Women on boards and monitoring tasks: an empirical application of Kanter ’ s theory on and

Purpose – This paper examines the relationship between women on boards and board monitoring tasks depending on group categories identified in the Kanter ’ s theory. Design/methodology/approach – Using a sample of the largest listed companies in Spain, Italy and France during the period 2007 – 2017, this study tests the effect of women ’ s presence based on the following board categories: (1) skewed boards with a percentage of women that is less than 20%; (2) tilted boards with a percentage of women that ranges from 20% to 33%; (3) tilted boards with a percentage of women that is more than 33%; and (4)balanced boardswith an equal or quasi-equal gender distribution.The authors use the case of the gender board quota regulation in different European Union countries. Findings – The results suggest that tilted boards engage in stronger firm monitoring and that the effect of women on board monitoring tasks is positive and statistically significant when the percentage of female directors reaches the threshold of 33%. Practical implications – The outcomes of this study help policymakers identify the minimum threshold that quota regulations should mandate in order for boards to be effective. Originality/value – This paper moves forward the ongoing debate about the effect of women on corporate boards, shifting the focus from the ratio or presence of female directors to the size of the group they form within the board. To the best of authors ’ knowledge, this is the first study to test Kanter ’ s theory by investigating the relationship between women on boards and board monitoring.


Introduction
The effects of having more women in decision-making positions have been discussed for the last decade. There is an overall consensus that increasing the share of women in areas of power and influence, such as on corporate boards, is important for reasons that range from business utility to justice and equality (Seierstad et al., 2017). However, there is an ongoing debate about the circumstances that must exist so that the presence of more women on boards translates to more effective boards that improve performance of their tasks.
A major assumption regarding the presence of female directors on corporate boards is that on a male-dominated board, women may bring different resources, qualities and managerial practices. This increased heterogeneity helps avoid group-thinking problems related to the decision-making process. According to the seminal work of Janis (1972), when group thinking occurs, the group comes to a consensus without critical reasoning or consideration of the Women on boards and monitoring tasks possible alternatives and their consequences. At the board level, these dynamics may be particularly detrimental and may affect a number of decisions and firm outcomes. As a corporate governance mechanism, boards are aimed at monitoring the opportunistic behaviour of managers. According to the classical principal-agent framework (Jensen and Meckling, 1976;Fama and Jensen, 1983), managers can make decisions that maximise their own utility, thereby reducing shareholder wealth. In this context, a board of directors can be used as a tool to monitor managers and reduce agency costs. Specifically, the board scrutinises the behaviour of top executives and actively discusses and challenges managers' decisions (Hillman and Dalziel, 2003).
Most of the prior research on women on boards aims to prove the value of having female directors, looking foremost at the association between women on boards and firm financial performance (see the survey by Kirsch, 2017). However, another interesting research question is related to how gender diversity in the boardroom enhances board performance. In this paper, we analyse how gender diversity helps the board perform its monitoring tasks.
Prior studies on women on boards document the effect of female directors using different measures, such as the ratio of female directors, the number of women in the boardroom or a dummy that identifies the presence of women (Post and Byron, 2015;Gabaldon et al., 2016). Findings on the effects of gender diversity are not conclusive because most of the relationships are not linear and the above measures may not be appropriate (see the reviews by Terjesen et al., 2009;Kirsh, 2017). The emerging empirical literature (Konrad et al., 2008;Torchia et al., 2011;Rossi et al., 2017;Srivastava et al., 2018;Birindelli et al., 2019) proposes that female directors become effective for board performance when a critical mass is reached. This proposal is based on the theoretical work of Rosabeth Moss Kanter (1977) which argues that women in organisations change their behaviour according to their numerical representation. Specifically, Kanter identifies different group categories based on numerical proportion between the dominant sub-group and the minority. In each category, women, as a minority, face different group dynamics that may influence their contribution to group performance.
Despite its popularity, Kanter's theory has rarely been empirically tested. While studies on gender diversity often explicitly refer to Kanter's work, they rarely test the effect of the different group categories she identifies. We aim to fill this gap. We study the difference in board monitoring depending on the type of group identified by Kanter: (1) skewed groups dominated by men in which women comprise up to 20%; (2) tilted groups with a more moderate distribution of men and women (percentage of women is from 20% to 40%); and (3) balanced groups with an equal or quasi-equal gender distribution (percentage of women is from 40% to 50%). Our study shows that board monitoring tasks change across the types of groups identified by Kanter. The association between thresholds of female directors and board monitoring tasks is positive and statistically significant in tilted boards when the percentage of women is more than 33%. At this threshold, board members perceive female directors as equal colleagues, and male directors are less likely to dismiss comments made by a woman. Our results indicate that tilted boards engage in stronger firm monitoring as the female directors reach the threshold of 33%.
