NATIONAL CULTURAL DIFFERENCES AND THE RELATION BETWEEN FEMALE DIRECTORS AND FIRM PERFORMANCE

NATIONAL CULTURAL DIFFERENCES AND THE

RELATION BETWEEN FEMALE DIRECTORS

AND FIRM PERFORMANCE

Boris van Oostveen University of Groningen Faculty of Economics and Business

Supervisors Prof. Dr. C.L.M. Hermes and Dr. L. Dam December 2012

ABSTRACT

Using a unique dataset, this study addresses endogeneity issues and national cultural differences in examining the relation between the extent of female board directors and firm performance within Europe. The first part of the research shows that the extent of female directors positively affects long term performance and that endogeneity must be circumvented as it biases results. The second part of the research shows that national cultural differences explain inter-country variation in the extent of female directors. Within cultures characterized by Hofstede’s power distance, the extent of female directors negatively affects firm performance. Within cultures characterized by individualism, masculinity and uncertainty avoidance, no relation exists.

I. INTRODUCTION

Theory articulates that the extent of female board directors positively affects firm performance. The central premise is that as boards consist primarily out of men, increasing the extent of female directors improves board heterogeneity and ultimately firm performance. However, empirical studies report mixed results. By using a unique dataset and an instrumental variables approach, this study circumvents endogeneity issues and addresses national cultural differences in estimating the relation between the extent of female directors and firm performance.

The fierce debates on governance reform efforts draw attention worldwide and gender diversity within the boardroom is the topic on which the normative discussion focuses. Norway is at the forefront of regulating gender diversity, as it was the first country to introduce a representation quota for female directors on corporate boards. In 2004, the Norwegian government passed a law mandating state-owned and inter-municipal firms to hold at minimum 40% of both genders at their boards (Storvik and Teigen, 2010). Following this example, Spain introduced a comply-or-explain quota in 2007, Iceland followed in 2011, and at the moment Germany, Italy, the Netherlands and the United Kingdom (UK) are contemplating the same (Storvik and Teigen, 2010; Werdigier, 2011). In 2010, PLCs in Finland need to have at least one female director on the board and in 2011 France passed a law stipulating a similar gender parity around 2015 to 2017 (Reding, 2011).

These reform efforts result from the low extent of female board directors worldwide. Catalyst (2007) reports that female directors account for 14.8% of the Fortune 500 board seats in 2007. At the same time in Australian, Canadian and Japanese boards, female directors account for 8.7%, 10.6% and 0.4% of all seats respectively (Adams and Ferreira, 2009). Table I shows that European boards consist of 6.6% female directors in 2004 and 9.3% in 2011. Although the average percentage of female directors has increased over time, for some individual countries the opposite is true.

Singh, Vinnicombe and Johnson (2001), Carter, Simkins and Simpson (2003), Erhardt, Werbel and Shrader (2003), Smith, Smith and Verner (2006), Catalyst (2007), McKinsey (2007), and Campbell and Minguez-Vera (2008) conclude on a positive relation between the extent of female directors and firm performance. However, more recent studies of Farrell and Hersch (2005), Adams and Ferreira (2009) and Bøhren and Strøm (2010) conclude on a non-existent or negative relation.

Given the continued low representation of female directors within the boardroom and all mixed results provided by literature, the focal point of this study is to examine how the extent of female directors relates to firm performance. This study consists of two parts. The first part of the research circumvents endogeneity issues by incorporating firm-fixed and time-fixed effects, and by using an instrumental variables approach. The extent of female directors is identified with two instruments, the number of board directors that have a seat in other boards with female directors and the percentage of female labor participation. The second part of the research addresses

Table I

Percentage of female directors for 19 European countries in the period 2004 to 2011

national cultural differences to explain the relation between the extent of female directors and firm performance. As the shaping of corporate governance is embedded in cultural norms and values, that are governed by a complex set of laws, rules and regulations, codes of conduct and policies, national cultural differences might explain how board functioning varies between countries. Literature shows that (professional) behavior differs between countries as a result of national cultural differences (Parsons and Shils, 1951; Kluckhohn, 1962; Hofstede, 2001; Adler, 2002; House et al., 2004). As values are considered the most fundamental aspect of culture and shape peoples’ behavior, cultural values shape the behavior of all individuals in society. Hence, the behavior of and interaction between male and female directors depends on culture as well and has implications for board functioning. Chang and Cheung (2008) find evidence that culture plays an important role in establishing productive negotiations among individuals. As it is directors’ behavior that determines board functioning and differences in behavior determine board heterogeneity, societal and cultural systems must also determine how the extent of female directors improves board functioning and heterogeneity, and consequently firm performance. Examining how Hofstede’s cultural dimensions affect the extent of female directors reveals whether culture is a factor that explains country differences. Limiting the sample to firms headquartered in countries whose power distance, individualism, masculinity and uncertainty avoidance scores are in the upper quartile, provides the opportunity to examine how the extent of female directors relates to firm performance depending on national cultural differences.

differences. Being the first to congregate these topics, this study paves the way for more research on the extent of female directors and national cultural differences.

The first part of the research shows that the extent of female directors positively affects long term firm performance and that endogeneity issues must be circumvented for these bias results. The second part of the research shows that national cultural differences determine the relation between the extent of female directors and firm performance. In cultures characterized by power distance, a negative relation is found. In cultures characterized by individualism, masculinity and uncertainty avoidance no relation between the extent of female directors and firm performance results.

The next section reviews two main stream theories that provide rationale for a positive relationship between the extent of female directors and firm performance. Key is how the extent of female directors affects board heterogeneity and functioning, and consequently firm performance. The second section also discusses national cultural differences and how these affect the relation between the extent of female directors and firm performance. The third section describes the methodology and the instrumental variables approach in particular. The instruments acquire special attention as these form the back bones on which the model builds. The fourth section elaborates on the unique European dataset and the fifth section on the results. Finally, the sixth section provides a conclusion and discusses limitations on this work and suggestions for further research.

II. LITERATURE REVIEW

resources to these firms. This section first elaborates on how, from both theories, a positive relation between the extent of female directors and firm performance is expected. Subsequently, this relation will be discussed in the context of national cultural differences.

Agency theory

The monitoring and disciplining function of the board refers to boards being an essential tool for shareholders to monitor the performance of managers and discipline them when necessary (Jensen and Meckling, 1976; Fama, 1980; Fama and Jensen, 1983; Eisenhardt, 1989). Managers who do not perform to their potential or disappoint shareholders by not achieving expected returns, risk being sanctioned and replaced by the board. If we assume shareholders to value maximization of firm performance, managers who act in the interest of the shareholders are expected to do the same. Ceteris paribus, the better the board functions and aligns the interests of managers with those of shareholders, the better a firm is expected to perform.

interpretation of the risky situation. Explanation for the difference in social preferences originates from psychology. It is suggested that females are more sensitive to social signals in determining appropriate behavior. Gender differences in competitiveness result from (the interaction between) nature and nurture. On the one hand, it is in females’ nature to rationalize avoiding a negotiation in certain situations. On the other hand, the fact that some gender differences exhibit only later in life exemplifies an environmental cause.

