Women on Boards in Portuguese Listed Companies: Does Gender Diversity Influence Financial Performance?

Abstract

Corporate sustainability integrates financial performance with environmental, social, and governance (ESG) performance. Due to their personal characteristics, women can play an important role in promoting sustainable actions and strategic decision-making, creating positive effects on a company’s financial performance. The main objective of this paper is to analyze the effect of Board gender diversity on financial performance, in the context of Portuguese listed companies. Financial performance was measured through an accounting-based measure, the return on assets (ROA), and a market-based measure, the Tobin’s Q. Board gender diversity was proxied through several measures identified in the literature and the information needed to construct them was manually collected from the corporate governance reports. A regression analysis was conducted on panel data consisting of 29 companies from the period from 2010 to 2019. The results show that female presence is positively related with ROA when there are at least two women on the Board, or when the proportion of women is, at the least, 20%. A positive relationship was also found between the Tobin’s Q and the presence of at least 20% women on the Board. The results confirm the existence of the positive effects of Board gender diversity on financial performance, as predicted by resource dependence theory and agency theory, but only when a critical mass of women is reached. In general, we conclude that, although gender diversity promotion measures underlying Portuguese legislation are not yet fully consolidated, the imposed threshold of 20% women on Boards is a signal sent to the market and one that is valued by investors.

1. Introduction

Boards of directors have an important role in corporate governance, as they are responsible for monitoring management; supervising and advising on strategic decision-making; preparing and disseminating information to the various stakeholders; and also for providing valuable contact networks with entities outside the company [1]. For this reason, a large body of research has analyzed the extent to which certain characteristics or practices of the Board have positive economic effects for companies, particularly in terms of the company’s performance and market value [2]. More recently, particular attention has been given to Board diversity, which can be defined as the variety inherent in the Board’s composition [3]. Because women are generally under-represented in “old boys’ networks”, Board gender diversity refers to the presence of more female directors on the Board and in other top management positions as Chief Executive Officer (CEO) or Chief Financial Officer (CFO). Diversity in terms of gender, age, nationality, or ethnicity allows for different opinions, perspectives, and experiences, more connections with entities outside the company, and problem-solving skills which may contribute to a greater ability to monitor and resolve conflicts of interest between different stakeholders [4]. Due to their personal characteristics, women are more concerned with social causes, environmental protection, and ethical issues [5]. For this reason, gender diversity in companies is an important driver for sustainable actions [6] and contributes to enhancing their reputation [7,8].

In 2012, the European Union presented a proposal for a Directive “on improving the gender balance among non-executive directors of companies listed on stock exchanges and related measures” [9]. Although its approval has not yet occurred, the Directive has triggered several legislative actions by member states. In Portugal, Law 62/2017 [10] mandated a gender quota for the Boards of state-owned companies and listed companies. For state-owned companies, the proportion of people of each gender appointed to each management and supervisory body may not be less than 33.3% from 1 January 2018, and this proportion shall apply to both executive and non-executive directors. For listed companies, that proportion may not be less than 20% as from the first elective general meeting after 1 January 2018 and 33.3% as from the first elective general meeting after 1 January 2020. The law provides for sanctions in case of non-compliance with the minimum thresholds. At the same time, the transposition into national law of Directive 2014/95/EU [11], through the Decree Law 89/2017 [12], amended the Portuguese securities code requirements on the corporate governance content, requiring a description of the diversity policy applied by the company in relation to its management and supervisory bodies, namely in terms of age, gender, qualifications, and professional background.

This study analyzes the effect of Board gender diversity on financial performance, based on a sample of Portuguese listed companies from the period from 2010 to 2019. For this purpose, several measures of gender diversity are tested and both accounting-based and market-based measures of financial performance are applied. This investigation seeks to make several contributions to the existing literature. First, because previous evidence on the relationship between Board gender diversity and financial performance is mixed [13], the use of the alternative measures of gender diversity and accounting-based and market-based measures of financial performance will make it possible to identify the types of relationships that are applied to the Portuguese context. Second, we provide empirical evidence on the Portuguese context, which has been scarcely analyzed. To the best of our knowledge, in Portugal, the effects of Board gender diversity on financial performance are documented using a limited set of gender variables, both for family and non-family listed companies, by Vieira [14], and for family-owned companies, by Fėlix and David [15]. Finally, we provide evidence based on a long and recent period covering the publication of the proposal of Directive on Board gender balance and the entry into force of the gender quota law in Portugal.

2. The Relationship between Board Gender Diversity and Financial Performance

2.1. Theoretical Framework

Several organizational, economic, sociological, and psychological theoretical approaches predict that a female presence in management positions and more gender diverse Boards lead to positive economic effects for companies [16,17,18]. Kirsch [19] presents two causal paths that can explain why Board gender composition, in particular the presence of female directors, may positively affect organizational outcomes. Firstly, Board gender composition affects the Board as a group, and this causality draws upon agency theory [20] and resource dependence theory [21,22]. Agency theory predicts that the presence of female directors provides the Board with greater insight and closer monitoring of management, because women are more active and more inclined to ask questions that would not be asked by male directors [23], and they tend to be more independent than their male counterparts, as they tend to be more often outside directors and not part of “old boys’ networks” [19]. Resource dependence theory is used to argue that diverse Boards have superior resources because female directors provide resources that male directors are not able to provide. Gender differences in values and traits may also affect Board processes and organizational outcomes, because women adopt more ethical, risk-averse, and long-term oriented points of view [19]. Considering that the presence of women on Boards enhances its independence and improves its efficiency, a positive relationship between Board gender diversity and financial performance is expected [24]. Additionally, if female directors improve management control and enhance the independence of corporate Boards, Board gender diversity may improve the transparency and disclosure quality and reduce earnings management practices [23].