The paper contributes to the research literature in three ways. First, we test Kanter's theory in the context of corporate governance, a topic that has drawn limited attention from empirical studies (Konrad et al., 2008;Torchia et al., 2011;Joeck et al., 2013;Rossi et al., 2017;Srivastava et al., 2018). We show in which of the groups identified by Kanter women influence board monitoring tasks. For this type of analysis, country selection is crucial. In recent years, the European Commission has put pressure on Member States to introduce quota regulations that foster the presence of female directors in the boardroom. Although quota regulations have been introduced in several European countries, countries differ in identifying the minimum mandatory threshold of female directors that a board should meet. Spain was the first country in the European Union to introduce a gender quota regulation. Enacting this law has increased the number of women on corporate boards, but female representation on corporate boards is below 20% (Gabaldon and Gimenez, 2017). The quota law in Italy identifies 20% and 33% as the minimum thresholds of the underrepresented gender (Rigolini and Huse, 2017;De Masi, 2019). Specifically, the law mandated a gender quota of 20% for Italian-listed companies and state-owned enterprises in the first board term after the enforcement of the law. In the subsequent board term, boards are expected to have 33% of their seats held by women. The law is well-enforced (Desana, 2017) as there is a wide range of sanctions, such as a warning, a fine and the dismissal of all board members. French regulation is stricter, requiring the boards of listed companies to have 40% of their seats occupied by women (Bohren and Str€ om, 2010). After the enforcement of the quota regulation, France was one of the countries in the European Union with the highest percentage of women on corporate boards (Zenou et al., 2017). These three countries have common features in terms of the ownership concentration and governance of companies (Hall and Soskice, 2001) and the legal system (La Porta et al., 1999). Specifically, Weimer and Pape (1999) group countries using a taxonomy of corporate governance systems based on characteristics such as the main stakeholders who control managerial decisions; the importance of the stock market in the national economy; the role the market plays within corporate control; the corporate ownership structure; the executive compensation system; and the time perspective of economic relationships. They document that France, Italy and Spain belong to the same category. This gives us an accurate setting to run a natural experiment and test the effect of the groups identified by Kanter. Second, we contribute to the stream of research on gender diversity and corporate governance. We move beyond the studies linking female directors with firm performance (Adams and Ferreira, 2009;Rossi et al., 2017;Srivastava et al., 2018). Our study aims to understand the effect of having women on boards with respect to one of the main board tasks, that is, board monitoring. Identifying three different situations according to the numerical representation of female directors (skewed, tilted and balanced), the paper documents the influence of women on board monitoring depending on the size of the group they form.
Third, our findings provide empirical support for policymakers at the European Union (EU) and national levels. There are different approaches to increasing the presence of women on corporate boards, and in the case of quota regulations, the discussion about the mandatory thresholds of female directors is far from reaching a common consensus. Our results shed light on the threshold that enhances board monitoring.
The rest of the article is organised as follows. Section 2 discusses the related literature on Kanter's theory and women's contribution to board monitoring. Section 3 describes the sample, the variables and the methodology. Section 4 presents the estimation results and a discussion of the findings. Section 5 provides robustness checks and Section 6 gives conclusions and implications for policymakers and regulators.
2. Literature review and hypothesis development 2.1 Kanter's theoretical framework and women on boards According to Kanter's theory, the benefits of diverse boards may be hindered by the dynamics between the dominant sub-group and the minority. Building on the seminal work of Kanter (1977), our analysis identifies three different categories according to their composition: (1) skewed groups with a percentage of women up to 20%; (2) tilted groups, with a more moderate distribution of men and women (percentage of women is from 20% to 40%); and (3) balanced groups with an equal or quasi-equal numerical representation of men and women (percentage of women is from 40% to 50%).

