THE IMPACT OF GENDER QUOTAS ON FIRM
PERFORMANCE
STUDENT INFORMATION
Name Lotte Grol
Student number 10752188
Programme BSc. Economics and Business
Track Finance and Organization
Field of BSc. Thesis Organization Economics
Name supervisor Silvia Dominguez Martinez
Date final version January 31st, 2017 ABSTRACT
This paper contributes to existing literature on the relationship between corporate board gender quotas and the financial performance of firms. Using OLS regressions, firm-specific data for the years 2006 and 2010 was compared between affected Norwegian public limited liability firms and Danish public limited liability firms that were unaffected by the rule. The first research question relates to the impact of gender quotas on firm performance, whereas the second research question looks at gender diversity and its effect on firm performance. For the first research question, no significant relationship between gender quotas and firm performance was found. Regarding the second research question, it was found that an increased percentage of female board members has a marginally significant positive impact on firm performance in Norway. The optimal percentage of female board members was found to be 32%. For Denmark, the results were inconclusive.
STATEMENT OF ORIGINALITY
This document is written by Lotte Grol, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
INDEX
1. INTRODUCTION ... 3
2. LITERATURE REVIEW ... 4
2.1 Context: The Gender Quota in Norway ... 4
2.2 Advantages of Gender Quotas ... 6
2.3 Disadvantages of Gender Quotas ... 7
2.5 Hypotheses ... 9
3. DATA AND METHODOLOGY ... 9
3.1 Dependent Variable ... 11
3.2 Independent Variables ... 11
3.3 Control Variable ... 12
3.4 Regression Analysis ... 12
4. EMPIRICAL RESULTS AND ANALYSIS ... 13
4.1 The Impact of Gender Quotas on Firm Financial Performance ... 13
4.2 Gender Diversity and Firm Financial Performance ... 17
5. DISCUSSION ... 20 5.1 Sample Size ... 20 5.2 Board Characteristics ... 20 5.3 Selection Bias ... 20 5.4 Time Frame ... 20 5.5 Industry ... 21 6. CONCLUSION ... 22 7. REFERENCES ... 23 8. APPENDIX ... 25
1. INTRODUCTION
The underrepresentation of women on corporate boards is currently one of the most significant corporate governance issues. Despite substantial progress over recent years, women are still outnumbered by men in leading positions all over the world. According to research done by the European Commission in 2016, women in Europe hold on average only 23.3% of board positions in publicly listed companies and only 5.1% of CEO positions are occupied by women (European Union, 2016). The intangible barrier to advancement in profession for women is often referred to as the glass ceiling. In an effort to ‘break’ this glass ceiling, many policymakers have begun adopting gender quotas for corporate boards of directors.
The political debate on female representation and gender quotas is largely influenced by the discussion of whether the quota adds value to the firm. Some papers conclude that gender quotas have a negative impact on firm value. Ahern and Dittmar (2012) conclude that stock prices decline at the announcement of a legislative quota, followed by a large decrease in Tobin’s Q in the years thereafter. Matsa and Miller (2013) also found a negative correlation between gender quotas and firm performance. However, other researchers have found positive effects of gender quotas on firm performance. For example, Ferrari et al (2016) found a positive correlation between the introduction of a quota law and stock prices. Reguera-Alvarado et al (2017) conclude that the increase in female board members that follows from the introduction of a quota results in better financial performance. Thus overall, current research on the correlation between gender quotas and firm performance is inconclusive. The aim of this paper is to contribute to existing research by providing new insights on the causal relationship between legislative quotas for women and firm performance. Therefore, the first and main research question addresses the direct correlation between gender quotas and firm performance, and is formulated as follows: “Do legislative quotas for women on corporate boards have a positive impact on the financial performance of firms?” Since Norway was the first country to adopt a mandatory gender quota, this question studies the effect of the Norwegian gender quota on firm performance, comparing affected Norwegian firms to Danish firms that were unaffected by the rule. The years 2006 and 2010 were chosen because the mandatory gender quota was introduced in 2008 in Norway. This lies in the middle of the time frame.
