BOARD GENDER DIVERSITY AND FIRM PERFORMANCE:
EMPIRICAL EVIDENCE FROM COUNTRIES IN ASIA
PACIFIC REGION
Tesalonika Devi University of Groningen Faculty of Economics and Business
Supervisor Drs. Boris van Oostveen June 2017
ABSTRACT
While many studies concentrated in US and Europe, this research examines the link between board gender diversity and firm performance in four Asia-Pacific countries: Australia, India, Japan, and Singapore from 2006 to 2015. In assessing the board gender diversity-firm performance relationship, this study deals with endogeneity issue using the instrumental variable and fixed effect estimations, and it also takes into account the effect of the differences of society’s perception towards women in the economic role between countries. The first part of the study shows that increasing numbers of female directors on the board have a negative effect on short-term performance, but no effect on long-term performance. The second part of the study shows that the effect of board gender diversity on firm performance is consistently negative regardless of any perceptions prevailing in the society towards women in the workplace.
I. INTRODUCTION
Female representation on corporate boards is often linked to improved firm performance. This notion is based on the idea that gender-diverse boards are considerably better at bolstering corporate governance and subsequently enhancing firm performance. Although the idea is intuitively appealing, empirical evidence remains inconclusive, and this is subject to a heated debate. To address this issue, this study attempts to examine the link between board gender diversity and firm performance in four Asia-Pacific countries: Australia, India, Japan, and Singapore by taking into consideration the endogeneity problems and the effect of the differences of society’s perception towards women in the economic role between countries.
The calls for greater gender diversity in the boardroom are getting louder, and it becomes a global concern over time. Not only in European Union (EU), United Kingdom (UK) and the United States (US), several implemented policies to address this priority have also become apparent across the Asia-Pacific region in recent years. For example, Malaysia introduced a gender quota system for women on boards in 2011, and India followed it in 20131. Furthermore, Australia also enacted self-disclosure policies requiring listed companies to publicly disclose board gender composition in 2010, Japan followed in 2013, and Singapore followed in 20152. These initiatives show a serious commitment of the regulators to push companies towards board gender balance since women remain substantially underrepresented in the boardroom, in particular for Asia-Pacific (10.2%) that still far lags other developed economies such as UK (26.1%), EU (20.8%), and US (18.7%).3
Despite the strong support for gender diversity on boards, mixed results raise a concern whether board gender diversity matters. Some empirical studies show positive link between gender diversity and performance (Carter, 2003; Campbell and Minguez-Vera, 2008) while others conclude negative relationship (Adams and Ferreira, 2009; Bøhren and Strøm, 2010; Darmadi, 2013), or even no effect between the two (Bonn et al., 2004; Farell and Hersch 2005). The actual relationship between board gender diversity and firm performance thus remains an open question signaling an urgency for further research. This study aims to extend the literature in
1 Deloitte, 2014. Women in the Boardroom: A Global Perspective. Deloitte Touche Tohmatsu Limited. 2 PwC Singapore, 2016. Board Diversity Disclosures in Singapore: A Good Practice Guide. PwC
3 Korn Ferry, 2016. Korn Ferry Diversity Scorecard 2016: Building Diversity in Asia-Pacific Boardrooms. Korn
this regard, and it consists of two main analysis. The first part of the study focuses on assessing the general link between board gender diversity and firm performance, while the second part brings the analysis to next level by addressing cultural differences between countries when examining the board gender diversity-firm performance relationship. Given the national culture differences, society’s shared values and beliefs associated with gender roles may differ between countries. As national culture may influence the social process in the boardroom (Li and Harrison, 2008), these values and beliefs may affect how women directors are stereotyped, and consequently may influence whether board gender diversity improves board dynamics and functioning, and ultimately firm performance. Thus, this study argues that even in those countries located in the same region, the board gender diversity-firm performance relationship may vary depending on the national culture, which might also explain why there are mixed results in the literature.
This research contributes to the literature in three ways. First, this study helps to fill the gaps in the existing studies by using non-US and non-European data from four Asia-Pacific countries: Australia, India, Japan, and Singapore in the period 2006 to 2015. This can be considered as an improvement since the existing studies are still mostly concentrated in the leading economies and using a smaller time frame (Low et al., 2015). Second, this study is one of the few attempts that deal with the endogeneity issues by using both the fixed effects and instrumental variable approach. As the endogeneity issues are rarely addressed, the existing studies are lacking since they may often produce biased causal interpretations (Adams and Ferreira, 2009). Third, this study considers the effect of cultural differences on the link between board gender diversity and firm performance. Unlike other studies that heavily used Hofstede or Schwartz cultural model dimensions to explain national cultural settings differences (Li and Harrison, 2008; Breuer and Salzmann, 2012), this study defines national culture differences specifically based on society’s attitude towards women in the economic role, and it is proxied by Economic Participation and Opportunity (EPO) score as suggested by Low et al. (2015). Since gender-stereotyping views are formed based on the society’s attitude towards women role (Eagly and Mladinic, 1989; Brannon, 2002), this approach may effectively predict the interactions between male and female directors and assess its implications on board functioning and firm performance.
However, not as expected, the second part of the analysis shows that the impacts of board gender diversity on firm performance are consistently negative for both in settings where society’s attitude towards women is positive and in settings where society’s attitude towards women is negative.
This study is organized as follows. Section two elaborates the relevant theories on the link between board gender diversity and firm performance, empirical evidence, and hypothesis developments. Section three describes the endogeneity issue and the empirical models. Section four describes the data, variables, and an overview of the descriptive statistics. Section five presents the results and robustness test. Finally, section six presents the conclusions, limitations, and suggestions for further research.
II. LITERATURE REVIEW
Agency theory and resource dependence theory are known as the dominant theories that predict a positive relationship between female representation in the boardrooms and firm performance, whereas social psychological theory offers the possibility of either a positive or negative link between the two depending on the group dynamics. In this section, those aforementioned theories and the empirical evidence are first elaborated to give an insight of what we may expect from the relationship between board gender diversity and firm performance. Subsequently, the potential effects of cultural differences between countries on the link between gender diversity and firm performance are also discussed.
Agency theory
Agency theory focuses on the agency problem, which arises from the conflict of interest between owners (the principals) and manager (the agent) due to the separation of ownership and control in organizations. It is based on the premise that managers will not always act in the best interest of owners since both parties incline to maximize their utilities (Jensen and Meckling, 1976). Instead, managers are more likely to choose the alternative decision that deviates from owners’ interest to pursue their private benefits (Core et al., 2006; Finegold et al., 2007). Consequently, this problem creates agency cost that may reduce firm value.
Board of directors as a central part of corporate governance mechanism play a pivotal role to align the interest of owner and managers (Campbell and Minguez-Vera, 2008). It helps to overcome agency problem through monitoring managers’ behavior, managing compensation and undertaking disciplinary actions (Fama and Jensen, 1983; Shleifer and Vishny, 1986; Campbell and Minguez-Vera, 2008). Furthermore, it is widely seen that one of the ways to enhance the ability of boards to perform those tasks is by having greater board gender diversity (Carter et al., 2003; Adam and Ferreira, 2009; Gul et al., 2011; Low et al., 2015).