Women on boards and monitoring tasks
With regard to interaction problems and their influence on group outcomes, the skewed category is the most problematic. The effect of women may be reduced because they are treated as tokens. The dominant group (men) tends to see women first as female, giving them a sex role stereotype. This makes it difficult for women directors to be heard and, importantly, to be perceived as equal colleagues. In the skewed group, women face three possible situations that affect their contribution and, consequently, the group performance: visibility, polarisation and assimilation (Kanter, 1977). Visibility takes place when women find themselves being constantly watched. In this case, they are afraid to make mistakes, and they experience pressure not to outperform men. Polarisation occurs when men exaggerate the similarities among men and emphasise the differences between men and women to isolate women and exclude them from the decision-making process. Assimilation relates to women's attitudes of accepting the gendered stereotypes defined by men. These stereotypes make it difficult for women to be included in the decision-making process, creating isolation, discomfort and a lack of confidence among minority members (Tajfel and Turner, 1986;Carli and Eagly, 2002). Nielsen and Huse (2010b) confirmed that gender stereotypes may limit the potential influence of women on boards working as a group. They demonstrated that the perception of women as unequal board members may limit their potential contribution to board decision making. In a skewed group, the contribution of women to group outcomes is thus hindered because of all of these possible situations. We used Spain as a proxy for boards in which women face most of the challenges of being members of a skewed group. In tilted groups, women are perceived more equally. Unlike in skewed groups, they experience inclusion and comfort. They can ally with other women and influence the culture of the group (Joeck et al., 2013). Even if female directors may still face tokenism, they contribute more to the discussion, bringing different perspectives that are more likely to be heard by the other board members. In Italy, where the quota law mandates a minimum threshold of 20% in the first board term and 33% in subsequent terms, boards behave like tilted groups in line with Kanter's theory.
In balanced groups in which the threshold of 40-50% of female directors on the board is reached, gender-based differences become less important than in the other subgroups, and the focus is on the skills and competencies of each individual independently of gender. Here, the balance between men and women allows for more likely effective board performance as the group may benefit completely from the advantages that come from diversity. We use France, where the board quota law mandates 40% of the board seats to be held by women, as the context in which to test the association between women on boards and board monitoring in balanced groups.
There are a limited number of empirical studies that verify Kanter's theory in the context of women on boards. One of the first studies to identify the behavioural consequences of being a token on the board is Konrad et al. (2008). The authors documented the different behaviour of women according to their number on the board. Specifically, they identified different dynamics in groups with one, two or three women. In groups with a solo female director, women experienced the following situations: (1) hypervisibility, that is, constantly being in the spotlight; (2) invisibility, that is, being ignored and not taken seriously; (3) stereotyping, that is, viewed as representing all women and not as being an individual; and (4) needing to work hard to be heard and impact the group. In groups with two female directors, the situation of women is better, but tokenism still exists. Women experience increased feelings of inclusion. Although women may still be stereotyped, their contribution is more pronounced than in the case of solo female directors. In groups with three women, gender is no longer a barrier, and women find themselves more self-confident and less constrained by what the men think. They also ally with each other to support different arguments and discuss alternatives. Torchia et al. (2011) tested the effects of one, two and three women on boards with respect to firm innovation using primary data from Norwegian boards. Specifically, they showed that a solo woman on boards is not sufficient to impact organisational innovation. They provided empirical evidence that gender diversity cannot matter if women are only tokens. Joecks et al. (2013) is the first study to test the different thresholds of female directors, focusing on their effect on firm performance. They found that tilted boards outperformed skewed boards. Specifically, they showed that female directors reaching the threshold of 30% positively affected firm performance, providing empirical evidence that the contribution of women changes according to the size of the group they form. As it is critical to move the debate that in the past has focused on the direct link between the representation of women on boards and firm performance (Kirsch, 2017;Terjesen et al., 2009), we examine the relationship between board monitoring tasks and women on boards depending on their percentage thresholds as an indication of a certain board category.

Women on boards and board monitoring
Monitoring tasks are seen as one of the key tasks for boards. Previous studies have identified different sets of monitoring tasks, such as management behaviour control, budget control, strategic control and CSR control (Baysinger and Hoskisson, 1990;Stiles and Taylor, 2001;Huse 2005;Huse et al., 2009). We focus on the first set of monitoring tasks which is the monitoring of CEO and top management team behaviour.