In addition to this question, it is interesting to discover whether increased gender diversity increases firm value.Previous papers on the impact of gender diversity on firm performance provide varying results as well. On the one hand, Adams and Ferreira (2009) find gender diversity has a negative impact on firm performance. On the other hand, Hoogendoorn et al (2013) conclude
that diverse teams perform better than teams with more male members. Carter et al (2003) find similar results. This paper intends to find new evidence on the relationship between gender diversity in corporate boards and firm performance. Therefore, the second question is as follows: “Does gender diversity in corporate boards have a positive impact on the financial performance of firms?” By including this second research question, this paper distinguishes itself from previous research considering either the impact of gender quotas on firm performance, or the impact of gender diversity on firm performance. In this paper, the link between gender quotas, gender diversity and firm performance is investigated.
In Figure 1, a visualization of the research question is provided, in which the numbers indicate the two different research questions.
Figure 1: Visualization of the research questions
The structure of this thesis is as follows. In chapter two, a review of the existing literature will be given. Thereafter, chapter three will describe the used dataset and variables. It will explain the methodology behind the regression model, and the hypotheses will be stated. In chapter four, a conclusion is drawn from the results. Also, possible limitations and suggestions for further research will be discussed in section five, followed by the final conclusion in chapter six.
2. LITERATURE REVIEW
This section will elaborate on the origin, advantages and disadvantages of legislative gender quotas. Gender diversity will also be evaluated. Following existing literature, the hypotheses of this thesis will be stated and explained.
2.1 Context: The Gender Quota in Norway
Norway was the first country to officially respond to the issue of female
underrepresentation. In 1981, Norway introduced the first gender quota. A quota is “a percentage
1
target that mandates a proportional representation of a particular group” (Kogut et al, 2014, p. 892). The gender quota in Norway legislated a minimum representation of female board members. This quota was restricted to government appointed boards, councils and committees. In 2001, the government started evaluating a more expansive quota policy, which would also apply to the boards of public limited liability companies. The first proposal for this legislative quota was submitted in 2002, and the law was passed in December 2003. At that time, only 9% of directors were women (Ahern & Dittmar, 2012, p. 137). According to the new law, all public limited liability companies (or ASA firms) were required to have at least 40% representation of each sex. At this point, there were no sanctions for companies that did not comply with the law. As a result, by 2005 the percentage of female board members was still only 17%. Therefore, the law became compulsory on January 1, 2006. The Norwegian government introduced sanctions to firms that did not
conform to the female quota of 40%. Those firms were sanctioned with forced dissolution if they did not comply with the law on female representation by January 1, 2008. The sanctions proved to be successful, as by 2009, the median percentage of female board members was 40% (Bertrand et al, 2017, p. 3). Following Norway, many other European countries adopted similar quotas. Belgium, France, Germany, Iceland, Italy and Spain are now also subject to binding laws on female
representation.
The Norwegian government imposed the gender quota law without the consent of the affected firms. Overall, ASA companies were opposed to the introduction of the quota. They were mainly complaining that there was a lack of talented and willing female directors. Furthermore, the purpose of the quota was to increase gender equality, not to improve firm performance, which was not in line with the main interests of most firms. Between 2003 and 2009, the percentage of female board members in public limited liability firms increased by 37 percentage points. In contrast, for private limited liability firms not affected by the gender quota, this increase was merely 3 percentage points (Ahern & Dittmar, 2012, p. 145). This implies that the increase in female directors was pushed by the implementation of the gender quota law. According to prior studies, the quota policy even induced exit out of the ASA form. Research done by Ahern and Dittmar (2012) and Bøhren and Staubo (2014) shows that delisting from the Oslo Stock Exchange was more common among firms with a smaller share of female board members before the introduction of the quota. 46.3% of the ASA firms that had no female directors in 2002 had delisted by 2009 (Ahern & Dittmar, 2012, p. 185). This further implies that the law was introduced without the support of ASA firms.