Literature suggests that gender diversity is beneficial to improve board functions through several ways (Carter et al., 2003; Adam and Ferreira, 2009; Gul et al., 2011). First, gender diversity leads to improved monitoring and oversight of managers’ behavior (Carter et al., 2003; Adams and Ferreira, 2009; Gul et al., 2011). From the interview with women board members of top US companies, Selby (2000) find evidence that having women seated on the boards positively affects the board's interactions as women are keen to raise critical questions. As a result, board gender diversity can induce managers to be more transparent (Gul et al., 2011). Additionally, women are considered to carry their duty more seriously, and it is corroborated empirically from the evidence that women tend to join monitoring committees, and they have better attendance behavior than men board members (Adams and Ferreira, 2009). Second, gender-diverse boards may produce better compensation management. It is shown that gender diversity positively affects the fraction of the equity-based pay, which may encourage managers to perform in alignment with shareholders interest beyond the effect of fixed cash compensation (Adams and Ferreira, 2009). Third, gender-diverse boards are found to positively affect CEO turnover-performance sensitivity (Adams and Ferreira, 2009) that may improve the disciplinary functions of boards in the case of poor management performance.
turn may reduce firm value. In general, although agency theory does not give a full support towards a positive relationship between board gender diversity and firm performance, it indeed predicts that the positive relationship may occur.
Resource dependence theory
Resource dependence theory posits that companies are dependent on environmental resources, and this leads to the interdependence between organizations (Pfeffer and Salancik, 1978). In this context, the board of directors performs an important function to connect firms with the resources needed. According to Pfeffer and Salancik (1978), boards can perform their job as a linkage mechanism by three means: (1) providing advice and consultation, (2) creating legitimate recognition via firm's public image, (3) providing signals for communicating information and for gaining support from important stakeholders. In this way, they argue that board of directors may help the firm to minimize the dependency to external entities and uncertainties, thereby increasing firm value. Considering the importance of boards function as a linkage mechanism, it thus becomes necessary to select board characteristics that may bring the most benefits to the company (Hillman, 2000), and it is being said that a more diverse board can provide more benefits to the company (Carter, 2010).
(Brammer et al., 2007). In turn, this positive firm reputation may help the firm to obtain access to relevant stakeholders who control the resources needed (Hillman et al., 2000).
As reviewed above, resource dependence theory provides the rationale for the positive association between board gender diversity and firm performance. Lückerath-Rovers (2011) find support for this as they show that gender diversity positively affects firm performance in Dutch companies, suggesting that a more gender-diverse boards may have a better connection
with all relevant stakeholders, which may also induce firm reputation. Similarly, Erhardt et al. (2003) also find a positive relationship between gender diversity and company
performance in US companies, arguing that gender diversity may better reflect the diversity of firm’s customers and labor pool. Those findings above thereby confirm the prediction of resource dependence theory.
Social psychological theory
From the social psychological theory viewpoint, the net effects of board gender diversity on firm performance depends on the social context and group interaction. Supporting this view, studies find that although female directors may bring potential benefits to the board, the possibility of whether this potential can be unleashed depends on various circumstances, such as women’s social connection with the majority directors (Westphal and Milton, 2000), the ability of women to fit in and gain credibility from male counterparts, and the effectiveness of chairperson in supporting the participation of female directors (Kakabadse et al., 2015). While numerous studies document a positive relationship between board gender diversity and firm performance, some empirical studies also find either no effect or a negative link between the two. Studies such as Shrader et al. (1997), Haslam et al. (2010), and Bøhren and Strøm (2010) find a negative relationship between gender diversity and firm value. Consistent with the social psychological theory, they argue that this relationship is negative due to negative stereotyping towards the women capability (Haslam et al., 2010), and ineffective decision-making process (Shrader et al., 1997; Bøhren and Strøm, 2010). Also, Carter et al. (2010) find no evidence of the effect of board gender diversity and firm performance, arguing that valuable benefits provided by female directors may have been offset by social psychological dynamics such as conflicts and exclusion.
Gender stereotype bias
In summary, all the theories above generally give an indication that there is indeed a relationship between board gender diversity and firm performance. However, the direction of the link being either positive or negative is not clear, and the prediction from existing empirical studies are still not able to give conclusive results. Following agency theory and resource dependence theory, the author acknowledges that women may bring benefits to the firms through a better board monitoring and relationship with relevant stakeholders. However, the author believes that a negative stereotyping towards women’s capability are still dominates in the society. According to The Economist Corporate Networks (2016), this negative stereotyping view towards women leaders indeed exists in Asia-Pacific, which complicates women to blend in and gain credibility from male counterparts. In turn, this view may negatively affect boards dynamics as there are barriers for women to give a full contribution to the boards (Ragins, 1998). Given the existence of negative stereotypical views against women, this study thus hypothesizes that although gender diversity may bring benefits to the firm performance, there are possibilities that these benefits may fail to balance the adverse effects of social psychological interaction within a group such as conflicts and marginalization of women. Therefore, the first hypothesis is formulated as in the following.
H1: Board gender diversity negatively impacts firm financial performance
Attitude towards women’s participation in economy
Several studies have been suggesting that corporate governance phenomena should be effectively studied with proper attention to the effects of cultural differences between countries (Licht et al., 2005; Cheung and Chan, 2007). As national culture determines society’s beliefs about what is ideal and desirable, the way women directors are stereotypes and its implications on the social-psychological interactions in the boardroom may not necessarily be the same across countries. Hence, the effects of cultural differences need to be taken into account when examining the efficacy of board gender diversity on firm performance. Since society’s attitude towards women role underlies the creation of gender-stereotyping (Eagly and Mladinic, 1989; Brannon, 2002), this study defines culture specifically to social attitude toward women in the workplace.
counterparts, which enable them to unleash their real potentials to contribute to the boardroom. Hence, the presence of female directors may lead to healthy discussions and better decision-making process, and ultimately it may improve firm performance. Accordingly, the following hypothesis is proposed:
H2a: The effect of board gender diversity on firm performance is positive for firms located in countries characterized by positive society’s attitude toward women in economic roles.
In contrast, negative stereotypical views towards women leaders may dominate in countries characterized by skeptical society’s attitudes toward women in the workplace (Low et al., 2015). In this case, male directors tend to underestimate women directors’ capabilities and reluctant to give opportunities for women to contribute in the boardroom (Ragins, 1998). Beyond that, this prejudice may potentially damage women’s confidence, which leads to women’s underperformance (Bergeron et al., 2006). As such, this gender-stereotyping bias may be detrimental to firm performance. Thus, the following hypothesis is posited:
H2b: The effect of board gender diversity on firm performance is negative for firms located in countries characterized by negative society’s attitude toward women in economic roles.
III. METHODOLOGY
This section begins with the identification of endogeneity problems that may arise in the relationship between board gender diversity and firm performance. Subsequently, the empirical model employed in this study is described.
Endogeneity problems
Omitted variable bias occurs when the residuals include unobservable factors that influence both dependent and independent variables. For example, Adams and Ferreira (2009) point out that the unobservable factors can be in the form of firm characteristics such as corporate culture. They illustrate that companies with more progressive culture may have better performance as well as more female directors, which leads to spurious correlations between board gender diversity and firm performance. In this case, a fixed or random effect estimators can be employed to eliminate omitted variable bias.