Women on boards may be particularly effective in monitoring management (Samara et al., 2019). There are several reasons for this which are rooted in the agency theory and resource dependence theory (Jensen and Meckling, 1976;Fama, 1980;Daily and Dalton, 2003;Pfeffer and Salancik, 1978). The agency theory describes the relationship between a principal (e.g. shareholders and other stakeholders) and an agent (e.g. managers). According to this theory, a principal cannot observe management behaviour (Jensen and Meckling, 1976) and, with these informative asymmetries, managers are likely to make decisions that improve their own utility but may be detrimental to the company (Fama and Jensen, 1983). A common assumption in agency theory is that independent directors would act as good monitors for shareholder interests since they are independent from their inside director counterparts. Women on boards are more likely to be appointed as independent directors (Carter et al., 2010) that are nominated from the outside of "old boys' networks" (Kirsch, 2017). Thus, they strengthen existing board mechanisms for monitoring manager behaviour. In the light of Kanter's theory, the strong monitoring effect requires a certain percentage of women so that the board operates like a tilted group. When there are not enough women, boards operate like skewed groups in which women count to less than 20% of the board composition. They are then treated as tokens. Their voices are not heard and their competences and experiences are not exploited. Nielsen and Huse (2010a) confirm that the potential contributions of women are likely to be disregarded when women are seen as tokens. Therefore, we posit the following Hypothesis: H1. On skewed boards which have a percentage of female directors up to 20%, women behave as tokens and their presence does not affect board monitoring tasks.
Building further on Kanter's theoretical framework, the increase in the percentage of women on boards affects the value added brought by women to the boardroom. Specifically, women on boards might particularly influence board monitoring tasks because they are more likely to be positively linked to the diffusion of the managerial control system, which leads to a professionalisation of the firm (Songini and Gnan, 2009). More professionalism can reduce the asymmetry between the principal and the agent, decreasing the possibilities that managers behave opportunistically. This is because a professional board is more likely to set up managerial control systems that can prevent underperforming managers from receiving Women on boards and monitoring tasks unmerited privileged treatment or from using company resources for private benefits (Samara and Arenas, 2017;Samara et al., 2019). Women are also more likely to attend board meetings than are men (Adams and Ferreira, 2009) and they carefully scrutinise information provided by management before attending meetings (Izraeli, 2000;Huse and Solberg, 2006). Higher attendance and higher scrutiny imply greater control over managers, which would improve the probabilities that accounting irregularities and opportunistic managerial behaviours are detected (Al-Mudhaki and Joshi, 2004;Pucheta-Martunez et al., 2016). Moreover, women are more likely to question business practices that are unethical (Franke et al., 1997), and they tend to apply stricter ethical standards (Pan and Parks, 2012). All of these results suggest that women could be particularly active in monitoring management behaviour and could positively affect the monitoring attitude of the board.
Resource dependence theory is used to argue that female directors bring to boards unique resources in terms of knowledge, skills, experience, competencies and specialisation (Kirsch, 2017). Empirical studies indicate that women and men on boards have different competencies and experiences (see survey Terjesen et al., 2009). Women are more likely to have professional backgrounds related to fields such as accounting or law that are particularly helpful for enhancing board monitoring. In this respect, boards with women are more capable of challenging management decisions (Judge and Zeithaml, 1992;Hillman et al., 2000), and they may be particularly effective in monitoring management behaviours. All these benefits are subject to the percentage of women on boards. An increase in percentage changes the women's status in the boardroom from tokens to more equally accepted board members. Based on the above discussion, we present the following hypotheses: H2a. On tilted boards which have a percentage of female directors from 20% to 33%, women are not seen as tokens and they facilitate more diligent board monitoring tasks.
H2b. On tilted boards which have a percentage of female directors of more than 33%, women are perceived more equally and they facilitate more diligent board monitoring tasks.
In balanced groups which reach the threshold of 40%-50% of women on the board, women are perceived as completely accepted board members. Each board member is seen as an individual, independently from his/her gender. In this group, the competences and knowledge of women are likely to be extensively used. The benefits that come from a diverse group are more likely to be further exploited and women contribute significantly to board monitoring tasks. For this reason, we posit the following hypothesis: H3. On balanced boards which have a percentage of female directors from 40% to 50% and women are perceived as equal to male board members, the presence of women improves board monitoring tasks.