Despite initial negative reactions to the quota, there are many arguments in favour of a female quota, described in the next section.
2.2 Advantages of Gender Quotas
As elaborated on in the first subsection of this chapter, gender quotas were imposed to increase the number of female board members. There are several arguments in favour of legislative gender quotas. First of all, according to agency theory women are overall better monitors and more independent directors. The rationale behind gender quotas can be explained through the lens of agency theory, as defined by Fama and Jensen (1983). Agency theory is a theoretical framework that is often used to understand the correlation between board characteristics and firm value. In an agency framework, the role of the board of directors is to monitor the actions of the managers and shareholders, in order to resolve agency problems. The main question is whether a board with more female directors is better at monitoring management, and less likely to undermine the interests of shareholders. Adams and Ferreira argue that female directors are tougher monitors than male directors (2009, p. 301). This finding is supported by the argument that women are more willing to break with established patterns, ask more questions, and bring new perspectives (Matsa & Miller, 2013, p. 161). Also, it is important that outside directors are independent. Since female directors do not belong to the informal system used by powerful men to help others with the same educational or social background referred to as the “old boys network”, they are likely to be more independent directors (Adams & Ferreira, 2009, p. 292).
Another advantage of legislative quotas for women is that it helps break the glass ceiling, as was discussed in the introduction. There are two possible explanations why women have been underrepresented in corporate boards. Firstly, women are not part of the “old boys network”. Quotas give talented women who are not supported by this network the opportunity to still climb the corporate ladder (Bertrand et al, 2017, p. 4). Another explanation for the underrepresentation of women is that business prejudice towards women prevents them from being hired. By
implementing a quota, discrimination against women is counteracted.
Third of all, according to Comi et al, quotas may result in spillover effects in the labour market (2016, p. 2). Those spillover effects have two implications. First of all, having more women on the corporate board may result in the hiring or promotion of other women. This does not only hold for board positions, but also for other managerial positions within the firm. Secondly, a higher representation of women could lead to a more convenient working environment for women, in which policies take their family lives into account. This standpoint is, however, not supported by all researchers. According to Bertrand et al, legislative quotas do not create large positive spillover effects in the short to medium term (2017, p. 22).
Finally, in the long run, quotas may incentivize young women to choose an education in line with their aspired career. Young women that become aware of the introduction of the quota
may expect to professionally benefit from the reform, and choose to graduate in fields such as management, finance, economics and law (Comi et al, 2016, p. 2). Research done by Bertrand et al (2017) partially supports this argument. Bertrand et al found that among undergraduates in Norway, there was a significant increase in females choosing business majors after the introduction of the quota. However, there was no evidence of a significant such increase among graduate degrees (2017, p. 26).
2.3 Disadvantages of Gender Quotas
As mentioned in the introduction, many business leaders were opposed to the law. This section will elaborate more on the disadvantages of gender quotas.
One of the main arguments against gender quotas is that not enough qualified women can be found to serve on boards. In Norway, this concern was so severe that many public limited liability firms publicly expressed their concern after the proposal for the gender quota. The Norwegian government responded by creating a database of talented women interested in a position on a corporate board (Bertrand, 2017, p. 7).
Secondly, opponents of gender quotas argue that, if talented female directors cannot be found, the effect of the quota will become counterproductive. Negative stereotypes may be
reinforced, and women may believe that they do not have to be very skilled or talented to become a board member. This may even lead to a demotivating environment for women, in which they invest less in their careers (Bertrand et al, 2017, p. 4).
Additionally, gender quotas may not have a positive effect on firm value. The main argument for this is that the purpose of most boards is to maximize firm value and they compile their boards accordingly. Therefore, it would be sub-optimal to introduce binding constraints in terms of board composition (Comi et al, 2016, p. 2).
