged

by one period. This is control for potentially endogenous choices made by the board in

view of expected performance.

4BoardEx reports the network size of individual directors which is equal to the number of otherdirectors a given individual is "related" to. A relation between two individuals is established if one ormore of the following is true:They have graduated in the same classThey have worked in the same firm togetherThey sit on the same boards at the same timeThey share familial relationship.

10

Page 11: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

5 5. Results and Analysis

5.1 5.1 Female Directors and Firm Performance

In Table 3, we present the results of female representation on corporate boards and its

effect on firm performance. In column (1) we estimate the effect of proportion of female

directors on boards, lagged by one period, on firm performance (ROA). We progressively

add firm-level and board-level characteristics in columns (2) and (3), and finally country

and year fixed effects in column (4). In all the specifications, proportion of female directors

on board is positively associated with firm performance. This suggests that having more

female on corporate boards may be value-enhancing. However, this association may be

endogenous; better performing firms may appoint more female directors.

To address the concern that the positive correlation between female directors and firm

performance might be endogenous, in Table 4 we present the headline results of female

board representation and firm performance using a range of econometric techniques. In

all the specifications, the dependent variable is ROA. In specifications (1) and (2) we

present the OLS and fixed effects estimates respectively. There is a positive and signif-

icant association between %Female directors and firm performance. If endogeneity were

driving these results, we would expect to see different results in specifications (3) and

(4) where we present the results of IV and GMM estimates. The positive association

of firm performance and female representation on boards persists even when we correct

for endogeneity. In specification (4), this association is large but not statistically signifi-

cant. Therefore it does not seem that our central result is driven solely by endogenously

determined female representation on boards and firm performance.

To test the robustness of the results we use three different measures of female represen-

tation. In Table 5, we present the results of these regressions with ROA as the dependent

11

Page 12: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

variable. The measures of female representation in boards used in the three specifications

are % female directors on board, at least one female director in a given firm-year and

% female directors on committees, respectively. All measures of board gender diversity

are positively associated with firm performance. However, the positive effect of a larger

proportion of female directors on key committees is an order of magnitude higher than

that of the baseline specification. It seems that having female directors more integrated

in the functioning of the boards generates value over and above symbolic tokenism.

We use a range of firm performance measures to test the robustness of the results. In

Table 6, we present the Arellano-Bond one-step estimates using three different measures

of firm performance: ROA (column 1), Tobin’s Q, approximated by market-to-book value

ratio (column 2) and return on equity (column 3). All variables are lagged and lead to loss

of precision of the estimates. However, the positive association between our measure of

gender diversity (% female on the board) and all three firm performance measures persist.

Thus our results suggest that gender diverse corporate boards is associated with pro-

ductivity gain for firms which is manifest in the form of enhanced firm performance.

However, our results may not necessarily support the idea of gender quotas for boards

as short term supply constraints may counterbalance any productivity gained from such

diversity.

5.2 5.2 Committee Assignments of Women Directors

The role of the committees is to specialize in narrowly-defined tasks. The number and

functions of these committees vary across firms and functions are sometimes combined.

We focus on three key committees- the audit committee that focuses on appointment of in-

dependent auditors and managing internal financial performance, nomination committee-

12

Page 13: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

that recommends appointment of new directors on board and remuneration/compensation

committee-that focus on compensation and benefits for executives. A priori, a director

who is a member of one or more of these committees has more influence on the strategic

choices made by the firm. We test whether the likelihood of female directors being on

these committees is different from that of men conditional on the proportion of female on

the board. This allows an examination of the mechanism of the impact of female board

representation. The sample is restricted to only non-executive directors. We present the

results in Table 7. The key variable of interest is Female, an indicator for an individual

director being of the eponymous gender. The number of observations varies across specifi-

cations because not all firms in the sample have all the three committees. The dependent

variable in each specification is a binary indicator of whether an individual director is a

member of any of the three committees (1), audit committee (2), nomination committee

(3) and remuneration committee (4) respectively. All specifications present the marginal

effects from probit regressions with firm fixed effects and year dummies and the standard

errors are clustered at firm level.