The second causal pathway linking Board gender composition with organizational outcomes relates to how that composition, more specifically the presence of women, affects stakeholders inside and outside the company, or both [19]. This causality is based on signaling theory [25] and considers that appointing women to Boards signals legitimacy and adherence to social values, thus enhancing the company’s reputation [26,27]. This signal, sent by the company to their stakeholders, leads them to take certain actions (such as buying the company’s shares or products, promoting female managers, initiating acquisitions, or working for the company) that, predictably, have a positive effect on financial performance [28]. There is evidence that the market reacts positively to the announcement of female Board appointments, suggesting that investors on average believe that female directors add value to the company [3]. The presence of women on Boards also has a positive effect on the quantity and quality of public disclosure and therefore companies with higher gender diversity on their Boards show lower levels of information asymmetry in the market [23,29]. If gender diversity leads to an improvement in information disclosure, and greater transparency and disclosure reduces information risk for market participants, a positive effect of Board gender diversity on stock prices and market value is expected.

The causal paths presented consider that Board diversity is beneficial because there is “value-in-diversity”, but a diversified composition of the Board can be viewed negatively, i.e., “diversity-as-process-loss” [19]. On the one hand, diverse groups can make better decisions because they consider many points of view, which generates more ideas and improves creativity, innovation, and adaptability, resulting in better decisions. On the other hand, diverse groups can be more conflictual, have trouble communicating, and have more divergence of opinions; therefore, decision-making will usually be more time-consuming and less effective [30].

Additionally, according to critical mass theory, minority gender (women) directors cannot fully play their role in a group (a Board) until a certain threshold or critical mass of women in the group is reached [31,32]. When only a marginal number of women directors are present, they are treated as female representatives (“tokens” or “symbols”), bringing about tokenism issues, such as stereotyping, suspicion, isolation, and exclusion, which discourage the women directors and the incumbent men directors to speak and perform in Board interaction processes. Critical mass theory posits that once a certain threshold or critical mass of women in the Board is reached, their presence becomes normalized. In this context, gender-unbalanced Boards are more problematic and even less efficient than more gender uniform ones [31,32]. The critical threshold often pointed to is of three women or around 30% of the group [33,34].

The negative effects of a diversified Board composition could also arise when the decision to appoint women to certain positions is motivated by societal pressure for greater equality between sexes, i.e., they are hired as “tokens” and not because of their competence. If men perceive women as “tokens”, it could be very possible that women will be marginalized and their opinions will not be considered [31,32]. This can lead to conflicts among Board members and a loss of effectiveness of decisions, with consequential negative economic effects. The presence of women is also associated with negative economic effects when they are appointed to Boards of companies experiencing periods of underperformance or other turmoil, a phenomenon known as the “glass cliff” [35,36].

2.2. Empirical Evidence

The literature on Board gender composition has expanded significantly in recent years, with the literature analyzing the effects of the female presence on corporate Boards in aspects such as decision-making, risk-taking, financial and non-financial performance, market value, corporate social responsibility, transparency, and disclosure (see [4,16,17,18,19,37,38]). Two factors may explain this growing trend. First, despite the rapid increase in female participation in the labor force during the 20th century, the “glass ceiling effect” still exists, with women remaining underrepresented in leading positions, including on corporate Boards [39]. Second, following the Directive proposal, several countries have introduced regulatory requirements to increase the number of women on corporate Boards [40]. In this context, research on Board gender diversity economic effects is needed to empirically test the theoretical reasons given for increasing female presence in corporate management positions and to assist regulators in gender policy-making.

In this study, we focus on the Board gender diversity. However, the effects of gender diversity on financial performance have also been studied, considering the presence of a woman in the chair position and in other top management positions, such as CEO and CFO (e.g., [41,42,43,44]). More recently, along with gender diversity, other characteristics of women have been analyzed, such as independence, multiple directorships, nationality, tenure, age, education level, and business education (e.g., [45]).

The United States of America has often been pointed out as the context where the effects of Board gender diversity have been most studied, as it is there that research on the role of the Board of directors in corporate governance has flourished [13,46,47,48]. However, research on the effects of Board gender diversity on financial performance has spread out to other settings such as Europe (see Appendix A), Asia (e.g., [49,50]), Africa (e.g., [44]), Latin America (e.g., [51]), and Russia (e.g., [52]). Considering the scope of our study, we focus on the empirical evidence from European countries, which is summarized in the Appendix A. The studies reveal mixed results concerning the effect of Board gender diversity on financial performance, showing positive, negative, or no effect. We consider that this may be explained by two reasons. The first reason concerns the differences in the time periods, countries, and/or companies studied that represent the institutional contexts with different agency problems, information asymmetries, or resource needs. The period “before” or “after” the introduction of a recommendation on Board gender diversity or a law on women’s quota may also influence the relationship between female presence on the Board and financial performance, with studies finding a positive relationship “before” and a negative relationship “after” [53], in Italy) or no effect “before” and a positive effect “after” but only for highly performing companies, suggesting that the costs of appointing more women could be high for low performing companies [54], in France). In Romania, Ionascu et al. [55] also found differences in Board gender diversity effects according to a company’s profitability, observing a positive effect only for profitable, small, and less well governed companies. In Portugal, Vieira [14] found a positive effect of female presence on Board on financial performance for family firms and a negative relationship for non-family firms. In specific contexts, the market reacts negatively to higher proportions of women on Boards, stating that more female presence means more expensive and less experienced Boards [56,57] in Norway; [54] in France). For a similar period and in the same country, the United Kingdom (UK), a positive relationship between Board gender diversity and financial performance was found for listed companies by Brahma et al. [45] and a negative relationship was found for small and medium sized companies (SME) by Shehata et al. [58]. Finally, recent evidence has shown that the economic effects of female presence are evident only when women have an active involvement in corporate governance, occupying decision-making positions such as CEO [59,60].