In the board gender diversity-firm performance relationship, the direction of causality becomes another concern. Adams and Ferreira (2009) argue that board gender diversity may not only affect firm performance, but the improvement of company performance may also in turn positively impact firm’s incentives to hire female directors. This may happen because well-performing companies are more eager to take risky strategy in hiring minorities such as female directors (Smith et al., 2006). Consequently, the causality may not only run from board gender diversity to firm performance, but it can also flow the other way around. Hence, the instrumental variable approach needs to be utilized in this case since the fixed effect estimator may not be adequate to address reverse causality problems.
Empirical models
The first objective of this study is to identify the link between board gender diversity and firm performance, which is formulated as below model:
PERFORM𝑖𝑡 = 𝛼 + 𝛽1FEMALE𝑖𝑡+ 𝛽2CONTROL𝑖𝑡+ 𝑐𝑖+ 𝑣𝑡+ 𝜖𝑖𝑡 (1) where PERFORM𝑖𝑡 represents firm performance, FEMALE𝑖𝑡 is a proxy for female representation on the boards, CONTROL𝑖𝑡 represents control variables, i is a firm index, t is a time index, 𝛼 is the constant, 𝛽1 and 𝛽2 are the coefficients for FEMALE𝑖𝑡 and CONTROL𝑖𝑡, 𝑐𝑖 is a firm-fixed effect, 𝑣𝑡 is a time-fixed effect, and 𝜖𝑖𝑡 is the error term.
Nevertheless, due to the potential problem of reverse causality, two ways fixed effects model may be inadequate to provide unbiased and consistent results. As previously discussed, although the board gender diversity may affect firm performance, the reverse may also be true that firm performance may be affected by the firm. Hence, the instrumental variable approach can be implemented to circumvent this problem (Wang, 2015). This is done by initially estimating the first-stage regression as shown in the model (2) to find the fitted value of board gender diversity from the regression of gender diversity on the exogenous instrument and control variables. Thereafter, in the second-stage regression, firm performance is regressed on the fitted value of board gender diversity and control variables based on model (1).
FEMALES𝑖𝑡 = 𝛼 + 𝛽1INSTRUMENT𝑖𝑡+ 𝛽2CONTROL𝑖𝑡+ 𝑐𝑖+ 𝑣𝑡+ 𝜖𝑖𝑡 (2) where, FEMALE𝑖𝑡 is a proxy for female representation on the boards, INSTRUMENT𝑖𝑡 represent the instrument variable, CONTROL𝑖𝑡 represents control variables, i is a firm index, t is a time index, 𝛼 is the constant, 𝛽1 and 𝛽2 are the coefficients for INSTRUMENT𝑖𝑡 and CONTROL𝑖𝑡, 𝑐𝑖 is a firm-fixed effect, 𝑣𝑡 is a time-fixed effect, and 𝜖𝑖𝑡 is the error term.
In the second section of the study, the analysis includes the consideration of culture. Initially, the instrumental variable approach as in the first part are estimated, but now by including EPO score as the proxy of culture. This is done to see whether the inclusion of culture (EPO) may change the direction of the general relationship estimated in the first section. The first-stage regression model is shown as below:
FEMALES𝑖𝑡 = 𝛼 + 𝛽1INSTRUMENTS𝑖𝑡+ 𝛽2CONTROL𝑖𝑡+ 𝛽3EPO𝑖𝑡 𝑐𝑖+ 𝑣𝑡+ 𝜖𝑖𝑡 (3) After the fitted value is estimated based on model (4), the second-stage regression is employed: PERFORM𝑖𝑡 = 𝛼 + 𝛽1FEMALES𝑖𝑡+ 𝛽2CONTROL𝑖𝑡+ 𝛽3EPO𝑖𝑡+ 𝑐𝑖+ 𝑣𝑡 + 𝜖𝑖𝑡 (4)
IV. DATA
The dataset utilized in this study consists of firms geographically located in Australia, India, Japan, and Singapore in the period from 2006 to 2015. The sample countries used in this analysis are chosen to represent countries in Asia-Pacific that has different society’s attitude towards women in the economy. In this case, Australia and Singapore represent countries characterized by positive society’s attitude towards women in the economy (High EPO), while Japan and India represent countries with the negative society’s attitude (Low EPO). The sample is constructed by first selecting all active and listed companies operating in those countries from Bureau van Dijk’s Orbis. Afterward, firms with no board gender diversity data, and not listed for all ten years are excluded, bringing about 935 firm observations as the initial dataset. This consists of 287 firm observations from Australia, 97 from India, 397 from Japan, and 154 from Singapore. Consequently, the final sample is an unbalanced panel data with 9,350 firm-year observations. In this study, board characteristics data is mainly retrieved from Datastream Asset 4 ESG (Environmental, Social, and Governance), whereas firm-level and country-level data are retrieved from Datastream and The Global Gender Gap Report of World Economic Forum respectively. The definition of each variable and the data source are summed in Table I.
Improving normality conditions
To improve the normality conditions, logarithm transformation of the variable is used for all variables in this study, except for three variables: female indicator, EPO, and educational attainment. The skewness and kurtosis comparisons of all variable used in this study are presented in Table A.I (see Appendix). To preserve the sign of the data, this study employs the alternative technique of the natural logarithm transformation, namely logarithm modulus transformation (John and Draper, 1980). By using this method, any variables with a value equal to or below 0 are no longer dropped from the observation, thus leads to more observations.
Firm performance
Table I
Definition of board, firm, and country characteristics for Australia, India, Japan, and Singapore in the period 2006 to 2015
Definitions Source
Board characteristics
Percentage of females (log) Logarithm transformation of a total number of female directors on the board divided by the total number of board of directors at the end of the year.
Datastream Asset 4 ESG and annual report Number of females (log) Logarithm transformation of a total number of
female directors on the board at the end of the year.
Datastream Asset 4 ESG and annual report Female indicator A dichotomous variable that equals 1 when the
board has one or more female directors at the end of the year, 0 otherwise.
Datastream Asset 4 ESG and annual report Board size (log) Logarithm transformation of a total number of
directors on the board at the end of the year.
Datastream Asset 4 ESG and annual report
Firm characteristics
Return on equity (log) Logarithm transformation of annual net income divided by the book value of shareholder equity at the end of the year.
Datastream
Market to book (log) Logarithm transformation of the market value of equity divided by book value of equity at the end of the year.
Datastream
Firm age (log) Logarithm transformation of the number of years since incorporation
Orbis Total assets (log) Logarithm transformation of book value of total
assets at the end of year
Datastream Net sales (log) Logarithm transformation of net sales at the end
of year
Datastream Leverage (log) Logarithm transformation of total debt divided by
total equity at the end of year
Datastream
Country characteristics
Culture (EPO) Economic Participation and Opportunity score at the end of the year.
The Global Gender Gap Report
Educational attainment Educational attainment score at the end of the year.