The results suggest that the likelihood of female directors being appointed on com-

mittees are different from that of male directors. Female directors are only more likely to

be appointed on audit committees. Unsurprisingly, female directors are more likely to be

appointed on committees when the proportion of female directors on the board is high.

The point estimate of the likelihood of female directors to be appointed on nomination

committee is negative, although this is not statistically significant at conventional levels.

Intuitively, if more female directors were to be appointed on nomination committees, it

might be plausible that more females would be nominated to be on the boards. However,

we can’t provide any definitive results on this.

13

Page 14: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

5.3 5.3 Additional Results and Robustness

5.3.1 A. U.K. vs Non-U.K. sample

About 35% of our observations are for UK firms. To ensure that our results are not

driven by the disproportionate presence of firms from one country, we run our baseline

regressions separately for UK and non-UK firms. The results are presented in Table 8. We

provide results for both measures of board gender diversity, viz. percentage of females on

the board, and the percentage of females on the committee. This exercise provides some

interesting insights. First, the effect on firm performance of percentage female directors

on the board is much stronger for the non-UK sample. The parameter estimate for the

UK sample is statistically insignificant. This is consistent with the results of Gregory-

Smith et al (2014) who finds no evidence of board gender diversity on the performance of

UK firms. Therefore, it seems our baseline results are downard-adjusted due to the large

proportion of UK-firms in the sample.

Second, and perhaps more interesting insight comes from the effect on firm perfor-

mance of the proportion of female directors on key committees. As above, the effect is

much stronger for the non-UK sample. However, the parameter estimate for the UK sam-

ple is now both significant and positive. This suggests that whilst the traditional measures

of board gender diversity has no effect on firm performance for UK firms, having female

directors integrated in the governance mechanism is value-enhancing.5 These results un-

derscores the fact that the current evidence on female representation on corporate boards

shows the returns to tokenism, rather than the full economic benefits of internalizing the

diversity.

5The regressions for Table 8 contains the full set of controls as the baseline regressions. In the interestsof brevity, we only report the estimates of the key variables.

14

Page 15: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

5.3.2 B. Excluding financial crisis years of 2008 and 2009

Even though our baseline results include year dummies, to ensure that our results are not

affected by the financial crisis, we exclude observations from 2008 and 2009. The results

with the reduced sample are very similar to that of the baseline estimates. For the sake

of parsimony, the tables are not presented here, but are available on request.

6 6. Conclusion

Gender diversity on corporate boards is likely to be one of the central themes of future

governance reforms. However there is no clear agreement on the economic benefits of re-

cruiting more women on boards. A growing body of literature examines the consequences

of increasing board gender diversity. In this paper we attempt to provide new evidence

on the impact of board gender diversity on firm performance using data from large Euro-

pean firms with a wider dispersion in women representation on board. We also examine

a possible transmission mechanism of female representation on firm performance. The

results of this paper suggest that increasing female representation on corporate boards

is associated with enhanced firm performance. This association is particularly strong in

firms with a high proportion of female directors ion key committees.

More generally, our results add to the literature in providing evidence of how increasing

female representation on boards, over and above symbolic tokenism, might positively

impact upon firm performance. This augments the existing case for increasing gender

diversity of boards from a standpoint of morality and equity and extends an economic

case for it. These results also provide support to policy initiatives aimed at increasing

board gender diversity.

15

Page 16: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

References

[1] Adams, R.B. and Ferreira, D. (2009) ’Women in the boardroom and their impact on

governance and performance’, Journal of Financial Economics, vol. 94(2), pp.291-309

[2] Adams, S.M., Gupta, A., Haughton, D.M. and Leeth, J.D. (2007). ‘Gender differences

in CEO compensation: evidence from the USA’,Women in Management Review, vol.

22(3), pp. 208—24.

[3] Ahern, K.R. and Dittmar, A.K. (2012) ’The changing of the boards: the impact

on firm valuation of mandated female board representation’, Quarterly Journal of

Economics, vol. 127(1), pp. 137-97.

[4] Arellano, M. and Bond, S. (1991). ’Some tests of specification for panel data: Monte

Carlo evidence and an application to employment equations’, Review of Economic

Studies, vol. 58(2), pp. 277-98.