Based on board heterogeneity caused by differences between male and female directors, boards might behave differently depending on the extent of female directors. Adams and Ferreira (2009) provide evidence by showing that increasing the extent of female directors affects board input and output measures, which suggests that boards behave become tougher monitors. They find that female directors are less likely to experience attendance problems and that increasing the extent of female directors results in less attendance problems for male directors. Thus, boards are expected to do a better job carrying out their duty and gathering information when the extent of female directors increases. Additionally, female directors are more likely to join the audit, nominating and corporate governance committees than male directors. So if directors affect board governance more easily via committees, female directors have the potential to affect board governance more easily than their male counterpart. Adams and Ferreira (2009) also find that the extent of female directors positively affects the sensitivity of turnover to stock performance. Additionally, the extent of female directors is found to positively affect equity-based pay for directors. These findings suggests that increasing the extent of female directors causes boards to discipline managers and align the interests of managers with those of shareholders (via equity-based pay) more easily. Hence, increasing board heterogeneity is expected to positively affect board functioning.

Resource dependence theory

directors provide three primary benefits to firm performance: 1) information in the form of advice and counsel, 2) legitimacy, and 3) access to channels for communicating information and obtaining commitments, support or preferential access to resources from important elements in their environment.

Board directors are able to provide advice and counsel to managers in support of strategic decision making. According to Hillman, Shropshire and Canella (2007), boards set the parameters for strategic decision making, boards are often involved in strategy initiation, and participate in all phases of the strategic decision making process. This increased effort by directors to provide managers with better strategic decision making is expected to benefit firm performance. Kor and Misanyi (2008) support this as they find a negative relation between top management’s and board’s collective levels of experience, suggesting that board directors supplement executive directors with advice and counseling regarding strategic decision making. Increasing the extent of female directors improves gender diversity and board heterogeneity. Following Hillman, Shropshire and Canella (2007), heterogeneous groups show an increased interest in information search and a greater range of perspectives, they generate more alternative solutions to problems and have more different kinds of perceptions about the environment. This also emphasizes the creative nature of more diverse groups, suggesting increased innovative group behavior and facilitating the advice and counsel function. Additionally, since female directors are less likely to be insiders, increasing the extent of female directors may bring a broader range of perspectives and non-traditional approaches to problems, which benefits the advice and counsel function (Carter, D’Souza, Simkins and Simpson, 2010).

2004). Additionally, female directors also serve as a role model within the firm. Following Singh, Vinnicombe and Johnson (2001) and Hillman, Shropshire and Canella (2007), they symbolize career possibilities, inspire current and potential employees, and thereby contribute to the work ethos of these employees. Retaining female employees by providing career growth and opportunities can also decrease costs resulting from turnover.

Hofstede’s cultural dimensions

Power distance has implications for the relation between the extent of female directors, board functioning and firm performance. In societies with high power distance, power is unequally distributed among members of society. The less powerful members accept and expect power

Table II

Description of Hofstede's cultural dimensions

This table describes Hofstede’s cultural dimensions. It emphasizes the high and low characteristics that typify each cultural dimension individually. This format is adapted from Culture GPS: Professional Edition - 5D Model (ITIM, 2011), where originally long term orientation was added as well. For the purpose of this study it is excluded from the table.

High characteristics Low characteristics

Power distance

The extent to which the less powerful members of society expect and accept power being unequally distributed

- High dependence - Low dependence needs

- Inequality accepted - Inequality minimized - Hierarchy needed - Hierarchy for convenience - Superiors often inaccessible - Superiors accessible - Power holders have privileges - All should have equal rights - Change by revolution - Change by evolution

Individualism

The extent to which people look after themselves and their immediate family only, versus belonging to groups who look after them

- “I” consciousness - “We” consciousness

- Private opinions - Relationships have priority over tasks - Fulfill obligations to self - Fulfill obligations to family, group,

society

- Penalty imp - Penalty implies loss of “face” and shame

Masculinity

The extent to which

achievement and success are dominant values in society, versus caring for others and quality of life

- Performance ambition, a need to excel - Quality of life, serving others - Tendency to polarize - Striving for consensus

- Live in order to work - Work in order to live - Big and fast are beautiful - Small and slow are beautiful - Admiration for the successful achiever - Sympathy for the unfortunate

- Decisiveness - Intuition

Uncertainty avoidance

The extent to which people feel threatened by uncertainty and ambiguity and to try to avoid such situations

- Anxiety, greater stress - Relaxed, less stress

- Inner urge to work hard - Hard work is not a virtue per se - Showing of emotions accepted - Emotions not shown

- Conflict is threatening - Conflict and competition seen as fair play - Need for agreement - Acceptance of dissent

- Need to avoid failure - Flexibility

- Need for laws and rules - Less need for rules

being distributed unequally and are afraid of disagreeing with the powerful members, the established order (Hofstede 1980; 2001). In societies with low power distance less powerful members are more willing to disagree with the established order and their opinions are valued highly. All members believe everyone should be treated the same way. Chan & Cheung (2008) provide evidence that a big inequality gap between powerful and less powerful members of society impedes negotiations in good corporate governance practices. With respect to the unequal distribution of board directors, the powerful established order consists primarily out of men and presumably experiences little opposition when it comes to sharing power and board seats with female directors. As female directors are considered to be the less powerful minority within the boardroom, it is expected that power distance negatively affects the extent of female board directors. Additionally, board heterogeneity and functioning is not expected to derive benefits from an increased extent of female directors. The less powerful female directors, are expected to be restricted by male directors in sharing (opposing) opinions, thoughts, advice and counseling. Hence, within societies characterized by power distance culture, no relation between the extent of female directors and firm performance is expected.

Masculinity has implications for the relation between the extent of female directors and firm performance. Masculinity purports the extent to which competitiveness, assertiveness, achievement and success are dominant values in society. A society that emphasizes caring for others is considered more feminine (Hofstede 1980; 2001). In a feminine society, a more supportive and social orientation prevails together with a strong concern for the preservation of existing relationships. As in a masculine society, managerial decisiveness, a performance orientation and an emphasis on competitive behavior is highly valued, a negative relation between masculinity and the extent of female directors is expected. Additionally, no relation between the extent of female directors and firm performance is expected in a society with high masculinity. As managerial decisiveness, a performance orientation and an emphasis on competitive behavior is valued, presumably a more supportive, social and quality of life orientation is suppressed. This makes it likely that female directors add little or no value to board functioning or heterogeneity, as feminine behavior is constrained and repelled.

consisting out of men are expected to be already relative risk averse and adding female directors is expected not to improve board functioning and heterogeneity as it does in societies with low uncertainty avoidance.