The second reason why empirical evidence on the effects of Board gender diversity is mixed may be related to the use of different measures of financial performance and gender diversity [13]. Regarding financial performance, the most popular measurements are accounting-based measures and market-based measures for the categories of the return on assets (ROA) and the Tobin’s Q, respectively. Tobin’s Q is defined as the market value of the company’s assets divided by their replacement cost. However, researchers commonly calculate Tobin’s Q as the sum of the market value of equity and the book value of debt divided by the book value of total assets. A value of Tobin’s Q which is greater than one reflects investors’ confidence in the company and its growth opportunities. Market-based measures capture the market expectations about the ability of the company to generate future cash flows and its associated risk. Accounting-based measures reflect past events and only those reflected in accounting. For this reason, diverse results can be found depending on whether an accounting-based or a market-based measure of financial performance is used (e.g., [42] in France; [14] in Portugal). Because it is unclear when the effects of the appointment of women to Boards become apparent [19], these measures should be combined and carefully analyzed, considering that changes in Board composition tend to be captured by accounting with some delay, while market-based measures may anticipate a future performance different from what will occur.

In relation to gender diversity measures, some studies use only one measure (e.g., [41] in France; [14,15] in Portugal), while others use several (e.g., [54] in France; [58] in the UK). The “total number” and the “proportion” of women on Boards consider a linear relationship between gender diversity and the economic effect. The “presence of at least X women on the Board” (gender “dummies”) forms a proxy of the critical mass that is needed to achieve the desired economic outcome. The Blau index and the Shannon index capture the balance between the two genders, being both qualitatively similar, but the Shannon index is a logarithmic measure of diversity, and thus is more sensitive to minor differences in Boards’ gender diversity. The use of more than one proxy may shed light on the most efficient level of gender diversity, as the relationships established may not be linear. For example, in the UK, Brahma et al. [45] only found a positive effect of financial performance for the presence of three or more women on the Board. More recent studies test the presence of a possible quadratic relationship between financial performance and gender diversity, using the square of the proportion of women, the Blau index, and the Shannon index ([61] in Spain; [59] in Turkey).

Considering the theoretical arguments that underlie the “value-in-diversity” perspective and the empirical evidence presented above, we formulate the following hypotheses:

Hypothesis 1 (H1).

The proportion of women on the Boards of Portuguese listed companies is positively related with financial performance.

Hypothesis 2 (H2).

The presence of women on the Boards of Portuguese listed companies is positively related with financial performance only when a critical mass of female presence is reached.

3. Research Methodology

3.1. Sample Selection and Data Collection

The empirical analysis is based on data from three sources: SABI database (Iberian Balance Sheet Analysis System), for economic and financial information; corporate governance reports, from which data about Board characteristics were hand-collected; and companies’ market data, available online. The initial sample consisted of 55 listed Portuguese companies included in the SABI database, with non-consolidated accounts for the period 2010 to 2019. The financial sector, football clubs, and companies with a fiscal year not ending on 31 December were excluded, due to the lack of comparability of accounting information. Companies with negative equity were also removed because extreme financial constrains might compromise the relationships tested. After this process, only 29 companies had corporate governance reports available, making a total of 290 observations. The regressions with Tobin’s Q as the dependent variable were carried out for only 250 observations, corresponding to 25 companies, due to lack of stock market data for some of the company–year observations.

Table 1. Board gender diversity over the years of the sample.

	Total Number Members of the Board			Total Number Women on the Board			Proportion of Women on Board			Chair Is Woman	Boards with at Least 1 Woman	Boards with at Least 2 Women	Boards with at Least 3 Women	Boards with at Least 20% Women	Boards with at Least 33% Women
Years	Min.	Median	Max.	Min.	Median	Max.	Min.	Mean	Max.	Number of Companies
2010	3	9	20	0	0	4	0%	7%	27%	0	14	2	1	3	0
2011	3	9	24	0	1	4	0%	7%	31%	0	15	3	1	3	0
2012	3	9	24	0	1	4	0%	9%	40%	0	18	5	1	4	1
2013	3	9	21	0	1	5	0%	10%	40%	0	19	5	2	6	2
2014	3	8	20	0	1	5	0%	11%	44%	0	18	7	3	5	3
2015	3	8	19	0	1	5	0%	13%	43%	0	18	9	4	8	4
2016	3	8	19	0	1	4	0%	15%	43%	1	19	10	7	10	5
2017	3	8	19	0	1	4	0%	15%	43%	1	19	11	7	10	5
2018	3	8	21	0	2	5	0%	19%	43%	1	24	17	8	18	4
2019	3	9	21	0	2	5	0%	22%	33%	1	27	21	11	24	1

presents an overview of gender diversity in the Boardrooms over the years of the sample. The Board size ranges from 3 members to 24 members and the number of women on Boards from 0 (zero) to a maximum of 5. Only one company in the sample has a woman as a chair, from 2016. We can see a positive evolution of Board gender diversity over the years, but in 2019, the most recent year of the sample, there are still companies with no female directors in their Boards. The size of the Boards has remained stable over the years, which is a similar scenario to that documented in Spain by Martínez and Rambaud [61]. After the enactment of the gender law in 2017, the number of women on Boards increased, on average, from one to two. In 2019, the 20% threshold of the gender quota law is exceeded only by 24 companies (from the 29 analyzed), which may be explained by the fact that the law only applies to appointments from 2018 onwards.