[5] Becker, G.S. (1957) The Economics of Discrimination,\chicago, IL: University of

Chicago Press

[6] Erhardt, N.L., Werbel, J.D. and Shrader, C.B. (2003). ‘Board of director diversity

and firm financial performance’, Corporate Governance, vol. 11(2), pp. 102—11.

[7] European Union (2012). ‘Women in economic decision-making in the EU: progress re-

port. A Europe 2020 initiative’, Commission StaffWorking Document, Luxembourg.

[8] Gregory-Smith, I., Main, B.G.M., and O’Reilly, C.A. (2014). ’Appointments, pay

and performance in UK boardrooms by gender’, The Economic Journal, vol. 124, pp.

F109-F128.

16

Page 17: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

[9] Storvik, A. and Teigen, M. (2010). ‘Women on board. The Norwegian experience’,

Friedrich-Ebert-Stiftung International Policy Analysis, Berlin, Germany, available at:

http://library.fes.de/pdf-files/id/ipa/07309.pdf (last accessed: 15 October, 2014).

[10] Nielsen, S. and Huse, M. (2010). ‘The contribution of women on boards of directors:

going beyond the surface’, Corporate Governance: An International Review, vol.

18(2), pp. 136—48.

[11] O’Reilly, C.A. and Main, B.G.M. (2012). ‘Women in the boardroom: symbols or

substance?’, Stanford Graduate School of Business, Palo Alto, California, available

at: http://ssrn.com/abstract=2039524 (last accessed: 15 October, 2014).

[12] Farrell, K.A. and Hersch, P.L. (2005). ‘Additions to corporate boards: the effect of

gender’, Journal of Corporate Finance, vol. 11(1/2), pp. 85—106.

[13] Hermalin, B. and Weisbach, M. (1988) ’The determinants of board composition’,

RAND Journal of Economics, vol. 19, pp. 589-606.

[14] Hermalin, B., and Weisbach, M. (2003) ’Board of Directors as an endogenously de-

termined institution: a survey of the economic literature’, Economic Policy Review,

vol. 9, pp. 07-26.

[15] Larcker, D.F., Richardson, S.A. and Tuna, A.I. (2007). ‘Corporate governance, ac-

counting outcomes and organizational performance’, Accounting Review, vol. 82(4),

pp. 963—1008.

[16] Linck, J., Netter, J., and Wang, T. (2008) ’The determinants of board structure’,

Journal of Financial Economics, vol. 87, pp. 308-328.

[17] Gregory, A., Jeanes, E., Tharyan, R. and Tonks, I. (2013) ’Does the stock market

gender stereotype corporate boards? Evidence from the markets reaction to directors

trades’, British Journal of Management, vol. 24(2), pp. 174-90.

17

Page 18: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

[18] Lazear, E.P. and Rosen, S. (1990) ’Male-female wage differentials in job ladders’,

Journal of Labour Economics, vol. 23(2), pp. 325-40.

[19] Milliken, F.J. and Martins, L.L. (1996) ’Searching for common threads: understand-

ing the multiple effects of diversity in organizational groups’, Academy of Manage-

ment Review, vol. 21(2), pp. 402-33.

[20] Mincer, J. and Polachek, S. (1974). ‘Family investment in human capital: earnings

of women’, Journal of Political Economy, vol. 82(2), pp. S76—108.

[21] Powell, G.N. and Butterfield, D. (1994). ‘Investigating the ‘glass ceiling’phenom-

enon: an empirical study of actual promotions to top management’, Academy of

Management Journal, vol. 37(1), pp. 68—86.

[22] Smith, N., Smith, V. and Verner, M. (2006). ‘Do women in top management affect

firm performance? A panel study of 2,500 Danish firms’, International Journal of

Productivity and Performance Management, vol. 55(7), pp. 569—93.

[23] Torchia, M., Calabr, A. and Huse, M. (2011). ’Women directors on corporate boards:

from tokenism to critical mass’, Journal of Business Ethics, vol. 102(2), pp. 299-317.