III. Methodology

PERFORMANCE = + FEMALES + CONTROLS + + + ϵ (1)

where, is an index for the firm, is a time index, PERFORMANCE is a proxy for firm performance, is a constant, and are coefficient vectors for FEMALES and CONTROLS, the extent of female board directors and the control variables respectively, is a firm-fixed effect and is a time-fixed effect.

The first part of the research has the following estimation strategy. Firstly, model (1) is estimated without a firm-fixed and time-fixed effect and servers as a benchmark for the other models. Secondly, model (1) is estimated with a firm-fixed and time-fixed effect. It circumvents problems resulting from dependence in the residuals, caused by a firm-fixed effect or conditions that affect the performance of all firms at a given point in time, a time-fixed effect (e.g. financial or economic crisis). 1 These factors are referred to as omitted variables that affect dependent and independent variables. This study will address this omitted variables issue, as they cause estimates to become inconsistent and biased. Hence, the estimates are adjusted for firm-fixed and time-fixed effects. Additionally, the model is estimated with robust standard errors to account for group correlation within firms and heteroskedasticity. Thirdly, the instrumental variables approach is used to address endogeneity issues and the results are compared with the previous models. For example, reverse causality may be an issue because of potential sorting of male and female board directors to firms based on performance (Adams and Ferreira, 2009). Following van Ees et al. (2003), it can be argued that in difficult times poor financial performance could trigger

1 _{We test for correlation between unobservable heterogeneity and the explanatory variables using the Hausman test.}

shareholders to intervene in the decisions of top management and impose a more diverse board with more independent female directors. Shareholders reason that a more homogenous and male dominated board is less critical than a more heterogeneous gender diverse board, and that this criticism improves firm financial performance. Hence, shareholders try to improve performance by increasing the number of female directors to make the board more diverse. A related example of reverse causality is provided by Demsetz and Lehn (1985) who argue that if a governance structure affects firm performance and an optimal structure exists which relies on an endogenous choice, value maximizing firms automatically choose this optimal structure. Consequently, when controlling for other variables, no variation in the values of the governance and performance variables would be observed. This makes it impossible to identify a relationship between these variables. As both examples show the need to account for endogeneity, this study adheres to this by using other variables indicating governance issues to instrument the estimation of the relationship between firm financial performance and board characteristics. Hence, this study relies on an instrumental variables approach. Model (2), referred to as the first stage model, estimates the fitted values of the extent of female board directors by using two instruments and various control variables:

FEMALES = + INSTRUMENTS + CONTROLS + + + (2)

where, is an index for the firm, is a time index, is a constant, and are coefficient vectors for INSTRUMENTS and CONTROLS, two instrumental variables and the control variables respectively, is a firm-fixed effect and is a time-fixed effect. In the first stage the measures of FEMALES, INSTRUMENTS and CONTROLS are used to estimate the fitted values of FEMALES. Consequently, model (1), referred to as the second stage model, will be used to regress PERFORMANCE on the fitted values of FEMALES and CONTROLS to obtain unbiased and consistent estimates.

The second part of the research uses a similar instrumental variables approach as used in the first part, to examine whether cultural differences explain inter-country variation with respect to the extent of female directors and its effect on firm performance. The second part of the research starts with estimating the first stage, in which cultural dimensions are added to the model:

The second stage will be estimated subsequently, using the fitted values of model (3):

PERFORMANCE = + CULTURE + FEMALES + CONTROLS + + (4)

The observant reader will notice that model (3) and (2) and model (4) and (1) are similar, except for the fact that culture is included and a firm-fixed effect is excluded. The exclusion results from the slowly changing, time-invariant nature of Hofstede’s cultural dimensions. Including a firm-fixed effect would exclude the cultural dimensions from the model, as a firm-fixed effect captures such time-invariant effects on the dependent and independent variable. Subsequently, the second part of the research continues with estimating the relation between the extent of female directors and firm performance within societies characterized by high power distance, high individualism, high masculinity and high uncertainty avoidance. Firstly, four subsamples are created from the unique European sample that consist of firms headquartered in countries that belong to the upper quartile scores with respect to the four cultural dimensions. Secondly, instrumental variables estimations are performed as in model (1) and (2) using these subsamples. Using this method, the relation between the extent of female directors and firm performance, within societies characterized by Hofstede’s four cultural dimensions is uncovered.

IV. Data

Firm performance

The choice of performance measure as dependent variable is important since performance can be measured based on accounting or market data. This study accounts for both by using return on assets and price to book ratio. Return on assets equals net income divided by total assets and is considered to measure short term accounting performance. Price to book ratio equals market

Table III

Definition of board, firm and country characteristics for 19 European countries in the period 2004 to 2011 Definitions Source Board characteristics

Number of females Total number of female directors on the board at the end of the year Capital IQ

Percentage of females Total number of female directors on the board divided by the total number of board directors at the end of the year

Capital IQ

Female indicator Dichotomous variable that equals 1 when the board has one or more female directors at the end of the year and 0 otherwise

Capital IQ Board size Total number of directors on the board at the end of the year Capital IQ

Number of directors connected

Total number of directors on the board at the end of the year that have a seat in another firm's board with female directors

Capital IQ

Firm characteristics

Return on assets (ln) Natural logarithm of the annual net income divided by the book value of total assets at the end of the year

Orbis

Price to book (ln) Natural logarithm of the stock price divided by the book value of equity at the end of the year

Orbis

Firm age (ln) Natural logarithm of the number of years since incorporation Orbis Total assets (ln) Natural logarithm of the book value of total assets at the end of the year Orbis Turnover (ln) Natural logarithm of the total operating revenues at the end of the year Orbis Number of employees (ln) Natural logarithm of the total number of employees at the end of the year Orbis

Country characteristics

Female labor participation Percentage of females active on the labor market World

Bank

Power distance Dichotomous variable that equals 1 when a firm is headquartered in a country that belongs to the 25% highest scores on the power distance cultural dimension as provided by Hofstede (1980) and 0 otherwise

Taras, Steel and Kirkman

Individualism Dichotomous variable that equals 1 when a firm is headquartered in a country that belongs to the 25% highest scores on the individualism cultural dimension as provided by Hofstede (1980) and 0 otherwise

Taras, Steel and Kirkman

Masculinity Dichotomous variable that equals 1 when a firm is headquartered in a country that belongs to the 25% highest scores on the masculinity cultural dimension as provided by Hofstede (1980) and 0 otherwise

Taras, Steel and Kirkman

Uncertainty avoidance Dichotomous variable that equals 1 when a firm is headquartered in a country that belongs to the 25% highest scores on the uncertainty avoidance cultural dimension as provided by Hofstede (1980) and 0 otherwise