The fourth argument against gender quotas regards the impact of the quota on the women themselves, after they are hired. Lehman (2013) researched discrimination of minority groups in the legal profession. He investigated law-firms that were constrained by a hiring policy that was aimed at hiring more minorities. He found that more minorities were hired in leadership positions, but that they were more likely to be given worse tasks and were less likely to be
promoted. The pressure that the law-firms were subject to in this research is similar to the pressure that ASA firms experienced through the quota in Norway. This suggests that female board
members that are hired might be discriminated by receiving less challenging tasks and by not being promoted, since they are only hired because of their gender.
The introduction of gender quotas may also cause conflicts within the corporate board. If male board members believe that the new female directors are only hired on the grounds of the law, and not based on their talents, they may not be very welcome to the new female directors (Comi et al, 2016, p. 2). The conflicts that arise can have negative impact on firm strategy and management.
2.4 The Link Between Gender Quotas, Gender Diversity and Firm Performance
As shown in Figure 1, the research question of this paper addresses the link between gender quotas, gender diversity and firm performance. Historic data shows that gender diversity in corporate boards has increased as a result of binding gender quota laws (Bertrand, 2017, p. 3). Knowing this, and taking into account the advantages and disadvantages of gender quotas, it is important to evaluate whether gender diversity is beneficial to firm performance and to find out if this is different in a country with a quota as compared to a country without a quota. Therefore, the advantages of gender diversity are evaluated.
The first argument in favour of gender diversity is that heterogeneous groups are more creative than homogeneous groups (Seierstad et al, 2017, p. 148). This argument can be illustrated by a quote from the CEO of Sun Oil, Robert Campbell, from his letter to the Wall Street Journal: “Often what a woman or minority person can bring to the board is some perspective a company has not had before— adding some modern-day reality to the deliberation process. Those perspectives are of great value, and often missing from an all-white, male gathering. They can also be
inspiration to the company’s diverse workforce” (Campbell, 1996). Moreover, it was found that diverse groups are better at problem solving (Carter et al, 2003, p. 2). The mixture of perspectives that follows from a diverse group composition promotes the evaluation of more alternative options. Also, the consequences of each option are explored more extensively. Another argument suggested by Harjoto et al (2015) is that by creating a more diverse corporate board, the interests of the different stakeholders are better attended to.
2.5 Hypotheses
To sum up, there are arguments for and against gender quotas, and current research on the impact of gender quotas on firm performance is inconclusive. From the literature review provided in this paper, the arguments in favour of gender quotas appear stronger and suggest that gender quotas have a positive impact on firm performance. Therefore, in this paper the following hypothesis is investigated:
H1: There is a significant positive relationship between gender quotas and firm performance. The research topic of this thesis is twofold. On the one hand, it investigates the impact of
legislative gender quotas, and on the other hand, it evaluates the impact of gender diversity. There is substantial evidence suggesting that gender diversity is beneficial to firm performance, as was discussed in the previous subsection. In order to address the second research question, the second hypothesis is formulated as follows:
H2: There is a significant positive relationship between gender diversity and firm financial performance. 3. DATA AND METHODOLOGY
In this section, the datasets and data sources that were used will be stated. Furthermore, the empirical model will be described and the statistical hypotheses will be given.
A panel of public limited liability companies from Norway and Denmark are analysed in the years 2006 and 2010, using (financial) data from the DataStream database and the annual reports of the companies. This time period was chosen for the following reason. As stated in the literature review, the gender quota was introduced on a voluntary basis in Norway in December 2003, made
mandatory in January 2006, and sanctions were introduced in January 2008. Therefore, in order to analyse the impact of the gender quota, a comparison will be made between 2006 and 2010: two years before, and two years after the introduction of sanctions for not complying with the quota. Firm-specific data will be collected for January 1st, 2006 and January 1st, 2010.