18

Page 19: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

Table1: Summary StatisticsThe sample consists of an unbalanced panel of 18499 director-level observations from118 firms for the period 2006-2013. Director level data is obtained from Execucompand firm level data is obtained from Datastream. All variables are winsorized at the1%-level.

Variable N Mean Std. Dev. Min Max

Firm Characteristics

Return on Assets 16647 6.643 6.108 -09.28 38.95Log Sales 16647 17.558 0.921 14.39 20.02Tobin’s-Q 16647 2.866 5.792 -58.37 86.00Stock Price Volatility 16647 23.939 6.091 13.05 49.38

Board Characteristics

Board Size 16647 16.963 5.942 6.00 36.00%Non-executive 16647 47.743 27.786 0.00 100.00% Female 16647 18.531 14.489 0.00 88.89Nomination Committee Size 16647 3.941 2.473 0.00 16.00Audit Committee Size 16647 4.208 1.461 0.00 8.00Remuneration Committee Size 16647 3.432 1.949 0.00 9.00

Director Characteristics

Total Pay 16647 1048.079 2483.93 0.00 67891.00Equity Pay 3861 2276.805 3356.398 0.00 64906.00Time on Board 16647 5.756 5.269 0.00 54.90Time in Role 16647 4.535 4.238 0.00 47.72Executive Age (years) 16647 58.115 8.097 26.00 90.00

19

Page 20: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

Table 2: Comparisons of Firms with and without at least one Female DirectorsThis table presents key summary statistics for firm-years with no female directors andfirm-years with at least one any female director. Firms with no-female directors aresmaller and have smaller boards. There is no statistically significant difference inany other attributes. All variables are winsorized at the 1%-level.

Variable Mean-No Mean-At Least One p-valueFemale Directors Female Director

Log Sales 17.296 17.614 0.000Tobin’s-Q 3.025 2.819 0.272Return on Assets 5.869 6.697 0.000Volatility 23.824 23.949 0.000

Board Size 15.140 17.152 0.272% Non-Executive 47.109 47.811 0.281Age (years) 59.035 58.013 0.000Nomination Committee Size 3.849 3.950 0.066Audit Committee Size 3.283 4.308 0.000Remuneration Committee Size 3.541 3.420 0.010

20

Page 21: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

Figure 1: Yearly Trends in Board Gender Diversity

This figure shows that there has been a rise in female representation

on corporate boards. There seems to be an increase in the proportion

of firms with more than 20% female directors.

21

Page 22: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

Figure 2: Yearly Trends in Female Representation

on Governance Committees

The fraction of female directors on key governance committees seems

to have remained relatively stable over our sample period.

22

Page 23: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

Figure 3: Female Representation on Boards:

UK vs Non-UK Firms

This figure compares the female representation on boards of UK and

non-UK European firms in our sample. It seems that a higher fraction

of non-UK firms have 10% female directors, whereas a higher fraction

of UK firms have over 50% female directors on board.

23

Page 24: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

Figure 4: Female Representation on Governance

Committees: UK vs Non-UK Firms

This figure compares the female representation on key governance

committees of UK and non-UK European firms in our sample. A higher

fraction of non-UK firms seems to have more female representation on

governance committees.

24

Page 25: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

Table 3: Female Representation and Firm PerformanceThis table presents the results of the impact on firmperformance of female representation on boardsusing three different measures of female representation,viz. %Female directors on the board (1), whether thereis at least one female director in a firm-year (2) and the% of Females on committees (3). All three measuressuggest a positive association of female directors withfirm performance. All specifications include year dummiesRobust standard errors in parentheses. Asterisks indicatesignificance at 0.01 (***), 0.05 (**) and 0.10 (*)levels.