Taras, Steel and Kirkman

price divided by book value per share. It is considered to indicate long term performance, as market expectations are included in the market price per share. The natural logarithm of both measures is used in our analyses since it improves the normality conditions. As a consequence of using the natural logarithm, values being equal to or below zero, drop from the data set resulting in less observations.2

The extent of female directors

This study uses three measures for the extent of female board directors, the number of female directors, the percentage of female directors and a dichotomous variable that equals 1 when a firm´s board has one or more female directors at the end of the year and 0 otherwise. As boards consist primarily out of men, adding a female increases gender heterogeneity. Since these measures are extensively used in other studies, using them all together maximizes robustness. Data regarding the extent of female directors is retrieved from Capital IQ as follows. Firstly, for the 4000 largest firms, profiles of the boards are individually downloaded. The International Securities Identification Numbers, as retrieved from Orbis, are used to ensure that the firm profiles are located adequately within Capital IQ database. Subsequently, the number of male and female directors is retrieved from these individual profiles.3 Although not directly reported, gender is relatively easy to determine from information that is provided in these profiles. References, such as he or she, are used as (indirect) evidence to determine a director’s gender. As the appointment tenure of all directors is also provided in these profiles making is relatively easy to determine in which years directors serve(d) on the board.

The instruments

It is essential that the instruments are correlated with the extent of female board directors, but in essence uncorrelated with firm performance, except through the variables that are controlled for. The first instrument measures the number of board directors that have a seat in other boards with female directors. According to Medland (2004) the foremost reason that female directors do not enter a board, is their lack of connections. Medland (2004) argues that the primary obstruction to

2 _{To add robustness the analyses are also performed by using the square root of both measures, while preserving the}

positive or negative signs. The results are available at request by the author.

female directorship is that the informal social network, the pool from which board directors drawn, consists primarily out of men. Ceteris paribus, the more females are connected with this social network, the more they are observed by directors and the higher the probability is that they will be selected to join the boardroom. Sheridan and Milgate (2005) follow a similar of reasoning and argue that women’s exclusion from informal social networks, and the exclusionary nature of “old boys’ network” in particular, is considered to be one of the major limitations for women’s access to boards. With their survey they find that both male and female directors explicate that most often their name is brought to the attention of the board of directors by a recommendation of an inside director. Following this line of reasoning, it can be argued that when a female director is more connected to inside board directors, she is more likely to become part of the (so called) informal social network, which in turn increases the probability that she becomes a board director. It is therefore hypothesized that the number of board directors that have a seat in other boards with female directors is positively associated with the extent of female directors on the board.

example, in the US in 1930 only 1.5% of police officers, 1.5% of architects and 2.4% of lawyers were female. In 2007 17.8% of police officers, 25.9% of architects, and 33.7% of lawyers were female. With respect to students, changes are even larger (from 7.2% to 47% in medical school and 2.8% to 49% in Law school). Sheridan and Milgate (2005) also elaborate on the unequal division of labor in the households, limiting females’ access to the boardroom. They argue that traditionally, board positions are considered to be the domain of men and only recently women occupy senior and higher level positions within firms. Women these days do not aspire household jobs, but instead are very active on the labor market. These women are very talented and possess many skills and much work life experience. Therefore it is expected that female labor participation is positively associated with the extent of female board directors.

Hofstede’s cultural dimensions

Hofstede’s cultural dimensions have been used in thousands of studies including a large number published in recent years. However, according to Taras, Steel and Kirkman (2012) cultural dimensions slowly change over time. Taras, Steel and Kirkman (2012) meta-analyze the relationship between Hofstede’s four cultural dimensions and several organizational outcomes. They compare the scores on cultural dimensions as provided by Hofstede (1980; 2001) with those from more recent literature. They find that there are only small differences and that scores only change slowly over time. This study therefore relies on the Hofstede (1980) scores, as standardized by Taras, Steel and Kirkman (2012). Countries are granted scores on power distance, individualism, masculinity and uncertainty avoidance. Table IV provides an overview of the scores used in this study.

TABLE IV

Scores on Hofstede’s cultural dimensions for 19 European countries in the period 2004 to 2011 Based on the work of Taras, Steel and Kirkman (2012), this table shows the scores used in this study. These reflect standardized scores on Hofstede’s cultural dimensions and a meta-analysis.

Power distance score Hofstede Power distance score Meta-analysis Individualism score Hofstede Individualism score Meta-analysis Masculinity score Hofstede Masculinity score Meta-analysis Uncertainty avoidance score Hofstede Uncertainty avoidance score Meta-analysis Austria -1.99 -1.29 0.27 -0.07 1.55 1.15 0.37 -0.03 Belgium 0.41 0.37 1.09 0.59 0.23 0.19 1.30 0.88 Cyprus 0.33 -0.02 -0.51 -0.56 0.07 0.30 1.48 0.85 Denmark -1.68 -1.17 1.05 0.48 -1.78 -0.92 -1.47 -1.31 Finland -1.01 -0.90 0.60 0.23 -1.25 -0.58 -0.06 -0.03 France 0.55 0.41 0.92 0.39 -0.35 0.06 0.99 1.05 Germany -0.92 -0.49 0.76 0.30 0.86 0.64 0.17 0.43 Greece 0.19 -0.12 -0.55 -0.72 0.39 0.23 2.01 1.29 Ireland -1.23 -0.70 0.88 0.42 0.97 0.84 -1.00 -0.56 Italy -0.25 -0.06 1.13 0.49 1.07 0.70 0.56 0.62 Luxembourg 0.26 0.26 0.98 0.48 0.12 0.21 1.02 0.87 Netherlands -0.79 -0.11 1.29 0.89 -1.89 -0.91 -0.30 -0.27 Norway -1.10 -0.94 0.84 0.57 -2.21 -1.14 -0.42 -1.37 Portugal 0.33 -0.14 -0.88 -0.83 -0.99 -0.70 1.69 0.32 Spain 0.06 0.16 0.10 0.05 -0.41 -0.13 0.99 1.17 Sweden -1.10 -0.76 0.92 0.69 -2.36 -0.95 -1.24 -0.94 Switzerland -0.96 -0.57 0.80 0.40 1.07 0.73 -0.10 0.14 Turkey 0.46 0.09 -0.47 -0.39 -0.25 0.37 0.95 0.41 UK -0.92 0.03 1.66 0.93 0.86 0.83 -1.00 -0.61 Mean -0.49 -0.31 0.57 0.23 -0.23 0.05 0.31 0.15 Upper Quartile 0.29 0.06 1.02 0.53 0.86 0.67 1.01 0.86 Control variables

with more resources and experience in the market in general have a better reputation, which consequently may have an effect on firm performance. Hence, a positive relation is expected between firm age and firm performance. Firm size is measured by three indicators, total assets, turnover (operating revenue) and the number of employees, Similar to firm age, firm size is considered to reflect some degree of resources and market experience (Miller and Triana, 2009). Hence, a positive relation with firm performance is expected (Carter, Simkins and Simpson, 2003; Erhardt, Werbel and Shrader, 2003; Campbell and Minguez-Vera, 2008; Carter, D’Souza, Simkins and Simpson, 2010). In this, the underlying assumption is based upon firms making effective and efficient use of these assets, operating revenues and employees. Hence, when systematically (one of) these different aspects of firm size are not managed efficiently and effectively, estimations may yield results that differ from expectations. For all firm characteristics, the natural logarithm is used to improve upon the normality constraints that should be met for analyses.