3.2. Model and Variables

The effects of Board gender diversity on financial performance are tested through a multivariate regression model, with the following specification:

Financial Performance_i,t = β₀ + β₁ × Gender_i,t + β₂ × Controls_i,t + β₃ × Year_i,t + ε_i,t

We chose two measures of financial performance for our dependent variable: ROA, as accounting-based measure; and Tobin’s Q as market-based. The independent variables include the Board gender diversity proxies (Gender) and an array of control variables (Controls) identified in previous studies (e.g., [14,45,62]). The gender proxies are included in the models one at a time and consider a set of “continuous” gender variables that capture the proportion of women on Boards, for testing Hypothesis 1; and “gender dummies” for testing Hypothesis 2. The models also include year dummy variables (Year) and the error term (ε_i,t) of the equation to be estimated. The definitions of all variables are presented in

Table 2. Variables definitions.

Dependent Variable (Financial Performance)
ROA	ROA is measured as the ratio of net income to average total assets during a financial year.
TOBIN	Tobin’s Q is defined as the sum of the market value of equity (share price multiplied by the number of ordinary shares in issue at fiscal year-end) and the book value of debt divided by book value of total assets.
Gender Variables
PWOM	Proportion of women on the Board, calculated by the ratio between the number of women and the total of members on the Board.
BLAU	Blau [63] Index, defined as: $Blau Index=1-{\displaystyle {\displaystyle \sum }_{i=1}^K}{P}_i^2$
	where K is the number of gender categories, i.e., 2 (women and men) and Pi is the percentage of each category. This variable ranges between 0 (the Board has only men or women) and 0.5 (there are the same number of men and women on the Board).
SHAN	Shannon [64] Index, defined as: $Shannon Index=-{\displaystyle {\displaystyle \sum }_{i=1}^K}{P}_i\times \ln (P_i)$
	where K is the number of gender categories, in this case two (women and men), and Pi is the percentage of each category. For companies with only women or only men on the Boards, the value of ln(Pi) is forced to be zero, because ln(0) equals infinity. This variable ranges between 0 (the Board has only men or women) and 0.69 (there are the same number of men and women on the Board).
1WOM	Dummy variable coded 1 if the company has at least 1 woman on the Board and 0 otherwise.
2WOM	Dummy variable coded 1 if the company has at least 2 women on the Board and 0 otherwise.
3WOM	Dummy variable coded 1 if the company has at least 3 women on the Board and 0 otherwise.
20%W	Dummy variable coded 1 if the company has at least a percentage of 20% of women on the Board and coded 0 otherwise.
33%W	Dummy variable coded 1 if the company has at least a percentage of 33% of women on the Board and coded 0 otherwise.
Control Variables
BOARD	Board size measured by the natural logarithm of total members.
SIZE	Company size measured by the natural logarithm of total assets.
LEV	Leverage ratio of total liabilities to total assets.

.

Considering that our data consist of a balanced panel, it is recommended to control for company-specific and time-invariant unobserved heterogeneity through fixed effects model (FEM) or random effects model (REM). Thus, for each regression, adequate tests are carried out to decide which estimation method produces the most efficient estimators: pooled ordinary least squares (OLS), fixed effects model (FEM), or random effects model (REM).

4. Empirical Findings and Discussion

4.1. Descriptive and Correlations Analysis

The descriptive statistics of the continuous variables to be included in the regressions are presented in

Table 3. Descriptive statistics.

	Min.	Mean	Max.	Standard Deviation
ROA	−2.912	0.017	0.554	0.245
TOBIN	0.003	1.154	7.001	1.119
PWOM	0.000	0.128	0.444	0.121
BLAU	0.000	0.195	0.494	0.165
SHAN	0.000	0.133	0.298	0.106
BOARD	0.477	0.927	1.380	0.198
SIZE	4.648	5.764	7.383	0.641
LEV	0.001	0.419	0.871	0.224

. ROA is, on average, close to zero because some companies have a high negative net income. The statistics for Tobin’s Q corroborate that most of the companies have a market value above their book value. As for the continuous gender diversity variables, the minimum values for the proportion of women (PMUL), Blau index (BLAU), and Shannon Index (SHANNON) corroborate the existence of companies without female directors, at least in one of the years of the period analyzed. Concerning the control variables, the only noteworthy fact is that, on average, the leverage ratio is around 42%, with some observations where this ratio reaches a maximum of 87% revealing the existence of highly leveraged firms.

After ensuring that the normality assumptions hold for the continuous variables, an independent samples t-test was conducted to compare the differences between the financial performance of companies, according to the gender diversity of their Boards. The results are presented in

Table 4. Differences in financial performance based on gender diversity measures.

		ROA				TOBIN
		N	Mean	T	Sigma	N	Mean	T	Sigma
PWOM BLAU SHAN	Above median	145	0.0469	−2.084	0.038 **	125	1.1369	0.016	0.986
	Below median	145	−0.0145			125	1.1392
1WOM	Yes	191	0.0236	−0.767	0.443	161	1.1913	−1.098	0.273
	No	99	0.0049			89	1.0361
2WOM	Yes	90	0.0549	−2.416	0.016 **	71	1.2541	−1.197	0.232
	No	200	0.0003			179	1.0858
3WOM	Yes	45	0.0566	−2.349	0.019 **	35	1.1486	−0.093	0.925
	No	245	0.0101			215	1.1361
20%W	Yes	91	0.0604	−2.691	0.007 *	73	1.3174	−1.903	0.058 ***
	No	199	−0.0024			177	1.0589
33%W	Yes	25	0.0475	−1.714	0.088 ***	19	1.3299	−0.849	0.404
	No	265	0.0144			231	1.1225

Variables definitions as per Table 2. *, **, and *** represent statistical significance at 1%, 5%, and 10%, respectively.