Dependent Variable ROA(1) (2) (3) (4)

% Femalet−1 0.052*** 0.0428*** 0.0148*** 0.005**(0.003) (0.003) (0.003) (0.002)

Log Salest−1 0.7290*** 0.3373*** 0.033***(0.0510) (0.049) (0.007)

Volatilityt−1 -0.2642*** -0.2158*** 0.000(0.0076) (0.007) (0.026)

Board Sizet−1 -0.4778*** -0.016***(0.0102) (0.003)

% Non-Executivet−1 0.066*** 0.0757***(0.002) (0.015)

Chairman-CEOt−1 -1.689***(0.989)

Constant 5.008*** 24.99*** 28.69*** 4.70***(0.804) (0.920) (0.884) (0.593)

Firm Fixed Effects No No No YesYear Dummies No No No YesObservations 16.647 16,647 16,647 16,647Adjusted R2 4.15 10.58 19.76 20.80

25

Page 26: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

Table 4: Female Directors and Firm Performance-Comparison

This table presents the results of the effect of female directors on firmperformance using different estimation techniques. Only the key covariatesare reported in all specifications. The main variables of interest is %FemaleColumn (1) presents the OLS estimates; column (2) includes firm fixed effects;column (3) presents IV estimates with network size of individual directors asan instrument and column (4) presents the result of Arellano-Bondone step regression. All specifications include year dummies. Robuststandard errors in parentheses. Asterisks indicate significance at 0.01 (***),0.05 (**) and 0.10 (*) levels.

(1) (2) (3) (4)

OLS FE IV GMM

Dependent Variable ROA

% Femalet−1 0.023** 0.005** 0.098**(0.003) (0.002) (0.044)̂%Female 0.364***

(0.057)Log Salest−1 -0.300*** 0.033*** 0.580*** 0.737**

(0.049) (0.007) (0.163) (0.306)Volatilityt−1 -0.206*** 0.000 -0.103** -0.0918**

(0.007) (0.026) (0.038) (0.0483)Board Sizet−1 -0.472*** -0.016*** -0.0301* -0.0244**

(0.010) (0.003) (0.015) (0.0095)% Non-Executivet−1 -0.064** 0.0757*** 0.237** 0.0100

(0.002) (0.015) (0.079) (0.0130)Constant 26.98*** 48.03*** 11.98** 16.77**

(0.898) (0.927) (4.166) (7.42)Firm Fixed Effects No Yes Yes YesYear Dummies Yes Yes Yes YesObservations 16,647 16,647 16,647 16,647Adjusted R2 25.50 20.80

26

Page 27: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

Table 5: Different Measures of Female RepresentationThis table presents the results of the impact on firmperformance of female representation on boardsusing three different measures of female representation,viz. %Female directors on the board (1), whether thereis at least one female director in a firm-year (2) and the% of Females on committees (3). All three measuressuggest a positive association of female directors withfirm performance. All specifications include year dummiesRobust standard errors in parentheses. Asterisks indicatesignificance at 0.01 (***), 0.05 (**) and 0.10 (*)levels.

Dependent Variable ROA(1) (2) (3)

% Femalet−1 0.005***(0.002)

Any Femalet−1 0.402**(0.148)

% Female in 0.035**Committeet−1 (0.013)Log Salest−1 0.033*** 0.098** 0.100***

(0.007) (0.048) (0.049)Volatilityt−1 0.000 -0.228*** -0.225***

(0.026) (0.007) (0.007)Board Sizet−1 -0.016*** -0.433*** -0.424***

(0.003) (0.014) (0.014)% Non-Executivet−1 0.0757*** 0.050*** 0.011***

(0.015) (0.002) (0.002)Constant 48.03*** 17.63*** 17.91***

(0.927) (0.978) (0.988)Firm Fixed Effects Yes Yes YesYear Dummies Yes Yes YesObservations 16,647 16,647 16,647Adjusted R2 20.80 29.24 28.13

27

Page 28: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

Table 6: Female Directors and Firm Performance-GMMThis table presents the results of GMM estimation of theeffect of female directors on firm performance. The dependentvariable in each specification is mentioned. All variables arelagged to control for potential endogeneity. Standard errorsare provided in brackets. Asterisks indicates significance at0.01 (***), 0.05 (**) and 0.10 (*) levels.