Descriptive statistics

Table V shows the descriptive statistics of all variables used in this study.4 It shows that all board characteristics have 17,680 observations. On average firms have 0.532 female board directors, which is remarkably low. The same holds for the average percentage of female directors and the average percentage of firms that have one or more female directors, 0.077 and 0.374 respectively. The average board size equals 6.623. The average number of directors that have a seat in one or more boards at other firms in this sample equals 1.401.

Presented in the Appendix, Table A.I shows that board characteristics are changing remarkably over time. Firstly, in Europe the extent of female directors increases. The average number of female directors shows a relatively large increase over time (85.5%), from 0.374 in 2004 to 0.694 in 2011. The average percentage of female directors also increases over time (46.0%), from 0.063 in 2004 to 0.092 in 2011. The same holds for firms that have one or more female board directors (55.0%), it increases from an average of 0.293 in 2004 to 0.455 in 2011. In other words, in 2011 less than half of all European firms have one or more female board directors. Secondly, in Europe the other board characteristics also show an increase. The average board size increases from 5.905 in 2004 to 7.112 in 2011, which is relatively small (20.4%). The number of directors that

have a seat in boards of other firms, show a relatively large increase (65.3%), from 1.012 in 2004 to 1.673 in 2011. Thirdly, the number of observations increases over time, reflecting the time consuming nature of gathering data regarding board characteristics. In contrast to the Investor Responsibility Research Center in United States, in Europe no overarching organization exists that specifically gathers such board data and provides an off the shelf dataset.

Table V shows that the average female labor participation equals 44% Table A.I shows that female labor participation increases (2%), from 43.592 in 2004 to 44.460 in 2010. This is a very low increase compared to the increase in the extent of female directors. In 2011 no data regarding labor participation is available.

Table VI shows pair wise correlation coefficients between all variables. Ignoring the three measures for firm size, there is one sign of multicollinearity. The correlation between power distance and uncertainty avoidance equals 0.75. According to Kennedy (2008) coefficients above

Table V

Descriptive statistics for board, firm and country characteristics for 19 European countries in the period 2004 to 2011

Mean Stdev Min Max Obs

Board-level characteristics

Number of females 0.532 0.828 0.000 9.000 17,680 Percentage of females 0.077 0.120 0.000 0.800 17,680 Female indicator 0.374 0.484 0.000 1.000 17,680

Board size 6.623 2.777 3.000 21.000 17,680

Number of directors connected 1.401 1.775 0.000 13.000 17,680

Firm-level characteristics Return on assets (ln) 1.654 1.125 -4.605 4.583 22,754 Price to book (ln) 0.348 0.932 -6.908 14.648 22,598 Firm age (ln) 3.295 1.116 0.000 6.290 29,538 Total assets (ln) 12.857 2.253 -2.259 21.674 28,958 Turnover (ln) 12.376 1.998 -1.789 19.711 28,857 Number of employees (ln) 6.967 2.081 0.000 13.369 25,292 Country-level characteristics

Female labor participation 44.000 4.905 25.412 48.477 27,916

Power distance 0.246 0.430 0.000 1.000 32,000

Individualism 0.831 0.375 0.000 1.000 32,000

Masculinity 0.537 0.499 0.000 1.000 32,000

Table VI

Pair wise Correlation Matrix [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [1] Number of females - [2] Percentage of females 0.88 - [3] Female indicator 0.82 0.82 - [4] Board size 0.34 0.06 0.29 - [5] Firm age (ln) 0.02 -0.01 0.01 0.07 - [6] Total assets (ln) 0.19 0.08 0.18 0.44 0.19 - [7] Turnover (ln) 0.21 0.09 0.19 0.42 0.22 0.87 - [8] Number of employees (ln) 0.20 0.09 0.18 0.38 0.23 0.72 0.88 - [9] Power distance -0.02 -0.04 0.00 0.10 0.02 0.16 0.13 0.11 - [10] Individualism -0.04 -0.06 -0.03 0.04 -0.18 -0.18 -0.13 -0.10 -0.27 - [11] Masculinity -0.29 -0.32 -0.25 0.05 -0.06 -0.05 -0.03 -0.02 -0.07 0.22 - [12] Uncertainty avoidance -0.09 -0.10 -0.08 0.02 0.14 0.22 0.18 0.15 0.75 -0.69 0.07 - [13] Number of directors connected 0.35 0.21 0.32 0.47 0.09 0.48 0.51 0.45 0.04 0.07 -0.18 -0.08 - [14] Female labor participation 0.11 0.10 0.09 0.03 -0.02 -0.08 -0.01 -0.02 -0.28 0.37 -0.21 -0.31 0.15

0.7 may reflect multicollinearity problems in regression analysis. Additionally, Hsiao (2003, pp 3-4) argues that with panel data models, multicollinearity problems occur less often than any other data set, since in general panel data contains more observations and variation. The possible correlation between the firm size measures is obvious. To a great extent these measures similarly measure firm size. However, the fact those correlations are far from perfect leaves room for the idea that they also measure some other, unknown firm characteristic.

V. Results

This study consists of two parts for two topics demand examination. In the first part the results from the OLS and instrumental variables estimations will be presented in order to determine the effect of the extent of female directors on firm performance and whether results differ when endogeneity issues are addressed. As a follow-up on the first part of the research, the second part elaborates on the inter-country variation with respect to the extent of female directors. In other words, results of instrumental variables estimations are presented in which Hofstede’s cultural dimensions are accounted for. Additionally, subsamples are created based on countries that belong to the upper quartile with respect to Hofstede’s cultural dimensions and new instrumental variables estimation results are presented.

First part of the research

and 6 also show positive estimates for all three measures at the 1% level (β=0.037; β=0.250; β=0.082).