Data and methodology
This investigation is based on a sample of the largest listed companies in Italy, France and Spain during the years 2007-2017. Specifically, we consider the classification of the national stock exchange that defines the largest listed companies in the country in 2015. We include EBEX 35 companies in Spain, FTSEMIB 40 companies in Italy and CAC 40 companies in France. These companies are known for being the most transparent within their own country, showing the highest level of disclosure required by European law.
The choice of the sample and the research period are in line with the aim of the study to examine the relationship between board monitoring tasks and women on boards depending on their percentage thresholds as an indication of certain board categories: skewed, tilted and MD balanced boards. Spain introduced the gender quota law in 2007. This law suggested that in listed companies, boards of directors have at least 40% of their board seats filled by women. The quota was not mandatory since there were no legal sanctions for non-compliance. The goal of 40% was never fulfilled, and the presence of women on boards stayed below 20% (Gabaldon and Gim enez, 2017). This result suggests that boards in Spain can be proxied as skewed groups and that female directors experience the behavioural challenges of tokenism. In this context, female directors may be considered tokens, and their contribution to board monitoring may not be relevant. Italy introduced a law in 2012 requiring a quota of female board directors in public companies. Specifically, every listed company had to reach the intermediate threshold of 20% of female directors on the board for the first post-law board term and 33% for subsequent board terms. In contrast to Spain, the law has been wellenforced as the sanctions for non-compliance are severe, and all listed companies in our sample have complied with this new board requirement. The thresholds of 20% and 33% of female directors identify the minority in tilted groups. We examined whether at those proposed thresholds of 20% and 33% of female directors, the benefits of a more diverse board exceed those of boards below the thresholds. The quota law in France proxies the case of balanced groups. The largest French companies must reach the threshold of 40% of female directors on their corporate boards (Zenou et al., 2017). In this case, the law is well-enforced since in cases of non-compliance, sanctions range from the non-payment of individual board member's remuneration to the annulment of director appointments.
To test our hypotheses, we used data from Thomson Reuters' Datastream ASSET4 ESG database. This database provides detailed information about corporate governance and financial statements. Specifically, the database collects information about certain corporate governance policies for guiding the behaviour of board members based on companies' annual reports on corporate governance, financial statements or any other publicly available documents related to companies' corporate governance.
To test the contribution of female directors for board monitoring, we estimated the following regression model for gender representation on the board of firm j: where α is the constant, β is the coefficient, ε is the residual term, and t 5 2005, 2006 ... 2015.
Our dependent variable is MONITORING, which is a proxy for board commitment towards the monitoring of management through the application of legal rules and best practices of corporate governance principles. It includes a set of items that assess whether a board has a policy that ensures effective monitoring; whether the policy is implemented; whether the board has internal information tools to develop monitoring of top management and the CEO; whether the board exercises monitoring tasks through the establishment of board committees; whether the board has financial experts on the auditing committee; the percentage of independent directors on board committees; and the number of board meetings and average board attendance.
All of this information is transformed by Datastream into a score that is calculated by equally weighting and z-scoring all underlying data points and comparing them against those of all companies in the Datastream ASSET4 ESG database. Z-score, or "standard score" (Kreyszig, 1979) is a relative measure that compares one company with a given benchmark. It reflects a firm's level of board monitoring task fulfilment relative to the average level of board monitoring task fulfilment measured for all other companies rated by Datastream ASSET4. It expresses the value in units of standard deviation of that value from the mean value of all companies (Wooldrige, 2009;Glantz et al., 2016). The z-scores are normalised, which entails that Datastream ASSET4 scales the z-scores in order to make them fit into the range of zero to one hundred. The resulting measure allows the creation of more distinction between values Women on boards and monitoring tasks that otherwise might be very close together (Kreyszig, 1979). These DATASTREAM scores, which are built using the secondary information data that is directly downloaded from official documents, helps us to easily make comparisons among countries and among companies, avoiding subjective assessments that may be an important limitation in the case of primary data (Thomson Reuters Datastream, 2019). Our explanatory variable WOMEN indicates a set of three dummy variables reflecting the percentage thresholds of women on boards identified by the country's gender quota law. In line with Kanter's theory (Kanter, 1977), we distinguish the following categories of boards: SKEWED BOARDS (assuming the value 1 if less than 20% of the board is comprised of women and 0 otherwise), TILTED BOARDS 20% (assuming the value 1 if at least 20% but less than 33% of the board is comprised of women and 0 otherwise), TILTED BOARDS 33% (assuming the value 1 if at least 33% of the board is comprised of women and 0 otherwise), and BALANCED BOARDS (assuming the value 1 if at least 40% of the board is comprised of women and 0 otherwise). The coefficient of the dummy variables measures the effect of having women on boards with respect to board monitoring tasks depending on the different board categories.