There are two types of limited liability firms in Norway. Private limited liability firms, abbreviated AS, and public limited liability firms, abbreviated ASA. The gender quota law in Norway applies only to public limited liability firms, or ASA firms. Therefore, the sample for this research consists of Norwegian ASA firms from the OBX Index. The OBX Index lists the 25 most
liquid firms on the Oslo Stock Exchange. The firms that are listed on the OBX Index were found using the DataStream database. In order to compare company results over time, only firms that were active in both 2006 and 2010 could be used. This resulted in a sample of 16 Norwegian firms. As explained in the second chapter, a possible explanation for the drop out of 9 ASA firms could be the gender quota, that caused various firms to change their legal status to private limited liability firm. However, after examination of the OBX Index, no evidence of this explanation was found. None of the firms that were dropped from the index after 2006 have become private. The main reasons for dropping out of the index were bankruptcy, possibly due to the financial crisis, and changing to a different stock exchange. The list of the 16 firms that were used is shown in Table 4 in the Appendix. In Table 5, given in the Appendix, the observable firm characteristics are given.
In order to answer the first research question, two types of tests are used. First of all, because of the small sample size, a non-parametric Wilcoxon matched-pairs signed-rank test is performed to measure the difference in the Tobin’s Q’s between 2006 and 2010. Additionally, an OLS regression is performed to control for firm size. For this regression, the
difference-in-differences comparison is used. This approach is used to compare two groups over two time periods, in which one of the groups is exposed to a treatment in the second period, whereas the other group is not exposed to a treatment in either of the periods. In this case, the treatment is the obligation of the gender quota in Norway, in January 2008. As the control group, Denmark is used, as there was no obligatory quota introduced in Denmark between 2006 and 2010. The difference-in-differences approach is favourable, because it removes biases from comparisons over time (NBER lecture, 2007, p. 1). The introduction of the quota in Norway took place during the financial crisis. During the financial crisis, the economic environment was very unstable, which may have impacted the financial performance of firms. However, the impact of the financial crisis is comparable for both countries. The difference-in-differences approach should correct for this bias. For the second research question, an OLS regression is done, which will be elaborated upon in the final part of this chapter.
The sample of Danish firms is derived from the OMX Copenhagen 20, consisting of the 20 most-traded stocks. The firms that are listed on the OMX Index were also found using the
DataStream database. After deletion of firms that were not active in both time periods, 13 Danish firms remained. For the OMX Index, it was found that the firms that dropped out after 2006 all switched to a different stock index.In Table 6, shown in the Appendix, the observable
characteristics for the Danish firms are presented.
Denmark was chosen as the control group, because of the various similarities between the two countries. Both countries are OECD high-income countries (World Bank Group, 2016, p. 43).
Also, the legal systems in Scandinavian countries are similar (Ahern & Dittmar, 2012, p. 163). In 2016 female labour force participation was 76% in both countries (World Bank Group, 2016, pp. 114 & 186).
3.1 Dependent Variable
The main issue of this thesis is the effect of the gender quota on the financial performance of firms. Therefore, the dependent variable used in the regression analysis is a measure of the firms’ financial performance. In previous studies, the measurements of firm performance vary considerably. Overall, there are two approaches used, either accounting measures or Tobin’s Q. Following literature from Adams and Ferreira (2009), Ahern and Dittmar (2012), and Carter, Simkins and Simpson (2003), Tobin’s Q is used as a measure of firms’ financial performance in this paper. Tobin’s Q is calculated as the ratio of a firm’s market value to the replacement value of its assets (Adams & Ferreira, 2009, p. 293). The focus on Tobin’s Q rather than accounting measures is because, in a way, it measures the existence of intellectual capital. Since Tobin’s Q measures the relationship between the market value of a firm and the replacement value of its physical assets, a high value implies greater intellectual capital. Greater intellectual capital, in turn, is correlated with higher financial performance (Reguera-Alvarado et al, 2017, p. 343). This is relevant for this research because it is interesting to discover whether an increase in female board members affects the intellectual capital, and thus financial performance, of firms. The Tobin’s Q was retrieved using the DataStream database. For robustness, the regressions are also run using return on assets as a dependent variable. Return on assets is calculated as net income over total assets.