As omitted variables possibly bias results from OLS estimations, these need to be accounted for with firm-fixed and time-fixed effects. Hence, Table VIII shows results of OLS estimations

TABLE VII

Firm performance and the extent of female directors in European firms, 2004-2011

This table shows the results of OLS estimations without inclusion of firm-fixed and time-fixed effects, using the European sample comprising 19 countries and covering the period 2004 to 2011. Data regarding board-level characteristics are retrieved from Standard and Poor’s Capital IQ database and firm-level characteristics are retrieved from Bureau van Dijk’s Orbis database. Columns 1, 2 and 3 (Columns 4, 5, and 6) show results from return on assets (price to book ratio) regressed on the number of female directors, percentage of female directors and female indicator respectively, and where board size, firm age, total assets, turnover and the number of employees are controlled for. Return on assets equals annual net income divided by the book value of total assets. Price to book ratio equals the stock price divided by the book value of equity. Number of females equals the number of female board directors, percentage of females equals the number of female board directors divided by the total number of board directors and female indicator is a dichotomous variable that equals 1 when a firm has one or more female board directors and 0 otherwise. Board size equals the total number of board directors, firm age equals the total number of years since incorporation, total assets equals the book value of total assets, turnover equals the total revenues and the number of employees equals the total number of employees. For all firm-level characteristics the natural logarithm is used and all variables are measured at the end of the year. ***, **, * Represent significance at the 1%, 5% and 10% levels respectively. Robust standard errors are shown in parentheses under the coefficients.

Return on assets (ln) Price to book (ln)

where firm-fixed and time-fixed effects are included.5 The results from Columns 1, 2 and 3 clearly show no significant effect of the extent of female directors on return on assets. This suggests that the number of female directors, percentage of female directors and whether one or more female directors have a seat at the board does not affect short term firm performance. In contrast to Column 1, 2 and 3 the estimates in Column 4, 5 and 6 show a positive effect of the number of female directors, percentage of female directors and female indicator on the price to book ratio, at the 10%, 5% and 10% respectively (β=0.030; β=0.269; β=0.042). This suggests that the extent of female directors positively affects long term performance. In other words, shareholders value female directors and more gender diverse boards.

The contrasting results regarding the OLS estimations with and without firm-fixed and time-fixed effects, suggest that omitted variables bias results. Hence, fixed effects should be incorporated in the model to account for this bias. Table IX shows the results of the first stage instrumental variables estimations with firm-fixed and time-fixed effects.6 As expected, the number of directors that have a seat in other boards with female directors positively affects the number of female directors, percentage of female directors and female indicator, at a 1% level, with respect to both performance measures (β=0.060; β=0.005; β=0.027 for return on assets and β=0.059; β=0.005; β=0.029 for price to book ratio). This suggests that when boards are connected to outside female directors via current directors, the extent of female directors in that board significantly increases. However, the significant negative effect of female labor participation on the extent of female directors is quite surprising. With respect to both performance measures, Columns 1, 2, 4 and 5 in Table IX show that female labor participation negatively affects the number of female directors and percentage of female directors, significant at a 1% and 5% level respectively. Columns 1, 3, 4 and 6 show that board size positively affects the number of female directors and female indicator at a 1% significance level. However, Columns 2 and 5 show that board size is not related to the percentage of female directors. This suggests that board size is

5 _{The results from the Hausman test for random correlated effects show that only the estimates of the fixed effects}

are consistent. The null hypothesis of the Hausman test, the difference in the coefficients is not systematic, is rejected at the 1% significance level. The results from the Hausman tests (6 in total) are available at request by the author.

6 _{Three tests are performed on each model to validate the instruments used. The Angrist-Pischke multivariate F test}

Table VIII

Firm performance and the extent of female directors in European firms in the period 2004-2011 This table shows the results of OLS estimations with inclusion of firm-fixed and time-fixed effects to deal with omitted variables. The European sample comprises 19 countries and covers the period 2004 to 2011. Data regarding board and firm-level characteristics are retrieved from Capital IQ and Orbis respectively. Columns 1, 2 and 3 (Columns 4, 5, and 6) show results from return on assets (price to book ratio) regressed on the number of female directors, percentage of female directors and female indicator respectively, and where board size, firm age, total assets, turnover and the number of employees are controlled for. Return on assets equals annual net income divided by the book value of total assets. Price to book ratio equals the stock price divided by the book value of equity. Number of females equals the number of female board directors, percentage of females equals the number of female board directors divided by the total number of board directors, and female indicator is a dichotomous variable that equals 1 when a firm has one or more female board directors and 0 otherwise. Board size equals the total number of board directors, firm age equals the total number of years since incorporation, total assets equals the book value of total assets, turnover equals the total revenues and the number of employees equals the total number of employees. For all firm-level characteristics the natural logarithm is used and all variables are measured at the end of the year. With respect to all estimations Hausman tests are performed to determine whether a firm-fixed effect is preferred over a random effect. All Hausman tests yield significant results and reject the null hypothesis that the difference in the coefficients is not systematic at a 1% level. Hence, only the coefficients of firm-fixed effect estimations are consistent. ***, **, * Represent significance at the 1%, 5% and 10% levels respectively. Robust standard errors are shown in parentheses under the coefficients.

Return on assets (ln) Price to book (ln)

(1) (2) (3) (4) (5) (6) Number of females -0.021 0.030* (0.020) (0.015) Percentage of females 0.062 0.269** (0.154) (0.110) Female indicator -0.039 0.042* (0.033) (0.0260) Board size -0.005 -0.007 -0.006 -0.005 -0.002 -0.004 (0.006) (0.006) (0.006) (0.006) (0.006) (0.006) Firm age (ln) 0.045 0.046 0.047 -0.134** -0.134** -0.135** (0.058) (0.058) (0.058) (0.056) (0.056) (0.056) Total assets (ln) -0.759*** -0.760*** -0.758*** -0.414*** -0.414*** -0.414*** (0.061) (0.061) (0.061) (0.150) (0.150) (0.150) Turnover (ln) 0.836*** 0.837*** 0.837*** 0.020 0.020 0.020 (0.075) (0.075) (0.075) (0.080) (0.080) (0.080) Number of employees (ln) -0.185*** -0.185*** -0.185*** 0.102 0.102 0.102 (0.037) (0.037) (0.037) (0.090) (0.089) (0.090) Constant 2.321*** 2.333*** 2.317*** 5.559** 5.546** 5.547** (0.572) (0.569) (0.573) (2.226) (2.223) (2.225) Number of observations 11,815 11,815 11,815 12,114 12,114 12,114 Number of firms 2,267 2,267 2,267 2,341 2,341 2,341 Within r2 0.147 0.147 0.147 0.344 0.344 0.344

Table IX

First stage regressions: Identification of the extent of female board directors

This table shows the results of the first stage instrumental variables estimations with inclusion of firm-fixed and time-fixed effects. The European sample comprises 19 countries and covers the period 2004 to 2010; data with respect to labor participation is unavailable for the year 2011. Board, firm and country-level characteristics are retrieved from Capital IQ, Orbis and the World Bank Group respectively. Columns 1, 2 and 3 (Columns 4, 5, and 6) show results from estimating the fitted values of the number of female directors, percentage of female directors and female indicator respectively, using two instruments, where return on assets (price to book ratio) is the dependent variable in the second stage and board size, firm age, total assets, turnover and the number of employees are controlled for. Number of females equals the number of female board directors, percentage of females equals the number of female board directors divided by the total number of board directors, and female indicator is a dichotomous variable that equals 1 when a firm has one or more female board directors and 0 otherwise. Number of directors connected equals the total number of board directors that have a seat in other boards with female directors. Female labor participation equals the percentage of female labor participation within the country in which a firm is headquartered. Board size equals the total number of board directors, firm age equals the total number of years since incorporation, total assets equals the book value of total assets, turnover equals the total revenues and the number of employees equals the total number of employees. For all firm-level characteristics the natural logarithm is used and all variables are measured at the end of the year. ***, **, * Represent significance at the 1%, 5% and 10% levels respectively. Robust standard errors are shown in parentheses under the coefficients.