The Global Gender Gap Report
financial performance (Hoskisson et al., 1994; Gentry and Shen, 2010). This study uses two
firm financial performance measure that is widely used in corporate governance literature: ROE and market to book value. ROE is an accounting-based measure that equals to
annual net income divided by the book value of shareholder equity. Market to book value is a market-based measure that can be defined as the market value of equity divided by book value of equity. For the robustness check, this study also replicates the regressions using ROA and ROIC to proxy firm performance.
Female representation on the corporate boards
boards at the end of the year, while others using female indicator (Carter et al., 2003; Darmadi, 2013), or number of females on the board (Carter et al., 2003). In this study, all those approaches are employed to measure female representation on boards better.
Data for board gender diversity are mainly derived from Datastream Asset 4 ESG. However, there are a lot of missing data in observations, especially data for Singapore. Thus, this study attempts to deal with this problem by hand-collecting the missing data from the firm annual report. Since not all the annual reports are available, many data are still missing, which is resulting in fewer observations.
Control variables
Hence, a greater level of debt may lead to a higher bankruptcy cost, which adversely affects firm value (Kılıç and Kuzey, 2016). Accordingly, the link between leverage and firm performance is expected to be negative.
Instrument variable
In using the instrumental variable approach, the instrument selected should be sufficiently correlated with the endogenous explanatory variable (board gender diversity), but uncorrelated with the dependent variable (firm performance). In this study, educational attainment score that captures the gap between women and men’s current access to education is used as the single instrument variable. The data are originated from The Global Gender Gap Report of World Economic Forum.
According to Cheung and Chan (2007), formal education is one of the important elements for women’s career advancement. It is because ones generally need to have sufficient education level to qualify for important positions in a firm. In other words, women who have more knowledge, skill, and a good education background are more likely to be chosen for boards. (Singh and Vinnicombe, 2004; Qian, 2016). Nevertheless, women traditionally tend to have less access to education, especially in the developing country (Islam and Amin, 2016), and this may be the reason for the absence of women on boards. Therefore, it can be argued that the higher education access for women, the higher the probability of women appointed as directors. In a study of 73 developing countries, Islam and Amin (2016) find evidence for this as they find that countries with higher access to education among women relative to men, also have a higher proportion of top women managers. Thus, it is hypothesized that educational attainment positively affects women representation on boards.
Culture
Programme’s Gender-Related Development Index, and Gender Inequality Index (Low et al., 2015), those measures are only available starting from 2010 and not specifically measure participation and opportunity of women in the economic role. Therefore, EPO score is chosen in this analysis.
The EPO score is one of the components of the Global Gender Gap index developed by World Economic Forum that was first published in 2006. It is reported annually and measured based on three indicators: the participation gap, the remuneration gap, and the advancement gap between of women and men in the economic roles. (Hausman et al., 2015). Table A.II presents the EPO score of Australia, Singapore, Japan, and India from 2006 to 2015 (see Appendix). It shows that Australia and Singapore consistently have EPO score higher than global median from 2006 to 2015, whereas Japan and India have lower EPO score than the global median over time. Therefore, Australia and Singapore are chosen to represent high EPO countries that more receptive toward women in the economic roles, while Japan and India represent low EPO countries with less favorable attitudes towards women in the economy.
Descriptive statistics
The descriptive statistics of all variables used in this analysis are presented in Table II. In total, there are 7,534 observations for board characteristics data. The mean value of the percentage of females (log), the number of females (log), and female indicator are 0.954, 0.322, and 0.349 respectively. Since the mean value of the board size (log) is 2.270, the average number of females (log) thus considerably low if it is compared with the mean of board size, indicating that females are underrepresented in the boardroom.
Table II
Descriptive Statistics for board, firm, and country characteristics for Australia, India, Japan, Singapore in the period 2006 to 2015
Mean Std.Dev Min Max Obs
Board characteristics
Percentage of females (log) 0.954 1.329 0.000 4.151 7,534 Number of females (log) 0.322 0.463 0.000 1.792 7,534 Female indicator 0.349 0.477 0.000 1.000 7,534 Board size (log) 2.270 0.363 0.693 3.555 7,534
Firm characteristics
Return on equity (log) 1.600 2.137 -8.609 7.495 9,036 Market to book value (log) 1.010 0.575 -5.479 6.214 9,038 Firm age (log) 3.504 0.912 0.693 5.823 9,350 Total assets (log) 14.514 2.353 3.091 21.686 9,207 Net sales (log) 13.716 2.801 -15.892 19.419 9,213 Leverage (log) 0.456 0.495 -3.942 5.557 9,203
Country characteristics
Culture (EPO) 0.638 0.119 0.383 0.814 9,350 Educational attainment 0.967 0.047 0.819 1.000 9,350
Table III
Pairwise correlation matrix
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (1) Return on equity (log) 1
(2) Market to book value (log) 0.253 1
(3) Percentage of females (log) 0.081 0.137 1
(4) Number of females (log) 0.099 0.146 0.9705 1
(5) Female indicator 0.096 0.144 0.9812 0.9513 1
(6) Board size (log) 0.149 0.014 -0.081 -0.032 -0.033 1
(7) Firm age (log) 0.112 -0.104 -0.157 -0.140 -0.150 0.362 1
(8) Total assets (log) 0.259 -0.142 -0.072 -0.033 -0.036 0.547 0.465 1
(9) Net sales (log) 0.371 -0.067 -0.103 -0.067 -0.070 0.544 0.490 0.845 1 (10) Leverage (log) 0.039 0.049 0.005 0.023 0.019 0.231 0.086 0.435 0.317 1
V. RESULT
This study is divided into two sections. In the first part, it focuses on answering the first hypothesis by examining the relationship between female representation on the corporate boards and firm performance in Australia, India, Singapore, and Japan. Recognizing that society’s attitude towards women may be unique for each country, two subsamples are created in the second part of the analysis to assess the link between board gender diversity and firm performance in two different settings: high and low EPO countries.
First section of the analysis
Initially pooled OLS regressions are estimated, and the results are shown in Table IV.4 In models (1)-(3), the results indicate a positive link between board gender diversity and return on equity at the significance level of 1% (β=0.159; β=0.490; β=0.465). The same results hold for models (4)-(6), where the estimated coefficients of board gender diversity are also positively related to market to book value at a significance level of 1% (β=0.052; β=0.160; β=0.154).
Due to the possibility of omitted variables, basic pooled OLS may produce insignificant and biased results. To address this problem, this study also considers including fixed effects or random effect to the models. Consequently, Hausman test for random correlated effects is applied to see whether the estimates of fixed effects are consistent. Based on the Hausman test results, the fixed-effect model should be employed. To provide further analysis, this study uses a joint F-test to see whether the firm-fixed, time-fixed, or two-way fixed effects should be used in the model. The results suggest accounting for both firm-fixed and time-fixed effects in the estimation.5
4 Several diagnostic tests are performed to ensure that OLS estimates are blue. The results of Jarque-Berra tests
(6 in total) reject the null of normality at 1% significance level. Hence, the residuals are not normally distributed. Since the sample size is large, the violation of normality assumption has no significant consequence (Brooks, 2014). The results of Breusch-Pagan tests (6 in total) reject the null of no heteroscedasticity at 1% significance level, and the results of Wooldridge tests (6 in total) reject the null of no first-order autocorrelation at 1% significance level. Accordingly, the robust cluster standard errors are used to correct heteroscedasticity and autocorrelation. The Variance Inflation Factor (VIF) tests indicate no sign of multicollinearity as all the values below 4. All the results are available upon request from the author.