We include control variables related to board characteristics that affect the ability of the board members to complete their tasks (Dalton et al., 1998;Minichilli et al., 2009;Zona and Zattoni, 2007;Nielsen and Huse, 2010b). BOARD SIZE, calculated as the number of members on the board, is assumed to significantly influence board monitoring tasks (Core et al., 1997;Bebchuck and Fried, 2004). INDEPENDENT DIRECTORS is measured as the percentage of independent directors reported by the company. The higher the number of independent directors, the stronger the ability of the board to monitor managers. CEO DUALITY is a dummy variable that is equal to 1 if the CEO is also the chairman and zero otherwise. CEOchairman separation may be beneficial for monitoring tasks since in the case of CEO duality, there would be constraints on board members' capacity to ask critical questions and obtain correct answers (Daily and Dalton, 1997). The BUSY DIRECTORSHIP variable is measured as the average number of other corporate affiliations of board members, and it captures the time that a board may devote to board activities (Lipton and Lorsch, 1992;Jiraporn et al., 2009). As a proxy of firm size, we use the logarithmic transformation of total assets. All variables are defined in Table 1.
To investigate the effect of different thresholds of female directors on board monitoring tasks, we considered panel data, which is more accurate in controlling for unobservable heterogeneity and omitted variable bias (Campbell and Minguez-Vera, 2008;Srivastava et al., 2018). As the estimation method, we used fixed effects. The results are clustered for firms. Stata 11 was used to run the analysis.

Empirical analysis
Before starting the regression analysis, in Tables 2 and 3 we first examined the descriptive statistics of the variables used in this study. Following Kanter (1977), we differentiated the data according to country. Each country identifies a different group with a specific degree of female director participation on boards. The three countries are similar in terms of average BOARD SIZE (14 members in Spain and France; 15 in Italy), BUSY DIRECTORSHIP (the average number of other directorships that board members have is less than 2) and FIRM SIZE. The countries are different in terms of the presence of women in the boardroom. The average representation of female directors is 12.26% in Spain, 16.46% in Italy and 24.29% in France. In Spain, most boards (76%) are skewed. In Italy, tilted groups comprise 68% of the sample, whereas in France, the percentage of balanced boards is 63%. There are also differences in the monitoring score. The mean value of the MONITORING TASKS variable is 47.04 in Spain, 56.07 in Italy and 65.59 in France. Table 4 shows the contribution of female directors to monitoring tasks.

MD
The effect of women on boards on monitoring tasks is not significant in the case of skewed boards (coeff. 5 2.02, p-value 5 0.70). This finding is in line with Kanter's theoretical framework which highlights the behavioural challenges that women face when they are tokens on skewed boards. The result suggests that when women are less than 20% of board composition, their presence does not affect board monitoring tasks. Hypothesis 1 is confirmed.