. For the continuous gender variables, the sample was divided according to the median of the gender variable (companies whose proportion of women, Blau index, or Shannon index is above or below the respective median for the year). For gender dummies, the sample was divided according to their definition. The results show statistically significant differences for ROA in all the groups analyzed (except for 1WOM), suggesting that more gender diverse companies have a higher ROA. The Tobin’s Q exhibits statistically significant differences only when comparing companies with at least 20% women (20%W) and the rest, with a higher Tobin’s Q for more gender diverse companies.

The correlations among the variables to be included in the regressions are presented in

Table 5. Correlation matrix (Pearson Correlations).

	ROA	TOBIN	BOARD	SIZE	LEV
ROA	1
TOBIN	0.195 *	1
BOARD	0.020	0.087	1
SIZE	0.106 ***	0.129 **	0.486 *	1
LEV	0.074	−0.019	0.348 *	0.342 *	1
PWOM	0.102 ***	0.074	0.010	−0.058	−0.112 **
BLAU	0.099 ***	0.072	0.053	−0.064	−0.105 ***
SHAN	0.089	0.071	0.091	−0.064	−0.108 ***

Variables definitions as per Table 2. *, **, and *** represent statistical significance at 1%, 5%, and 10%, respectively.

. The coefficients confirm a statistically significant and positive association between ROA and all gender variables, except for the Shannon index and the presence of at least 33% women (33%W); and between Tobin’s Q and the presence of at least 20% women (20%W). The correlations between the independent variables reveal low correlation coefficients with the highest being 0.486, suggesting that multicollinearity does not pose an issue in this study.

4.2. Regression Analysis

Table 6. Estimation results.

Panel A—Financial Performance Measured by ROA
	PWOM	BLAU	SHAN	1WOM	2WOM	3WOM	20%W	33%W
	FEM	FEM	FEM	FEM	FEM	FEM	FEM	FEM
Intercept	−2.740 * (0.438)	−2.751 * (0.440)	−2.748 * (0.441)	−2.718 * (0.439)	−2.732 * (0.436)	−2.761 * (0.447)	−2.749 * (0.434)	−2.714 * (0.439)
Gender	0.157 (0.182)	0.114 (0.140)	0.142 (0.222)	−0.004 (0.050)	0.077 *** (0.043)	−0.028 (0.055)	0.094 ** (0.039)	0.034 (0.063)
BOARD	0.485 (0.136)	−0.061 (0.136)	−0.065 (0.136)	−0.053 (0.142)	−0.103 (0.137)	−0.044 (0.137)	−0.023 (0.135)	−0.029 (0.144)
SIZE	0.485 * (0.082)	0.488 * (0.082)	0.488 * (0.082)	0.483 * (0.083)	0.492 (0.082)	0.491 * (0.084)	0.479 * (0.082)	0.479 * (0.083)
LEV	0.099 (0.129)	0.099 (0.128)	0.096 (0.128)	0.092 (0.128)	0.103 (0.127)	0.078 (0.131)	0.139 (0.129)	0.094 (0.128)
N	290	290	290	290	290	290	290	290
Adj. R2	8.24%	8.21%	8.12%	7.97%	9.12%	8.06%	10.09%	8.07%
F-Test	5.087 *	5.085 *	5.096 *	5.194 *	5.272 *	5.185 *	5.370 *	5.233 *
Haussman test	44.264 *	44.635 *	44.380 *	42.788 *	51.193 *	51.585 *	46.064 *	42.652 *
	Panel B—Financial performance measured by Tobin’s Q
	PWOM	BLAU	SHAN	1WOM	2WOM	3WOM	20%W	33%W
	REM	REM	REM	REM	REM	REM	REM	REM
Intercept	−0.571 (1.099)	−0.594 (1.101)	−0.595 (1.102)	−0.545 (1.101)	−0.509 (1.093)	−0.772 (1.118)	−0.629 (1.079)	−0.491 (1.096)
Gender	0.520 (0.616)	0.360 (0.457)	0.512 (0.717)	0.061 (0.155)	0.127 (0.140)	−0.212 (0.184)	0.468 * (0.130)	0.232 (0.225)
BOARD	0.391 (0.463)	0.366 (0.463)	0.351 (0.465)	0.333 (0.482)	0.288 (0.474)	0.470 (0.468)	0.574 (0.455)	0.525 (0.481)
SIZE	0.081 (0.209)	0.088 (0.209)	0.092 (0.209)	0.089 (0.209)	0.093 (0.208)	0.119 (0.211)	0.047 *** (0.205)	0.052 (0.211)
LEV	0.219 (0.367)	0.215 (0.367)	0.208 (0.367)	0.185 (0.365)	0.201 (0.365)	0.088 (0.373)	0.408 (0.361)	0.209 (0.365)
N	250	250	250	250	250	250	250	250
Adj. R2	27.00%	26.97%	26.94%	26.84%	26.98%	27.21%	30.65%	27.09%
F-Test	18.702 *	18.674 *	18.684 *	18.755 *	18.563 *	18.824 *	19.687 *	18.947 *
Haussman test	0.958	0.960	0.939	0.911	1.402	0.763	1.527	1.462

Variables definitions as per Table 2. Each column presents estimation results for the respective gender proxy and for the most efficient model, based on the F statistic and the Hausman test. *, **, and *** represent statistical significance at 1%, 5%, and 10%, respectively. Robust standard errors in parenthesis.

reports the results of the regression analysis, considering ROA and Tobin’s Q as dependent variables in Panel A and Panel B, respectively. The regression results presented correspond to the estimation method identified as the most efficient, based on the F statistic and the Hausman test: Fixed Effects Model (FEM) for ROA and the Random Effects Model (REM) for Tobin’s Q. The adjusted R² is higher in the models with Tobin’s Q than in the models with ROA as dependent variables, suggesting that gender diversity collectively with control variables have greater explanatory power for market-based performance compared to accounting-based performance.