(1) (2) (3)

Dependent Variables ROA Tobin’s Q ROE

% Femalet−1 0.0644** 0.059** 0.189**(0.003) (0.0102) (0.085)

yt−1 0.062** -0.076 0.774**(0.0275) (0.683) (0.375)

Log Salest−1 0.007*** 0.026** 0.008**(0.0005) (0.009) (0.003)

Volatilityt−1 -0.035 -0.048 -0.011**(0.029) (0.033) (0.005)

Board Sizet−1 -0.014** -0.006** -0.033***(0.006) (0.002) (0.008)

% Non-Executivet−1 0.067** 0.086* 0.009**(0.002) (0.043) (0.002)

Any Femalet−1 0.913* 0.674 0.620(0.448) (0.415) (0.388)

Observations 663 663 663No AR(2) 0.436 0.232 0.455Sargan Test 0.334 0.190 0.218

28

Page 29: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

Table 7: Assignment of Women Directors on Key CommitteesThis table presents the probability of individual female directors to be assigned to keygovernance committees-audit, nomination and remuneration. The sample consists of anunbalanced panel of 18499 directors for the period. 2006-2013. The main variable ofinterests are Female and Female*%Female. Female is a binary indicator for afemale director. The results suggest that female directors are more likely to bechosen on any committees when the % of Female directors on the board is high. Allcolumns show marginal effects of a probit model with firm fixed effects and yeardummies. Robust standard errors in parentheses. Asterisks indicate significance at0.01 (***), 0.05 (**) and 0.10 (*) levels.

(1) (2) (3) (4)

Dependent Variable Any Audit Nomination RemunerationCommittee Committee Committee Committee

Time in Role -0.000721* -0.00125 -0.000458 -0.00258(0.000432) (0.00177) (0.00221) (0.00202)

Age (Years) 0.000487** 0.00464*** 0.00693*** 0.00634***(0.000244) (0.00112) (0.00133) (0.00107)

Board Size -0.00248*** -0.00943*** -0.0105*** -0.0163***(0.000770) (0.00251) (0.00322) (0.00236)

Female -0.00578 0.101*** -0.0214 0.000932(0.00407) (0.0306) (0.0213) (0.0242)

% Female Directors 0.000243 0.00102 0.00228** 0.00105**(0.000168) (0.000630) (0.000935) (0.000497)

Female* 0.000314** 0.00021** 0.000366 0.000308% Female Directors (0.000156) (0.00012) (0.000358) (0.000317)% Non-Executive -2.00e-05 -0.000763* 0.000426 0.000884**Directors (0.000121) (0.000449) (0.000723) (0.000444)Return on Assets -0.000215 -0.000461 -0.000999 -0.00202

(0.000283) (0.00134) (0.00208) (0.00132)Log Sales -0.00469** -0.0102 -0.0199 -0.0154

(0.00236) (0.00955) (0.0197) (0.00946)Volatility -0.000377 0.000411 0.00171 -0.000165

(0.000305) (0.00116) (0.00263) (0.00111)Network Size 0.0023*** 0.0038** 0.0022 0.0016

(0.0011) (0.0017) (0.0019) (0.0023)

Year Dummies Yes Yes Yes YesObservations 16,171 16,171 16,171 16,171

29

Page 30: From Symbolism to Integration: New Evidence on Female ... · The time-covention of the measure of female representation is not immediately ap-parent. Some studies use contemporaneous

Table 8: UK vs. Non-UK SampleThis table presents the results of the impact on firm performanceof female representation on boards for the UK and the non-UKsub-samples. Two measures of female representations are used:viz. %Female directors on the board (1) and (3), and % of Femaledirectors on committees (2) &(4). The dependent variable in allspecifications is ROA. The results suggest a stronger impact offemale represenation on firm performance for the non-UK sample.All specifications include year dummies. Robust standard errorsin parentheses. Asterisks indicate significance at 0.01 (***), 0.05(**) and 0.10 (*) levels.

UK Non-UKDependent Variable ROA

(1) (2) (3) (4)

% Femalet−1 0.007 0.033***(0.004) (0.005)

% Female in 0.011* 0.037***Committeet−1 (0.006) (0.003)Constant 26.18*** 26.095*** 30.07*** 29.414***

(1.609) (1.621) (1.233) (1.262)Firm Fixed Effects Yes Yes Yes YesYear Dummies Yes Yes Yes YesObservations 5,794 5,794 10,580 10,580Adjusted R2 28.29 28.29 22.22 22.02

30