Number of females Percentage of females Female indicator Number of females Percentage of females Female indicator (1) (2) (3) (4) (5) (6)

Number of directors connected 0.060*** 0.005*** 0.027*** 0.059*** 0.005*** 0.029***

(0.007) (0.001) (0.005) (0.008) (0.001) (0.005)

Female labor participation -0.064*** -0.005** -0.002 -0.068*** -0.005** -0.003

(0.015) (0.002) (0.011) (0.015) (0.002) (0.010) Board size 0.083*** -0.001 0.033*** 0.083*** -0.001 0.031*** (0.005) (0.001) (0.003) (0.005) (0.001) (0.003) Firm age (ln) -0.025** -0.005 0.016 -0.055* -0.008 -0.024 (0.028) (0.004) (0.020) (0.031) (0.005) (0.021) Total assets (ln) 0.040** 0.007** 0.026** 0.015 0.001 0.006 (0.018) (0.003) (0.012) (0.016) (0.003) (0.010) Turnover (ln) -0.008 0.000 0.008 0.008 0.003 0.007 (0.017) (0.002) (0.011) (0.013) (0.002) (0.008) Number of employees (ln) -0.001 0.001 -0.009 0.006 0.002 -0.003 (0.011) (0.002) (0.008) (0.011) (0.002) (0.008) Number of observations 10,187 10,187 10,187 10,252 10,252 10,252 Number of firms 2,009 2,009 2,009 2,115 2,115 2,115 Centered r2 0.207 0.036 0.117 0.178 0.025 0.092 p-value Angrist-Pischke 0.000 0.000 0. 000 0.000 0.000 0.000 Kleibergen-Paap F-statistic 41.216 16.810 16.540 37.810 19.610 27.030 p-value Sargan-Hansen 0.347 0.347 0.328 0.667 0.585 0.201

level respectively. Columns 1, 2 and 3 show that total assets positively affects the percentage of female directors and female indicator at a 5% significance level, for return on assets.

the extent of female directors affect price to book ratio at a 5% significance level (β=0.277; β=3.079; β=0.521). However, Table X shows no evidence of a relation between the extent of female directors and return on assets. This suggests that the extent of female directors only affects long term performance and is valued by shareholders. Compared to the results from OLS estimations with fixed effects, the estimates from the instrumental variables estimations show a higher significance. This suggests that, in addition to omitted variables, other endogeneity issues bias results.

Table X

Second stage regressions: Firm performance and the extent of female directors in European firms in the period 2004-2010

This table shows the results of the second stage instrumental variables estimations with inclusion of firm-fixed and time-fixed effects. The European sample comprises 19 countries and covers the period 2004 to 2010 (data with respect to labor participation is unavailable for the year 2011). Board, firm and country-level characteristics are retrieved from Capital IQ, Orbis and the World Bank Group respectively. Columns 1, 2 and 3 (Columns 4, 5, and 6) show results from return on assets (price to book ratio) regressed on the instrumented number of female directors, percentage of female directors and female indicator respectively, and where board size, firm age, total assets, turnover and the number of employees are controlled for. Return on assets equals annual net income divided by the book value of total assets. Price to book ratio equals the stock price divided by the book value of equity. The number of females, percentage of females and female indicator are instrumented and result from the first stage. Board size equals the total number of board directors, firm age equals the total number of years since incorporation, total assets equals the book value of total assets, turnover equals the total revenues and the number of employees equals the total number of employees. For all firm-level characteristics the natural logarithm is used and all variables are measured at the end of the year. ***, **, * Represent significance at the 1%, 5% and 10% levels respectively. Robust standard errors are shown in parentheses under the coefficients.

Return on assets (ln) Price to book (ln)

Second part of the research

As the first part of the research shows, the extent of female directors positively affects long term firm performance. This makes it surprising to see the inter-country differences with respect to the extent of female board directors (e.g. Table I shows that in 2011 Portuguese boards have 4.3% female directors whereas Norwegian boards have 35.1%).

Table XI

First stage regressions: Identification of the extent of female board directors

This table shows the results of the first stage estimations with inclusion of Hofstede’s cultural dimensions and a time-fixed effect. The European sample comprises 19 countries and covers the period 2004 to 2010 (data with respect to labor participation is unavailable for the year 2011). Board, firm and country-level characteristics are retrieved from Capital IQ, Orbis and the World Bank Group respectively. Hofstede’s standardized scores are retrieved from Taras, Steel and Kirkman (2012). Columns 1, 2 and 3 (Columns 4, 5, and 6) show results from estimating the fitted values of the number of female directors, percentage of female directors and female indicator respectively, using the cultural dimensions power distance, individualism, masculinity, uncertainty avoidance and two instruments, where return on assets (price to book ratio) is the dependent variable in the second stage and board size, firm age, total assets, turnover and the number of employees are controlled for. Number of females equals the number of female board directors, percentage of females equals the number of female board directors divided by the total number of board directors, and female indicator is a dichotomous variable that equals 1 when a firm has one or more female board directors and 0 otherwise. Power distance, individualism, masculinity and uncertainty avoidance are dichotomous variables that equal 1 when a firm is headquartered in a country belonging to the upper quartile with respect to the corresponding cultural dimension, and 0 otherwise. Number of directors connected equals the total number of board directors that have a seat in other boards with female directors. Female labor participation equals the percentage of female labor participation. Board size equals the total number of board directors, firm age equals the total number of years since incorporation, total assets equals the book value of total assets, turnover equals the total revenues and the number of employees equals the total number of employees. For firm-level characteristics the natural logarithm is used and all variables are measured at the end of the year. ***, **, * Represent significance at the 1%, 5% and 10% levels respectively. Robust standard errors are shown in parentheses under the coefficients.