Regarding the effect of gender diversity on ROA (Table 6—Panel A), only the presence of at least two women (2WOM) and the presence of at least 20% women (20%W) on Boards show statistically significant coefficients, with positive signs, as expected. This result leads us to conclude that the effect of gender diversity on financial performance is not linear, because the proportion of women (PWOM) and the Blau and Shannon indexes (BLAU, SHAN) do not present statistically significant coefficients. In Portugal, for listed companies from the period from 2002–2013, Vieira [14] concluded that the proportion of women was positively related to ROA in family firms, and negatively related to ROA in non-family firms. This may be justified by the fact that in family firms, women are family members, with a long-term business outlook and the aim of maximizing the firm’s wealth in the long term [14], and in non-family firms, they may be merely seen as “tokens”. However, when considering the whole sample, no statistically significant relationship was found, which is consistent with our results.

Concerning the gender dummies variables, our findings are consistent with critical mass theory [31,32], confirming that gender diversity only influences financial performance when a critical mass of female presence is reached, or more precisely, two women or 20% women. The absence of statistically significant coefficients for the presence of at least one woman (1WOM) confirms that, unless there are two or more women on the Boardroom, a female presence is unlikely to motivate significant improvement in financial performance [33,34]. The lack of significant results for the presence of 33% of women (33%W) confirms that, in our context, the optimal number of women on Boards is of two or 20%, contrasting, for example, with Brahma et al. [45] in the UK, whose results concluded that the critical mass needed was of three women.

For the effect of gender diversity on Tobin’s Q (Table 6—Panel B), only the variable concerning the presence of at least 20% women (20%W) reports statistically significant coefficients, which are also positive. Considering that Tobin’s Q captures market expectations about companies’ future cash flows and risk, our results indicate that investors assign higher value to companies that have already reached the target of 20% imposed by the quota law, for two reasons. First, the presence of women improves economic performance and consequently contributes to higher future cash flows. Second, these companies pose less risk as they will not incur penalties for non-compliance with the law or reputational losses for being considered discriminatory or non-compliant. The 20% threshold for the proportion of women on the Board signals an adherence to social values and compliance with legal provisions, which in turn enhances the firm’s reputation [26,27].

Regarding the control variables, SIZE presents positive and statistically significant coefficients in most estimations, corroborating that financial performance is positively influenced by the company size in the Portuguese context [14]. The remaining control variables do not show statistically significant coefficients.

Thus, the results obtained lead to the rejection of hypothesis 1, which predicted a linear relationship between Board gender diversity and financial performance, and to the confirmation of hypothesis 2, for a critical mass of 20% for both measures of financial performance.

In line with previous studies conducted in contexts with a low presence of women on Boards, as in Portugal, we have not found statistically significant relationships between financial performance and the proportion of women on Boards, the Blau index, and the Shannon index [54,59,65,66]. Moreover, the presence of at least one woman on the Board does not seem to influence the financial performance of Portuguese companies, similarly to other studies [45,62,66,67,68]. In fact, few studies, and most of them carried out in specific contexts, have documented a positive effect on financial performance for the presence of at least one woman on the Board ([69] in banks; [15] for family firms; [70] only for ROE; [71] for SMEs). The proportion of women, the Blau index, the Shannon index, and the presence of at least one woman on the Board are measures frequently used in previous studies, some of which have concluded for the non-existence of a relationship between gender diversity and financial performance, on the basis of these measures alone [59,65,66]. Our results demonstrate the importance of using alternative measures of gender diversity as a means of identifying the optimal level of gender diversity for the context analyzed.

4.3. Robustness Tests

To assess the robustness of the results, the model was estimated with two modifications. The first consisted in introducing as independent variables lagged financial performance variables (1 year lag for ROA and Tobin’s Q), because managers and investors may adopt strategies based on past performance that may affect future performance [72]. In the second modification, an alternative accounting measure of financial performance was used—return on equity (ROE)—computed as the ratio of net income to equity [41,42,53,68,70]. The results, presented in Appendix B, in

Table A2. Models with one year lagged financial performance as independent variable.