3.2 Independent Variables
There are two research questions to be answered in this paper. The dependent variable is the same for both questions, namely Tobin’s Q. The independent variables, however, differ for the two research questions. The first research question investigates the impact of the quota. The first independent variable in this regression is a dummy variable for NORWAY. For Norway the dummy variable is 1, for Denmark it is 0. Moreover, to investigate the difference before and after the introduction of the mandatory quota, the variable 2010 is added. This is a dummy variable, which is 0 for 2006 and 1 for 2010. To interact the two variables, the interaction variable NORWAY*2010 is added.
The second research question focuses on the impact of gender diversity on firm
performance. As an independent variable, the percentage of female board members is used. This percentage could not be found through the Orbis database, as it lacks data on historic board
composition. Therefore, the composition of the Board of Directors for both years was collected from the annual reports of each individual firm. In some cases, the gender was unspecified in the report. In those cases, they were looked up on the firm’s website, to find indications of gender. Otherwise, the LinkedIn pages of the directors concerned were looked up. The percentage of female board members was calculated by dividing the number of female board members by the total number of board members.
Hoogendoorn et al (2013) found that the relationship between firm performance and share of women is non-linear. The correlation that they found is inverse and U-shaped: for a percentage of women between 20% and 50% firm performance increased, whereas for percentages higher than 50% firm performance decreased. To correct for this non-linearity, the percentage of female board members is squared.
Since the aim is to investigate whether the impact of gender diversity is different for a country with a legislative quota and a country without a legislative quota, this regression is performed for both countries separately and only for the year 2010 so that the quota is incorporated.
3.3 Control Variable
In addition to the dependent variable and independent variables, a control variable is added to each regression analysis. In accordance with Reguera-Alvarado (2017), firm size is used as a control variable for both regressions. Firm size is calculated as total assets. Firm size is chosen as the control variable to control for the possibility that the size of the firm financial performance influences the financial performance of the firm, since large firms are expected to have more market power. The data on firm size was derived from the DataStream database for each firm. Conform Adams and Ferreira (2012), and Carter et al (2003) the natural logarithm of total assets is used. Firm size is the only control variable that is used in the regression. Using many control variables is problematic because of the small sample size. The difference-in-differences approach is used to control for many factors.
3.4 Regression Analysis
To answer the research question, two regression analyses are performed. The first regression measures the impact of the quota on the Tobin’s Q of the firms in question. This first model is specified as:
Since it is expected that there is a significant positive relationship between gender quotas and firm performance, the statistical hypothesis following from this model is: H1: β3 > 0.
A second regression is used to measure the effect of adding more female directors to the corporate board on firm performance. This regression analysis is performed for Norway and Denmark separately, for the year 2010.
For each country, the second model is specified as:
TOBINQ = α +β1FEM + β2FEM2 + β
3LNSIZE + ε
For the second research question, it is predicted that gender diversity has a positive impact on firm performance. This implies that β1 should be > 0. Since it is expected that the relationship between the percentage of female board members and firm performance is non-linear and inverse,
β2 should be < 0. The statistical hypothesis following from this is: H1: β1 < 0 and β2 < 0. 4. EMPIRICAL RESULTS AND ANALYSIS
The results obtained from the aforementioned regressions are presented in this chapter. 4.1 The Impact of Gender Quotas on Firm Financial Performance
In this subsection, the impact of gender quotas on firm financial performance will be analysed in order to answer the main research question. The first step in addressing this question is evaluating the descriptive statistics of the relevant variables. In Table 1, the means of the relevant variables are shown. An elaborate summary of the descriptive statistics of the variables used can be found in Table 7, in the Appendix. From the descriptive statistics, the following observations are made.