Number of females Percentage of females Female indicator Number of females Percentage of females Female indicator (1) (2) (3) (4) (5) (6) Power distance -0.160*** -0.029*** -0.033 -0.192*** -0.035*** -0.047** (0.032) (0.005) (0.020) (0.032) (0.005) (0.020) Individualism -0.012 0.002 -0.022 0.014 0.007 -0.014 (0.042) (0.006) (0.026) (0.040) (0.006) (0.025) Masculinity -0.441*** -0.069*** -0.211*** -0.462*** -0.073*** -0.218*** (0.022) (0.003) (0.012) (0.022) (0.003) (0.012) Uncertainty avoidance -0.214*** -0.028*** -0.131*** -0.213*** -0.028*** -0.125*** (0.025) (0.004) (0.017) (0.025) (0.004) (0.017)

Number of directors connected 0.081*** 0.010*** 0.046*** 0.088*** 0.011*** 0.050***

(0.006) (0.001) (0.003) (0.006) (0.001) (0.003)

Female labor participation 0.006* 0.001 0.003 0.000 0.000 0.000

Table XII shows that the extent of female directors positively affects both measures for firm performance at a 1% level, while controlling for Hofstede’s cultural dimensions. Hofstede’s cultural dimensions also affect both measures of firm performance. Power distance positively affects return on assets and price to book ratio, at a 5% level in Column 1, a 1% level in Columns 2, 4 and 5 and 10% level in Column 6 (β=0.109; β=0.159 for return on assets and β=0.130; β=0.177; β=0.078 for price to book ratio). This suggests that firm performance directly benefits from a power distance culture. However, as power distance negatively affects the extent of female directors (which positively affects firm performance), firm performance indirectly suffers from a power distance culture. The same holds for masculinity as it positively affects both measures of firm performance at a 1% level. With respect to uncertainty avoidance, no direct relation is found with firm performance. This suggests that uncertainty avoidance solely (negatively) affects firm performance indirectly, namely via the extent of female directors. Table XII also shows that individualism negatively affects price to book ratio at a 1% level.

Table XII

Second stage regressions: Firm performance and the extent of female directors in European firms with Hofstede’s cultural dimensions, 2004-2010

This table shows the results of the second stage instrumental variables estimations with inclusion of Hofstede’s cultural dimensions and a time-fixed effect. The European sample comprises 19 countries and covers the period 2004 to 2010 (data with respect to labor participation is unavailable for the year 2011). Board, firm and country-level characteristics are retrieved from Capital IQ, Orbis and the World Bank Group respectively, and Hofstede’s standardized scores from Taras, Steel and Kirkman (2012). Columns 1, 2 and 3 (Columns 4, 5, and 6) show results from return on assets (price to book ratio) regressed on the instrumented number of female directors, percentage of female directors and female indicator respectively and the cultural dimensions power distance, individualism, masculinity, uncertainty avoidance. Additionally, board size, firm age, total assets, turnover and the number of employees are controlled for. Return on assets equals annual net income divided by the book value of total assets. Price to book ratio equals the stock price divided by the book value of equity. The number of females, percentage of females and female indicator are instrumented and result from the first stage. Power distance, individualism, masculinity and uncertainty avoidance are dichotomous variables that equal 1 when a firm is headquartered in a country belonging to the upper quartile with respect to the corresponding cultural dimension, and 0 otherwise. Board size equals the total number of board directors, firm age equals the total number of years since incorporation, total assets equals the book value of total assets, turnover equals the total revenues and the number of employees equals the total number of employees. For all firm-level characteristics the natural logarithm is used and all variables are measured at the end of the year. ***, **, * Represent significance at the 1%, 5% and 10% levels respectively. Robust standard errors are shown in parentheses under the coefficients.

Return on assets (ln) Price to book (ln)

Tables XV and XVI show results in line with expectations. Table XV shows that within a society characterized by high masculinity there is no relation between the extent of female directors and firm performance. Table XVI shows that there is no relationship between the extent of female

Table XIII

Second stage regressions: Firm performance and the extent of female directors in European firms headquartered in countries characterized by high power distance in the period 2004-2010 This table shows the results of the second stage instrumental variables estimations with inclusion of firm-fixed and time-fixed effects. The European sample covers the period 2004 to 2010 (data with respect to labor participation is unavailable for the year 2011) and is limited to those countries belonging to the 25% highest scores on the power distance cultural dimension. Board, firm and country-level characteristics are retrieved from Capital IQ, Orbis and the World Bank Group respectively. Hofstede’s standardized scores are obtained from Taras, Steel and Kirkman (2012). Columns 1, 2 and 3 (Columns 4, 5, and 6) show results from return on assets (price to book ratio) regressed on the instrumented number of female directors, percentage of female directors and female indicator respectively. Additionally, board size, firm age, total assets, turnover and the number of employees are controlled for. Return on assets equals annual net income divided by the book value of total assets. Price to book ratio equals the stock price divided by the book value of equity. The number of females, percentage of females and female indicator are instrumented and result from the first stage. Board size equals the total number of board directors, firm age equals the total number of years since incorporation, total assets equals the book value of total assets, turnover equals the total revenues and the number of employees equals the total number of employees. For all firm-level characteristics the natural logarithm is used and all variables are measured at the end of the year. ***, **, * Represent significance at the 1%, 5% and 10% levels respectively. Robust standard errors are shown in parentheses under the coefficients.

Return on assets (ln) Price to book (ln)

directors and firm performance within societies characterized by uncertainty avoidance. These findings suggest that both short and long term performance of firms within such cultures does not depend on the extent of female directors. Compared to the findings from the first part of research,

Table XIV

Second stage regressions: Firm performance and the extent of female directors in European firms headquartered in countries characterized by high individualism in the period 2004-2010 This table shows the results of the second stage instrumental variables estimations with inclusion of firm-fixed and time-fixed effects. The European sample covers the period 2004 to 2010 (data with respect to labor participation is unavailable for the year 2011) and is limited to those countries belonging to the 25% highest scores on the individualism cultural dimension. Board, firm and country-level characteristics are retrieved from Capital IQ, Orbis and the World Bank Group respectively. Hofstede’s standardized scores are obtained from Taras, Steel and Kirkman (2012). Columns 1, 2 and 3 (Columns 4, 5, and 6) show results from return on assets (price to book ratio) regressed on the instrumented number of female directors, percentage of female directors and female indicator respectively. Additionally, board size, firm age, total assets, turnover and the number of employees are controlled for. Return on assets equals annual net income divided by the book value of total assets. Price to book ratio equals the stock price divided by the book value of equity. The number of females, percentage of females and female indicator are instrumented and result from the first stage. Board size equals the total number of board directors, firm age equals the total number of years since incorporation, total assets equals the book value of total assets, turnover equals the total revenues and the number of employees equals the total number of employees. For all firm-level characteristics the natural logarithm is used and all variables are measured at the end of the year. ***, **, * Represent significance at the 1%, 5% and 10% levels respectively. Robust standard errors are shown in parentheses under the coefficients.

Return on assets (ln) Price to book (ln)