Panel A—Financial Performance Measured by ROA
	PWOM	BLAU	SHAN	1WOM	2WOM	3WOM	20%W	33%W
	FEM	FEM	FEM	FEM	FEM	FEM	FEM	FEM
Intercept	−3.845 * (0.540)	−3.854 * (0.544)	−3.842 * (0.546)	−3.771 * (0.543)	−3.864 * (0.535)	−3.824 * (0.540)	−4.009 * (0.536)	−3.777 * (0.535)
Gender	0.181 (0.192)	0.125 (0.149)	0.156 (0.238)	−0.001 (0.054)	0.078 *** (0.045)	−0.039 (0.056)	0.109 ** (0.042)	0.058 (0.066)
Previous Performance	−0.362 * (0.066)	−0.361 * (0.067)	−0.361 * (0.067)	−0.361 * (0.067)	−0.358 * (0.067)	−0.363 * (0.067)	−0.368 * (0.066)	−0.364 * (0.067)
BOARD	0.007 (0.142)	−0.005 (0.142)	−0.010 (0.142)	−0.001 (0.150)	−0.044 (0.144)	0.018 (0.144)	0.051 (0.141)	0.048 (0.152)
SIZE	0.665 * (0.100)	0.668 * (0.101)	0.668 * (0.100)	0.661 * (0.101)	0.067 * (0.100)	0.671 * (0.101)	0.067 * (0.099)	0.653 * (0.101)
LEV	0.101 (0.140)	0.100 (0.140)	0.097 (0.141)	0.097 (0.141)	0.100 (0.140)	0.078 (0.143)	0.131 (0.139)	0.101 (0.141)
N	261	261	261	261	261	261	261	261
Adj. R2	12.93%	12.86%	12.75%	12.57%	13.72%	12.77%	15.18%	12.88%
F-Test	5.811 *	5.797 *	5.802 *	5.878 *	5.935 *	5.879 *	6.147 *	5.938 *
Haussman test	41.022 *	39.010 *	42.631 *	68.821 *	75.006 *	38.626 *	84.558 *	67.416 *
	Panel B—Financial performance measured by Tobin’s Q
	PWOM	BLAU	SHAN	1WOM	2WOM	3WOM	20%W	33%W
	FEM	FEM	FEM	FEM	FEM	FEM	FEM	FEM
Intercept	4.220 * (1.364)	4.192 * (1.369)	4.196 * (1.372)	4.256 * (1.368)	4.254 * (1.364)	4.153 * (1.388)	3.993 * (1.325)	4.323 * (1.360)
Gender	0.381 (0.583)	0.253 (0.439)	0.325 (0.696)	0.022 (0.154)	0.041 (0.129)	−0.076 (0.169)	0.423 * (0.124)	0.214 (0.209)
Previous Performance	0.371 * (0.070)	0.372 * (0.070)	0.373 * (0.070)	0.371 * (0.070)	0.371 * (0.070)	0.367 * (0.071)	0.348 * (0.068)	0.365 * (0.070)
BOARD	0.314 (0.458)	0.294 (0.459)	0.284 (0.462)	0.289 (0.485)	0.281 (0.469)	0.335 (0.462)	0.559 (0.451)	0.450 (0.477)
SIZE	−0.695 * (0.263)	−0.687 ** (0.264)	−0.685 ** (0.265)	−0.692 * (0.266)	−0.690 * (0.264)	−0.677 ** (0.267)	−0.706 * (0.256)	−0.730 * (0.265)
LEV	1.118 * (0.393)	1.112 * (0.393)	1.105 * (0.392)	1.094 * (0.393)	1.101 * (0.392)	1.055 * (0.404)	1.271 * (0.384)	1.118 * (0.392)
N	225	225	225	225	225	225	225	225
Adj. R2	26.74%	26.70%	26.66%	26.58%	26.61%	26.65%	30.88%	26.98%
F-Test	3.307 *	3.294 *	3.287 *	3.285 *	3.288 *	3.321 *	3.821 *	3.375 *
Haussman test	85.143 *	84.871 *	84.684 *	84.571 *	84.736 *	85.130 *	101.97 *	86.575 *

Variables definitions as per Table 2. Previous performance corresponds to lagged financial performance variables (1 year lag for ROA in Panel A; 1 year lag for Tobin’s Q in Panel B). Each column presents estimation results for the respective gender proxy and for the most efficient model, based on the F statistic and the Hausman test. *, **, and *** represent statistical significance at 1%, 5%, and 10%, respectively. Robust standard errors in parenthesis.

and

Table A3. Alternative accounting-based measure of financial performance.

Financial Performance Measured by ROE
	PWOM	BLAU	SHAN	1WOM	2WOM	3WOM	20%W	33%W
	FEM	FEM	FEM	FEM	FEM	FEM	FEM	FEM
Intercept	−4.295 * (0.573)	−4.311 * (0.574)	−4.309 * (0.575)	−4.267 * (0.574)	−4.285 * (0.568)	−4.384 * (0.582)	−4.302 * (0.568)	−4.259 * (0.573)
Gender	0.223 (0.237)	0.165 (0.183)	0.217 (0.291)	0.002 (0.065)	0.121 ** (0.056)	−0.077 (0.072)	0.110 ** (0.050)	0.038 (0.083)
BOARD	0.004 (0.177)	−0.011 (0.177)	−0.017 (0.177)	−0.005 (0.185)	−0.077 (0.178)	−0.032 (0.179)	−0.035 (0.176)	−0.026 (0.189)
SIZE	0.770 * (0.108)	0.775 * (0.108)	0.775 * (0.108)	0.770 * (0.108)	0.782 * (0.107)	0.787 * (0.109)	0.763 * (0.107)	0.763 * (0.109)
LEV	−0.108 (0.168)	−0.109 (0.168)	−0.113 (0.168)	−0.119 (0.168)	−0.102 (0.167)	−0.157 (0.171)	−0.064 (0.168)	−0.116 (0.168)
N	290	290	290	290	290	290	290	290
Adj. R2	11.70%	11.68%	11.59%	23.96%	13.00%	11.82%	13.03%	11.47%
F-Test	5.228 *	5.194 *	5.207 *	5.379 *	5.586 *	5.699 *	5.512 *	5.674 *
Haussman test	50.325 *	51.072 *	50.970 *	48.388 *	60.390 *	59.881 *	63.195 *	45.08 *

ROE is return on equity, measured as the ratio of net income to equity during a financial year. Variables definitions as per Table 2. Each column presents estimation results for the respective gender proxy and for the most efficient model, based on the F statistic and the Hausman test. *, and **, represent statistical significance at 1%, and 5%, respectively. Robust standard errors in parenthesis.