Looking at tilted boards (column 2), when the percentage of women on boards is between 20% and 33%, women's effect on firm monitoring is not statistically significant. This finding documents that women on this type of board may continue to experience issues related to tokenism, and, hence, their contribution may be hindered. Thus, Hypothesis 2(a) is rejected. Looking at tilted boards (column 3), the regression results show that the threshold of 33% of women on boards matters for board monitoring. These results indicate that tilted boards engage in stronger monitoring tasks as long as the threshold of women on boards is met at 33%. Specifically, when the 33% threshold is reached, the score for the board monitoring tasks improves by 7.50. Hypothesis 2(b) is thus supported. Adding to Huse et al. (2009), who showed a positive impact from having women on boards with respect to monitoring managerial behaviour, we demonstrate that this effect was achieved when women held at least 33% of board seats. Our results provide evidence on the role of women in board Variable Description Monitoring tasks It is a Datastream category score that measures the board commitment towards the monitoring of management through the application of legal rules and best practices of corporate governance principles. It includes a set of the following indicator scores: (1) whether a board has a policy that ensures effective monitoring; (2) whether the policy is implemented; (3) whether the board has internal information tools to develop monitoring of top management and the CEO; (4) whether the board exercises monitoring tasks through the establishment of board committees; (5) whether the board has financial experts on the auditing committee ensuring auditing from the field of the internal control system (in the recommendation of Sarbanes-Oxley Act); (6) the percentage of independent directors on board committees; (7) the number of board meetings; and (8) the average board attendance Skewed board It is a dummy variable that assumes the value 1 if a board has less than 20% of female directors; 0 otherwise Tilted board 20% It is a dummy variable that assumes the value 1 if a board has at least 20% of female directors but less than 33% of female directors; 0 otherwise Tilted board 33% It is a dummy variable that assumes the value 1 if a board has at least 33% of female directors; 0 otherwise Balanced board It is a dummy variable that assumes the value 1 if a board has at least 40% of female directors; 0 otherwise Independent directors %

Percentage of independent board members as reported by the company CEO duality
It is a dummy equal to 1 if the CEO is also Chairman; 0 otherwise Board size The total number of board members at the end of the fiscal year Busy directors Average number of other corporate affiliations for the board member Firm size The logarithmic transformation of total assets as reported by the annual financial report by the company Note(s): The scores are calculated by equally weighting and z-scoring all underlying data points and comparing them against all companies in Datastream. The resulting percentage is therefore a relative measure of performance, z-scored and normalised to better distinguish values and position the score between 0 and 100%. A Z Score, or "standard score" expresses the value in units of standard deviation of that value from the mean value of all companies Table 1.

Variables description
Women on boards and monitoring tasks monitoring. When the 33% threshold is reached, the group dynamics change: women are not viewed as tokens. Their voices are heard, and they facilitate board monitoring tasks.
In column 4, we show the effect of women on boards when they comprise 40% of the board. Contrary to Hypothesis 3, which expects a difference between boards with 40% female directors and boards with a lower percentage, our results show no statistically significant difference. This finding documents that at 40% representation of female directors, board members do not change their commitment towards monitoring tasks. Thus, the tipping point is not in balanced boards in which women are perceived as equal to men. Table 5 presents the results controlling for busy directors. We include these results in a separate table since some observations are dropped because of missing data. The previous findings hold: the contribution of female directors is positive and statistically significant in titled boards at the threshold of 33%. BUSY DIRECTORSHIP has a negative impact on board  Table 3. Female representation on the board of directors: following Kanter (1977) MD monitoring in Italy and France. The higher the number of other affiliations held by board members, the lower is the commitment of boards to monitoring tasks.

Robustness check
In this section, we provide additional test results to ensure the robustness of our key finding that when women reach a threshold of 33% of board composition, there is a positive and significant association between women on boards and monitoring tasks. We employed the  (s): T-statistics are reported in brackets. *, ** and *** denotes significance at 10%, 5% and 1%, respectively. Robust standard errors clustered at the firm-division level. All tests are two-tailed Table 4. The contribution of female directors on monitoring tasks Women on boards and monitoring tasks variable board size. Previous research documents that large boards have inefficiencies in terms of board monitoring. The general argument is that smaller groups are more cohesive, more productive and can monitor the firm more effectively (Yermack, 1996). In contrast, a larger board is more likely to be associated with higher coordination costs, free riding behaviour and slow decision making that may hinder their capacity to monitor (Lipton and Lorsch, 1992;Coles et al., 2008;John et al., 2016). Based on that, we excluded very large boards from our sample. We also included an additional control for potential differences between boards that have a different structure. In Spain, all companies have one-tier boards (Gabaldon and Gimenez, 2017). France and Italy are countries that allow companies to choose between a one-tier structure and a two-tier structure which has a supervisory board and management board. For these countries, we included a dummy called TWO-TIER which is equal to 1 if the board has a two-tier structure, zero otherwise. The results in Table 6 are consistent with  Tables 4 and 5, and our findings are unchanged. The variable TILTED BOARD (>33% WOMEN) is still positive and statistically significant, showing that when the 33% threshold of women on boards is reached, the score improves for board monitoring tasks. The variable BALANCED BOARD (>40% WOMEN) is still not significant, proving that at the threshold of 40% representation of female directors, the board commitment towards monitoring of management does not change.