Table V shows the result of pooled OLS estimation with two-way fixed effects. In column (1) the percentage of women on boards is negatively related to return on equity at the significant level of 10% (β=-0.074). Column (2) and (3) show that both the number of females on boards
Table IV
Board gender diversity and firm performance in Australia, India, Japan, and Singapore, 2006-2015
This table presents the results of the basic pooled OLS without the inclusion of two-way fixed effects. Data for board characteristics are originated from Datastream Asset 4 ESG and annual report, while data for firm-level variables are from Datastream and Orbis. Columns 1,2,3 (Columns 4,5,6) show the results from return on equity (market to book value) regressed on the percentage of female directors, the number of females, and female indicator respectively, and the control variables. Return on equity is annual net income divided by the book value of shareholder equity. Market to book value equals the market value of equity divided by book value of equity. Percentage females equals the total number of female directors on the board divided by the total number of board of directors, number of females represents the total number of female directors, and the female indicator is a dichotomous variable that equals 1 when board has one or more female directors at the end of the year and 0 otherwise. Board size equals a total number of directors on the board, firm age is the total number of years since incorporation, total assets equal book value of total assets at the end of the year, net sales equal total net sales, leverage is total debt divided by total equity. For all variable, except female indicator, logarithm modulus transformation is used. ***, **, * indicate significance at the 1%, 5%, 10% levels respectively. Robust cluster standard errors to correct for heteroscedasticity and autocorrelation are shown in parentheses under the coefficient.
and female indicator have no significant effect on return on equity (β=-0.136; β=-0.165). Furthermore, the regression results in column (4)-(6) indicate that board gender diversity has no effect on the market to book value (β=0.008; β=0.031; β=0.023). Since the results of OLS estimation with and without the inclusion of two-way fixed effects are not consistent, this suggests that unobserved heterogeneity may exist resulting biased results. Therefore, the
Table V
Board gender diversity and firm performance in Australia, India, Japan, and Singapore, 2006-2015
This table presents the results of the basic pooled OLS with the inclusion of two-way fixed effects to address omitted variable problems. Data for board characteristics are originated from Datastream Asset 4 ESG and annual report, while data for firm-level variables are from Datastream and Orbis. Columns 1,2,3 (Columns 4,5,6) show the results from return on equity (market to book value) regressed on the percentage of female directors, the number of females, and female indicator respectively, and the control variables. Return on equity is annual net income divided by the book value of shareholder equity. Market to book value equals the market value of equity divided by book value of equity. Percentage females equals the total number of female directors on the board divided by the total number of board of directors, number of females represents the total number of female directors, and the female indicator is a dichotomous variable that equals 1 when board has one or more female directors at the end of the year and 0 otherwise. Board size equals a total number of directors on the board, firm age is the total number of years since incorporation, total assets equal book value of total assets at the end of the year, net sales equal total net sales, leverage is total debt divided by total equity. For all variable, except female indicator, logarithm modulus transformation is used. Hausman tests are performed for all estimations, and the results suggest that two-way fixed effects are preferred over random effects. ***, **, * indicate significance at the 1%, 5%, 10% levels respectively. Robust cluster standard errors to correct for heteroscedasticity and autocorrelation are shown in parentheses under the coefficient.
Return on equity (log) Market to book value (log) (1) (2) (3) (4) (5) (6) % females (log) -0.074* 0.008 (0.039) (0.009) # of females (log) -0.136 0.031 (0.099) (0.024) Female indicator -0.165 0.023 (0.100) (0.024)
two-way fixed effects estimations are preferred over the basic pooled OLS. Table VI shows the results of the first-stage regression model, where board gender diversity is regressed on the instrument and control variables. As predicted, educational attainment positively affects the presence of female directors on boards (β=6.824; β=2.135; β=3.024 for ROE and β=7.001; β=2.198; β=3.062 for market to book value) at the significance level of 1% for model (1),(3),(4),(6), and 5% for model (2) and (5). Additionally, the strength of the instrument is also
Table VI
First stage regressions: Fitted value identification of board gender diversity
This table presents the results of the first stage regression of instrumental variable approach with the inclusion of two-way fixed effects. Data for board characteristics are originated from Datastream Asset 4 ESG and annual report, while data for firm-level variables are from Datastream and Orbis, and the country-level data are from The Global Gender Gap report. Columns 1,2,3 (Columns 4,5,6) show the results from identifying the fitted values of the percentage of female directors, the number of females, and female indicator respectively, using educational attainment as the single instrument and control variables. Percentage females equals the total number of female directors on the board divided by the total number of board of directors, number of females represents the total number of female directors, and the female indicator is a dichotomous variable that equals 1 when board has one or more female directors at the end of the year and 0 otherwise. Educational attainment represents the country-level score of gender gap in education access, board size equals total number of directors on the board, firm age is the total number of years since incorporation, total assets equals book value of total assets at the end of year, net sales equals total net sales, leverage is total debt divided by total equity. For all variable, except female indicator and educational attainment, logarithm modulus transformation is used. ***, **, * indicate significance at the 1%, 5%, 10% levels respectively. Robust cluster standard errors to correct for heteroscedasticity and autocorrelation are shown in parentheses under the coefficient.
% of females (log) # of females (log) Female indicator % of females (log) # of females (log) Female indicator (1) (2) (3) (4) (5) (6) Educational attainment 6.824*** 2.135** 3.024*** 7.001*** 2.198** 3.062*** (2.379) (0.861) (0.966) (2.333) (0.848) (0.945) Board size (log) 0.272*** 0.140*** 0.127*** 0.303*** 0.150*** 0.138***
(0.063) (0.022) (0.024) (0.063) (0.023) (0.025) Firm age (log) -0.188 -0.076 -0.073 -0.184 -0.075 -0.077
(0.139) (0.047) (0.049) (0.144) (0.048) (0.050) Total assets (log) 0.081 0.039** 0.027 0.060 0.033* 0.019
(0.053) (0.020) (0.019) (0.054) (0.019) (0.019) Net sales (log) 0.016 0.005 0.004 0.015 0.005 0.003
23
evident from the post-estimation test result. The Sanderson-Windmeijer test rejects the null of under-identification and weak instruments for endogenous regressor at 1% significance level for all models, except for model (2) at 5%. This suggests that female representation on boards increases when gender inequality in education access closer to parity. Furthermore, the results show that board size has a positive effect on the presence of women on boards at 1% significance level for the model (1)-(6). Also, total assets positively affect the number of female directors at 5% significance level for the model (2) and 10% for the model (5).