, respectively, do not change the previous conclusions regarding the effect of gender diversity on financial performance. For the models with accounting-based measures of financial performance as dependent variables (ROA in Table A2—Panel A; ROE in Table A3), only the presence of at least two women (2WOM) and the presence of at least 20% women (20%W) on Boards show statistically significant coefficients, with positive signs. For the model with Tobin’s Q as the dependent variable (Table A2—Panel B), only the presence of at least 20% women (20%W) on Boards shows a positive and statistically significant coefficient.

The appointment of women to the Boards is not random and financial performance may causally affect female representation on Boards, as either female directors would prefer to join Boards of better performing companies or better performing companies may appoint a greater number of women to their Boards [73]. Because financial performance may affect the incentive for women to join corporate Boards and the motivation of Boards to appoint female directors, an endogeneity problem may arise from simultaneity/reverse causality between gender diversity and financial performance [47,73]. To address the reverse causality, a procedure similar to that of Ararat and Yurtoglu [59] was employed, consisting of testing whether ROA and Tobin’s Q of year t predicts female representation on Boards in years t + 1 and t + 2. For this purpose, a logit model was applied with 2WOM and 20%W (for years t + 1 and t + 2) as dependent variables and with ROA, Tobin’s Q, and the remaining variables of the model as independent variables (as per Table 6). The estimation results are presented in

Table A4. Test for reverse causality.

Panel A—ROA as Predictor of Gender Diversity
	2WOMt+1	2WOMt+2	20%Wt+1	20%Wt+2
ROA	0.158 (0.533)	0.047 (0.548)	0.353 (0.541)	0.368 (0.583)
N	261	232	261	223
Log—likelihood	−122.60	−113.34	−133.61	−124.02
Panel B—Tobin’s Q as Predictor of Gender Diversity
	2WOMt+1	2WOMt+2	20%Wt+1	20%Wt+2
Tobin’s Q	0.248 (0.157)	0.217 (0.161)	0.230 (0.143)	0.073 (0.152)
N	225	200	225	200
Log—likelihood	−89.73	−83.66	−105.71	−99.41

Variables definitions as per Table 2. Each column presents estimation results for logit models with gender diversity as dependent variable and financial performance as independent variable. The regressions include intercept, year dummies, and the independent variables of the models of Table 6. The analysis performed aims to test whether ROA and Tobin’s Q of year t predicts the presence of at least two women (2WOM) or at least 20% women (20%WOM) on Boards for the following year (t + 1) and two years ahead (t + 2). Standard errors in parenthesis.

and show no statistically significant coefficients for ROA and Tobin’s Q, indicating no evidence for reverse causality between gender diversity and financial performance.

5. Conclusions

Based on a sample of 29 Portuguese listed companies from the period from 2010 to 2019, this paper studied the effects of Board gender diversity on financial performance. Portuguese listed companies are subject to gender quota law, which imposes a minimum proportion of women on Boards of 20% from the first elective general meeting after 1 January 2018 and 33.3% from the first elective general meeting after 1 January 2020. Our results support that the presence of women exerts a positive effect on financial performance, corroborating the economic premises for more gender diverse Boards, based on resource dependence theory and agency theory. However, this effect is not observed linearly with the proportion of women on Boards or for the presence of only one woman, rendering support for critical mass theory. In fact, a positive effect on ROA and ROE was only observed when the Board has at least two women or when the minimum threshold of 20% of female directors is reached. Concerning the Tobin’s Q, a positive effect of Board gender diversity was found only for the presence of at least 20% of women. Consistently with signaling theory, this result leads to the conclusion that by complying with the gender quota law, companies send a positive signal to the market about their future risk and cash flows. The gender diversity underlying the 20% threshold not only ensures better economic performance and higher future cash flows, but also reduces the perceived risk for investors, as these companies will not incur penalties for non-compliance with the law or reputational losses for being considered discriminatory or non-compliant. Meeting the 20% threshold signals adherence to social values and compliance with legal provisions, which in turn enhances the firm’s reputation.

This research makes three primary contributions to the literature. First, it adds to the scarce literature [14,15] on the effects of Board gender diversity on financial performance, in the Portuguese context. Second, it presents empirical evidence on the levels at which female representation affects the financial performance of Portuguese listed companies, in a period in which two important moments occurred: in 2012, the European Union presented a proposal for a Directive on Board gender balance; and, in 2017, a law came into force in Portugal, imposing gender quota for the Boards of state-owned and listed companies. Third, it provides an economic justification for the gender diversity regulation imposed on companies, legitimizing the thresholds imposed and encouraging companies to meet them as soon as possible, which makes the evidence found particularly important for managers, investors, and policymakers.

Despite its contributions, this paper has some limitations, the first of which is the reduced sample size due to the small number of companies on the Portuguese stock market. A second limitation regards the period analyzed, which ends in the year 2019, not allowing the effects of reaching the 33.33% threshold to be fully understood. Finally, it should be noted that other personal characteristics of women which may affect the relationship between financial performance and gender diversity, such as education and experience, have not been studied due to the lack of information on these aspects, which has only more recently started to be disclosed in governance reports. Certainly, future studies will contemplate periods that will make it possible to remedy these limitations. Concurrently, future research may consider not only the female presence on Boards but also in specific committees and other top management positions as CFO and CEO (e.g., [41,43]). Additionally, other effects of female presence should be studied, such as on earnings management (e.g., [74]) and sustainability performance and disclosures (e.g., [75,76,77]).