6. Discussion and conclusion In our research, we studied the relationship between gender diversity in the boardroom and board monitoring through the lens of Kanter's theoretical framework identifying different group categories (Kanter, 1977). This paper attempts to understand how numerical proportion between the dominant subgroup on boards-menand the minority subgroupwomenimpacts monitoring tasks. According to the literature, women are more likely to ask for additional information (Loden, 1985;McInerney-Lacomb et al., 2009), question business practices that are unethical (Franke et al., 1997) and consider different viewpoints (Kirsch, 2017;Eisenhardt et al., 1997). Boards benefit from such contributions when women are not tokens. We provide evidence that the association between women on boards and monitoring  (s): T-statistics are reported in brackets. *, ** and *** denotes significance at 10%, 5% and 1%, respectively. Robust standard errors clustered at the firm-division level. All tests are two-tailed Table 6. Robustness check MD tasks is positive and statistically significant for tilted boards which have a percentage of women of more than 33%. At this threshold, board members perceive female directors as equal colleagues, and their different knowledge and capabilities are more likely to be utilised.
The paper makes several contributions to both theory and practice. First, we tested the validity of Kanter's framework in the context of corporate governance. We provide evidence that different thresholds of women have different effects on board outcomes. This result moves the ongoing debate from the effect of women on boards measured by the number of female directors or the presence of female directors towards the effect of the outgroups of women on board tasks.
Second, this study has important implications for corporate boards and policymakers. In recent years, several governments in Europe have introduced quotas to increase gender diversity on boards. The requirements regarding the minimum mandatory thresholds of women that corporate boards should incorporate differ across countries. On the one hand, our research documents the importance of having a law that pushes companies to increase the number of women on boards. On the other hand, it sheds light on the minimum threshold that a board should reach to exploit the benefits of gender diversity.
Third, we provide managerial implications of our study. More diverse boards improve monitoring tasks only if gender diversity on boards reaches a threshold of 33%. Considering the benefits associated with a more diverse board, companies may formulate policies to create a more diverse environment at the upper echelon of the organisation. Capturing the attention of male leaders about the benefits of women on boards and motivating them to support gender diversity seems crucial. This may be particularly important to change the business culture in countries such as Italy, Spain and France, where institutions are traditionally less supportive of women than in Nordic countries.
We also acknowledge some limitations of the study and directions for future research. On the one hand, diverse boards may be more efficient in monitoring; on the other hand, they may experience more conflicts, have trouble communicating and become split into fractions (Kirsch 2017). This could be particularly problematic in situations that require fast decisions (Williams and O'Reilly, 1998;Di Tomaso et al., 2007;van Knippenberg and Schippers, 2007). Research demonstrating how women contribute to better decision-making processes is lacking. More qualitative studies are needed to reveal the extent to which decisions change when more women are nominated to the board. Additionally, our study focuses on one set of board tasks and on one dimension of boards. It would be possible that the effect of women on boards differs across the board tasks and the dimensions of board monitoring. This is because different board tasks require different attitudes and skills which could have different effects on board performance. The effect of women, depending on the size of the group they form, on other board tasks and other board dimensions deserves further investigation.
Moreover, we focused on three countries that share common features in terms of corporate governance systems and legal frameworks (La Porta et al., 1999;Aguilera et al., 2015). Future research should also design comparative analyses considering other countries with different institutional settings. Scandinavian countries, for example, have a long tradition in terms of supporting women to reach decision-making positions. It is likely that in these countries, women in the boardrooms have roles that have a different influence on board dynamics and on board commitment to achieving its tasks. An extension of the study should consider other countries with different institutional settings.
Lastly, differences could exist between family and non-family businesses. Family firms have distinctive features that influence the contribution of women to board functioning (Samara et al., 2019). Specifically, one of the characteristics that differentiate family firms from other forms of companies is their willingness to preserve and increase their socioemotional wealth, which the family desires for intergenerational succession, to preserve family harmony and unity and to portray a good family image and reputation Women on boards and monitoring tasks (G omez-Mejia et al., 2007). These motives that drive the decisions as well as family member behaviour, together with the family relations within the boardroom, create different board attitudes compared to boards of non-family companies. Considering the agency-related problems in family firms, such as entrenchment and asymmetric altruism among family employees (Chua et al., 2009;Kellermanns et al., 2012), future research may explore the board monitoring attitude in these businesses and whether women family members are more or less able to increase the board monitoring capacity.