Table VII
Second stage regression: Board gender diversity and firm performance in Australia, India, Japan, Singapore, 2006-2015
This table presents the results of the second stage regression of instrumental variable approach with the inclusion of two-way fixed effects. Data for board characteristics are originated from Datastream Asset 4 ESG and annual report, while data for firm-level variables are from Datastream and Orbis. Columns 1,2,3 (Columns 4,5,6) show the results from return on equity (market to book value) regressed on the percentage of female directors, the number of females, and female indicator respectively, and the control variables. Return on equity is annual net income divided by the book value of shareholder equity. Market to book value equals the market value of equity divided by book value of equity. Percentage females equals the total number of female directors on the board divided by the total number of board of directors, number of females represents the total number of female directors, and the female indicator is a dichotomous variable that equals 1 when board has one or more female directors at the end of the year and 0 otherwise. Board size equals a total number of directors on the board, firm age is the total number of years since incorporation, total assets equal book value of total assets at the end of the year, net sales equal total net sales, leverage is total debt divided by total equity. For all variable, except female indicator, logarithm modulus transformation is used. ***, **, * indicate significance at the 1%, 5%, 10% levels respectively. Robust cluster standard errors to correct for heteroscedasticity and autocorrelation are shown in parentheses under the coefficient.
Return on equity (log) Market to book value (log) (1) (2) (3) (4) (5) (6) % females (log) -2.757** 0.264 (1.079) (0.175) # of females (log) -8.810** 0.8413 (3.835) (0.593) Female indicator -6.221*** 0.604 (2.260) (0.392) Board size (log) 0.509 0.990* 0.548 0.009 -0.037 0.006
(0.351) (0.575) (0.336) (0.056) (0.090) (0.057) Firm age (log) -1.105** -1.252** -1.038** -0.024 -0.009 -0.026
(0.487) (0.555) (0.407) (0.081) (0.087) (0.076) Total assets (log) 0.824*** 0.948*** 0.770*** -0.126** -0.138** -0.122**
(0.238) (0.299) (0.200) (0.061) (0.067) (0.059) Net sales (log) 0.223** 0.225** 0.200** -0.020 -0.020 -0.017
Table VII presents the results of the second-stage regression model. In line with expectations, it is shown that board gender diversity negatively affects ROE (β=-2.757; β=-8.810; β=-6.221) at 5% significance level for the model (1) and (2), and at 1% for the model (3). However, column (4)-(6) show no evidence of the impact of board gender diversity on market to book value (β=0.264; β=0.8413; β=0.604). This suggests that short-term financial firm performance deteriorates when the female representation on boards increases. Conversely, long-term performance does not depend on the presence of women directors on boards.
Second section of the analysis
In the first section, this study examines the link between board gender diversity and firm performance using the full sample from Australia, India, Japan, and Singapore. The result suggests that the presence of women on corporate boards has an adverse impact on short-term firm financial performance, indicating that there is enough evidence to support hypothesis 1.
In the second section of the analysis, the main objective is to test whether society’s perception towards women in economic roles can indeed explain the relation between board gender diversity and firm performance. At first, this study re-estimates the instrumental variables approach using the full-sample data, but now with the inclusion of EPO score as the additional control variables. This is to observe whether the inclusion of culture (EPO) may change the direction of the link between board gender diversity and firm performance. Secondly, this study estimates the instrumental variable approach as in the first section without the inclusion of EPO score but now based on two different subsamples: High and Low EPO to see the difference between the two settings.
result shows that EPO score negatively affect ROE in the model (1) and (2) at 10% significance level (β=-6.677; β=-7.744), but positively affect market to book value for the model (4) –(6) at 1% significance level (β=2.068, β=2.149; β=1.884). Table X and XI shows the result of instrumental variable approach for high EPO countries and low EPO countries respectively.
Table VIII
First stage regressions: Fitted value identification of board gender diversity
This table presents the results of the first stage regression of instrumental variable approach with the inclusion of two-way fixed effects and EPO score. The sample comprises Australia, India, Japan, and Singapore. Data for board characteristics are originated from Datastream Asset 4 ESG and annual report, while data for firm-level variables are from Datastream and Orbis, and the country-level data are from The Global Gender Gap report. Columns 1,2,3 (Columns 4,5,6) show the results from identifying the fitted values of the percentage of female directors, the number of females, and female indicator respectively, using educational attainment as the single instrument and control variables. Percentage females equals the total number of female directors on the board divided by the total number of board of directors, number of females represents the total number of female directors, and the female indicator is a dichotomous variable that equals 1 when board has one or more female directors at the end of the year and 0 otherwise. Educational attainment represents the country-level score of gender gap in education access, EPO equals a country score of Economic Participation and Opportunity subindex, board size equals total number of directors on the board, firm age is the total number of years since incorporation, total assets equals book value of total assets at the end of year, net sales equals total net sales, leverage is total debt divided by total equity. For all variable, except female indicator, educational attainment, and EPO score, logarithm modulus transformation is used. ***, **, * indicate significance at the 1%, 5%, 10% levels respectively. Robust cluster standard errors to correct for heteroscedasticity and autocorrelation are shown in parentheses under the coefficient. % of females (log) # of Females (log) Female indicator % of females (log) # of Females (log) Female indicator (1) (2) (3) (4) (5) (6) Educational attainment 7.786*** 2.476*** 3.328*** 7.878*** 2.508 *** 3.340 *** (2.388) (0.866) (0.977) (2.335) (0.850) (0.953) Board size (log) 0.293*** 0.147*** 0.134*** 0.324*** 0.158 *** 0.145 ***
(0.063) (0.023) (0.025) (0.064) (0.023) (0.025) Firm age (log) 0.004 -0.008 -0.012 -0.001 -0.011 -0.019
(0.136) (0.047) (0.048) (0.139) (0.047) (0.049) Total assets (log) 0.087* 0.042** 0.029 0.068 0.036* 0.022
(0.051) (0.019) (0.018) (0.052) (0.019) (0.018) Net sales (log) 0.012 0.004 0.002 0.011 0.004 0.002
26
Table X shows that there is a negative relationship between board gender diversity and return on equity at 10% significance level (β=-0.462; β=-1.198) in the model (1) and (2), but there is a positive relationship (β=8.065) in the model (3). However, the result of the model (3) cannot be inferred as it may be flawed due to an insignificant constant. In addition, the results show that there is a negative relationship between board gender diversity and market to book value at 1% significance level (β=-0.310; β=-0.803) in the model (4) and (5), but it is not significant
Table IX
Second stage regression: Board gender diversity and firm performance in Australia, India, Japan, Singapore with EPO score, 2006-2015
This table presents the results of the second stage regression of instrumental variable approach with the inclusion of two-way fixed effects and EPO score. Data for board characteristics are originated from Datastream Asset 4 ESG and annual report, while data for firm-level variables are from Datastream and Orbis. Columns 1,2,3 (Columns 4,5,6) show the results from return on equity (market to book value) regressed on the percentage of female directors, the number of females, and female indicator respectively, and the control variables. Return on equity is annual net income divided by the book value of shareholder equity. Market to book value equals the market value of equity divided by book value of equity. Percentage females equals the total number of female directors on the board divided by the total number of board of directors, number of females represents the total number of female directors, and the female indicator is a dichotomous variable that equals 1 when board has one or more female directors at the end of the year and 0 otherwise. EPO equals a country score of Economic Participation and Opportunity subindex. Board size equals a total number of directors on the board, firm age is the total number of years since incorporation, total assets equal book value of total assets at the end of the year, net sales equal total net sales, leverage is total debt divided by total equity. For all variable, except female indicator, logarithm modulus transformation is used. ***, **, * indicate significance at the 1%, 5%, 10% levels respectively. Robust cluster standard errors to correct for heteroscedasticity and autocorrelation are shown in parentheses under the coefficient.
Return on equity (log) Market to book value (log) (1) (2) (3) (4) (5) (6) % females (log) -2.507*** 0.195 (0.875) (0.142) # of females (log) -7.884*** 0.612 (3.002) (0.467) Female indicator -5.864*** 0.460 (1.970) (0.334) Board size (log) 0.478 0.904* 0.526* 0.018 -0.015 0.015
(0.313) (0.482) (0.312) (0.049) (0.075) (0.051) Firm age (log) -0.717* -0.787* -0.796** -0.138* -0.131* -0.129* (0.401) (0.427) (0.355) (0.072) (0.072) (0.071) Total assets (log) 0.816*** 0.925*** 0.768*** -0.126** -0.135** -0.123**
(0.221) (0.267) (0.193) (0.059) (0.063) (0.057) Net sales (log) 0.211** 0.211** 0.194** -0.016 -0.016 -0.015
Table X
Second stage regression: Board gender diversity and firm performance in countries characterized by High EPO score, 2006-2015
This table presents the results of the second stage regression of instrumental variable approach with the inclusion of two-way fixed (random) effects for model 1,2,4,5(3 and 6) in countries with high EPO score. The sample countries consist of Australia and Singapore. Data for board characteristics are originated from Datastream Asset 4 ESG and annual report, while data for firm-level variables are from Datastream and Orbis. Columns 1,2,3 (Columns 4,5,6) show the results from return on equity (market to book value) regressed on the percentage of female directors, the number of females, and female indicator respectively, and the control variables. Return on equity is annual net income divided by the book value of shareholder equity. Market to book value equals the market value of equity divided by book value of equity. Percentage females equals the total number of female directors on the board divided by the total number of board of directors, number of females represents the total number of female directors, and the female indicator is a dichotomous variable that equals 1 when board has one or more female directors at the end of the year and 0 otherwise. Board size equals a total number of directors on the board, firm age is the total number of years since incorporation, total assets equal book value of total assets at the end of the year, net sales equal total net sales, leverage is total debt divided by total equity. For all variable, except female indicator, logarithm modulus transformation is used. ***, **, * indicate significance at the 1%, 5%, 10% levels respectively. Robust cluster standard errors to correct for heteroscedasticity and autocorrelation are shown in parentheses under the coefficient.
Return on equity (log) Market to book value (log) (1) (2) (3) (4) (5) (6) % females (log) -0.462* -0.310*** (0.274) (0.090) # of females (log) -1.198* -0.803*** (0.714) (0.230) Female indicator 8.065** -3.821 (3.976) (0.529) Board size (log) 0.074 0.228 -2.817* 0.458*** 0.557*** 0.330***
(0.423) (0.476) (1.486) (0.117) (0.130) (0.197) Firm age (log) -0.280 -0.306 -3.250*** -0.164 -0.183 -0.125
(0.381) (0.383) (0.749) (0.113) (0.112) (0.119) Total assets (log) 0.817*** 0.813*** 0.314 -0.114* -0.114* -0.069
(0.153) (0.153) (0.241) (0.064) (0.064) (0.050) Net sales (log) 0.152** 0.151** 0.140 -0.015 -0.016 -0.019
(0.073) (0.073) (0.089) (0.012) (0.012) (0.013) Leverage (log) -0.791** -0.794** -0.499 0.610*** 0.603*** 0.604*** (0.358) (0.360) (0.413) (0.091) (0.091) (0.084) Constant -8.586*** -8.877*** 7.263 2.556*** 2.342*** 1.896*** (2.315) (2.323) (4.839) (0.898) (0.886) (0.651) # of observations 3,231 3,231 3,231 3,260 3,260 3,260 F-statistic 219.98*** 221.49*** 26.220*** 421.77*** 423.43*** 305.88*** Overall r2 0.102 0.105 0.018 0.003 0.004 0.012
Table XI
Second stage regression: Board gender diversity and firm performance in countries characterized by Low EPO score, 2006-2015
This table presents the results of the second stage regression of instrumental variable approach with the inclusion of two-way fixed effects in countries with low EPO score. The sample countries consist of India and Japan. Data for board characteristics are originated from Datastream Asset 4 ESG and annual report, while data for firm-level variables are from Datastream and Orbis. Columns 1,2,3 (Columns 4,5,6) show the results from return on equity (market to book value) regressed on the percentage of female directors, the number of females, and female indicator respectively, the control variables. Return on equity is annual net income divided by the book value of shareholder equity. Market to book value equals the market value of equity divided by book value of equity. Percentage females equals the total number of female directors on the board divided by the total number of board of directors, number of females represents the total number of female directors, and the female indicator is a dichotomous variable that equals 1 when board has one or more female directors at the end of the year and 0 otherwise. Board size equals a total number of directors on the board, firm age is the total number of years since incorporation, total assets equal book value of total assets at the end of the year, net sales equal total net sales, leverage is total debt divided by total equity. For all variable, except female indicator, logarithm modulus transformation is used. ***, **, * indicate significance at the 1%, 5%, 10% levels respectively. Robust cluster standard errors to correct for heteroscedasticity and autocorrelation are shown in parentheses under the coefficient.
Return on equity (log) Market to book value (log) (1) (2) (3) (4) (5) (6) % females (log) -1.660*** -0.205* (0.482) (0.117) # of females (log) -5.344*** -0.660* (1.695) (0.381) Female indicator -4.403*** -0.545* (1.343) (0.319) Board size (log) -0.015 0.091 0.094 0.057* 0.071* 0.071*
(0.152) (0.178) (0.176) (0.030) (0.036) (0.036) Firm age (log) 0.020 0.055 0.018 -0.073 -0.066 -0.077
(0.303) (0.321) (0.322) (0.121) (0.123) (0.118) Total assets (log) -0.303 -0.159 -0.383 -0.268*** -0.248** -0.275***
(0.286) (0.345) (0.288) (0.088) (0.100) (0.087) Net sales (log) 0.914*** 0.918*** 0.940*** 0.122* 0.122* 0.125**
(0.226) (0.233) (0.236) (0.063) (0.064) (0.064) Leverage (log) -1.749*** -1.750*** -1.718*** 0.253** 0.253** 0.254** (0.623) (0.646) (0.601) (0.101) (0.102) (0.101) Constant -5.106 -8.187*** -4.895 3.640*** 3.260** 3.680*** (3.358) (4.260) (3.384) (1.145) (1.305) (1.138) # of observations 4,197 4,197 4,197 4,224 4,224 4,224 F-statistic 335.20*** 315.30*** 315.56*** 1,036.7*** 991.25*** 987.48*** Centered r2 0.035 0.049 0.031 0.017 0.003 0.019
women in a high-level leadership position such as the board of directors. Catalyst (2005) find evidence for this as it shows that although women are perceived to have the capability in the workplace, they are still regarded as ineffective leaders when it comes to top-level positions. Hence, this negative stereotyping may negatively affect the interactions between male and female directors in the boardroom, thereby reducing firm performance.
In the case of low EPO countries, Table XI shows that board gender diversity negatively affects return on equity at 1% significance level for all models (β=-1.660; β=-5.344; β=-4.403), and it also negatively affects market to book value at 10% significance level (β=-0.205; β=-0.660; β=-0.545). In line with expectations, these findings support Hypothesis 2b that the effect of board gender diversity on firm performance is negative for companies located in countries characterized by negative society’s attitude toward women in economic roles.
Robustness Check
In this study, several robustness tests are performed6. Initially, all the regressions are re-estimated using two alternative measures of firm financial performance: ROA and
ROIC. Consistent with the reported findings, the first part of analysis shows that board gender diversity negatively affects firm performance in general, and these results still hold after controlling for culture. However, the second part of the analysis shows slightly different results with the reported results. In countries with positive society’s attitudes towards women at workplace, the results indicate that board gender diversity generally has no effect on firm financial performance. However, these results may be flawed due to an insignificant constant. Furthermore, in countries with a negative society’s perception towards women at workplace, the results consistent with reported findings indicating a negative link between board gender diversity and firm performance. All the robustness test results above hold for regression estimations both using ROA and ROIC. As previously discussed, this study applies logarithm modulus technique to do the data transformations. Hence, to add robustness of the findings, this study repeats the data transformations using natural logarithm, and all the regressions are replicated subsequently. Again, the results are generally still consistent with the reported findings.
VI. CONCLUSIONS
The benefits of board gender diversity on firm performance has been investigated for many years, yet no definite proof has been established. Therefore, to extend the existing literature, this study assesses the link between board gender diversity and firm performance in non-US and non-EU countries by employing data of four Asia-Pacific countries from 2006 to 2015. Additionally, this study deals with the endogeneity issue using instrumental variable approach and provides further insight into whether the country differences of society’s perception towards women in the workplace may explain the link between board gender diversity and firm performance.
process of socialization within the boards enabling women directors as the minorities to adapt and gain acceptance from the male counterparts in the long run, which leads to the improved decision-making process. Consequently, the loss from having women directors may not realize anymore in the long-term firm performance.
Given the cultural differences between country, society’s attitude towards women in the economic role may vary, which is argued to potentially bring different influences on how women directors are stereotyped in the boardroom. Accordingly, this study expects that even in those countries located in the same region, the link between board gender diversity and firm performance may be different depending on national culture. Thus, in the second part of the study, two subsamples are created, grouping the sample into countries with positive (Australia and Singapore), and negative (Japan and India) society’s attitudes towards women in the economic role. Surprisingly, the estimation results show that board gender diversity negatively affects firm performance in both settings. Considering the expected positive relationship, there are two possible explanations for a significant negative link between board gender diversity and firm performance in countries that more receptive towards women in the workplace. First, women directors in positive settings may be less tough and skillful than the ones in negative settings as they experience fewer challenges and requirements to reach top-level positions (Low et al., 2015). As such, they may bring limited contributions to the boards, thereby potentially reduce firm performance. Second, gender-stereotype bias, which hinders women contributions, may be more severe and difficult to eliminate in certain jobs such as executive positions (Catalyst, 2005). Although society may support female participation in the workplace, men are still widely seen as the ideal figures when it comes to leadership positions. Consequently, board gender diversity may not bring benefits to the company in this case.
Nevertheless, this study has several limitations that give rise to possibilities for further research. First, this study only employs four countries in Asia-Pacific due to limited board-level data availability in this region. Further studies should try expanding the sample by including more countries in Asia-Pacific to provide fewer distortions and better representation of the results. Second, this study only uses educational attainment as the single instrument variable to identify the fitted value of the presence of female directors, and it is a country-level factor. Hence, more instruments especially firm-level factor variables should be added to the model to capture exogenous variation in the explanatory variable better. Finally, this study does not find evidence that the country differences of society’s perception towards women in the economic role explain the link between board gender diversity and firm performance. Further research can clarify this finding by utilizing other potential proxies for society’s attitude towards women in the economic role beside Economic Participation and Opportunity (EPO) score.
REFERENCES
The Economist Corporate Network. (2016). Women in Leadership in Asia Pacific. The Economist Corporate Network Asia.
Abdullah, S. N. (2014). The causes of gender diversity in Malaysian large firms. J Manag Gov, 1137-1159.
Adams, R. B., & Ferreira, D. (2009 ). Women in the boardroom and their impact on governance and performance. Journal of Financial Economics 94, 291-309.
Bergeron, D. M., Block, C. J., & Echtenkamp, A. (2006). Disabling the Able: Stereotype Threat and Women's Work Performance. Human Performance , 133-158.
Bøhren, Ø., & Strøm, R. Ø. (2010). Governance and Politics: Regulating Independence and Diversity in the Board Room. Journal of Business Finance & Accounting, 1281–1308. Bonn, I., Yoshikawa, T., & Phan, P. H. (2004). Effects of board structure on firm performance: a comparison between Japan and Australia. Asian Business & Management, 105-125. Brammer, S., Millington, A., & Pavelin, S. (2007). Gender and Ethnic Diversity Among UK
Brannon, L. (2002). Gender: Psychological Perspectives. Boston: Allyn and Bacon .
Breuer, W., & Salzmann, A. J. (2012). National Culture and Corporate Governance . Corporate
Governance, 369-397.
Brooks, C. (2014). Introductory Econometrics for Finance 3rd Edition. Cambridge: Cambridge University Press.
Campbell, K., & Mínguez-Vera, A. (2008). Gender Diversity in the Boardroom and Firm Financial Performance. Journal of Business Ethics, 435-451.
Carter, D. A., D’Souza, F., Simkins, B. J., & Simpson, W. G. ( 2010). The Gender and Ethnic Diversity of US Boards and Board Committees and Firm Financial Performance.
Corporate Governance: An International Review, 396–414.
Carter, D. A., Simkins, B. J., & Simpson, W. G. (2003). Corporate Governance, Board Diversity, and Firm Value . The Financial Review 38 , 33-53.
Catalyst. (2005). Women "Take Care," Men "Take Charge" : Stereotyping of U.S. Business
Leaders Exposed. New York: Catalyst.
Cheung, H. Y., & Chan, A. W. (2007 ). How Culture Affects Female Inequality Across Countries: An Empirical Study. Journal of Studies in International Education, 157-179. Coad, A., Segarra, A., & Teruel, M. (2013). Like milk or wine: Does firm performance improve
with age? Structural Change and Economic Dynamics 24, 173-189.
Core, J. E., Guay, W. R., & Rusticus, T. O. (2006). Does Weak Governance Cause Weak Stock Returns? An Examination of Firm Operating Performance and Investors'Expectations .
The Journal of Finance , 655-687.
Daily, C. M., Certo, S. T., & Dalton, D. R. (1999). A Decade of Corporate Women: Some Progress in the Boardroom, None in the Executive Suite . Strategic Management
Journal, 93-99.
Darmadi, S. (2013). Do women in top management affect firm performance? Evidence from Indonesia. The international journal of business in society, 13(3), 288-304.