For Norway, it is notable that the percentage of female board members increased by 4.67% in 2010, with respect to 2006. However, evaluating this effect using the Wilcoxon matched-pairs signed-rank test, it is found that this increase was not statistically significant (p = .2502). For Denmark, the increase in the number of female board members was only 2.94%. Again, this effect is insignificant (p = .3621). Using the Wilcoxon rank-sum test, it is tested whether there is a significant difference between Norway and Denmark regarding the change in the percentage of female board members. The results derived from this test were also statistically insignificant.
A striking observation from Table 1 is that the mean percentage of female board members in Norway is less than 40% in 2010, namely 37.43%. It was expected that, with the implementation of the mandatory quota in January 2008, 40% of all corporate boards should be female by 2010. As explained in the second chapter, the mandatory quota law involved the punishment that
noncomplying firms had to be resolved. The minimum level of female board members in Norway was 11.11% in 2010, which is far from 40%.
From the descriptive statistics it is also found that, on average, the board size in Denmark is larger than in Norway. In 2006, the mean number of board members was 7.50 in Norway and 10.23 in Denmark. In 2010, the means were 8.25 and 10.62 respectively. Since a larger board size makes it more difficult to reach a higher percentage of female board members, the large board sizes in Denmark provide part of an explanation for the lower percentage of female board members in Denmark.
In Denmark, the mean percentage of female board members was lower than in Norway, for both time periods. The difference between Norway and Denmark was 17.24% in 2006 was and in 2010 it was 18.97%. In 2006, the noncompliance sanctions were not yet implemented in Norway, which suggests that not all firms were following the gender quota law. However, the mean percentage of female board members in Norway was already much higher than the mean percentage of female board members in Denmark at that time. A possible explanation for this could be a more welcoming atmosphere towards women in the workplace.
The mean size of firms in both countries, calculated as the natural logarithm of total assets, is comparable.
Table 1: Means of regression variables
Norway
2006 2010
Tobin’s Q 2.11 1.52
Total number of board
members 7.50 8.25
Number of female board
members 2.44 3.06
% of female board members 0.33 0.37
LnSize 17.32 17.57
Denmark
2006 2010
Tobin’s Q 2.69 2.14
Total number of board
members 10.23 10.62
Number of female board
members 1.77 2.00
% of female board members 0.16 0.18
LnSize 17.82 18.19
To answer the first research question, first of all, a non-parametric Wilcoxon matched-pairs signed-rank test is performed to determine the trend of the financial performance of the firms in both countries. A non-parametric test is chosen because of the small sample size. The results of the non-parametric test are presented in Table 9, in the Appendix. For both countries, it was found that the Tobin’s Q’s were significantly lower in 2010 than in 2006 (for Norway, p = .0052 and for Denmark, p = .0392). It is expected that this significant decrease in firm financial performance was mainly caused by the financial crisis of 2007-2008. The Wilcoxon matched-pairs signed-rank test for ROA reveals similar results; the returns on assets were significantly lower in 2010 than in 2006 for Norway (p = .0131). The same test for Denmark revealed a statistically insignificant negative effect (p = .8613). In order to find out whether the decrease in firm performance was significantly different for Norway than for Denmark, a Wilcoxon rank-sum test is performed. From this test, it is concluded that the difference in Tobin’s Q was not statistically significant (p = .7257). In contrast, the difference in return on assets is statistically significant (p = .0226). The finding that there was a strong decrease in the financial performance of firms during the time frame before taking into account the effect of a legislative quota predicts that it may be difficult to find statistically significant results for the impact of the gender quota.
Afterwards, in order to control for firm size, a parametric OLS regression is performed. The first OLS regression model uses Tobin’s Q as a dependent variable, and as independent variables, dummies for Norway, 2010 and an interaction of the two are used. The results of this regression are
shown in column (1a) of Table 2 on the next page. For robustness, in column (2a) of Table 2, the results are presented for a similar analysis, but return on assets (ROA) is used as a dependent variable instead of Tobin’s Q.
The majority of the findings from this first regression were not precise enough to reach statistical significance. For both dependent variables it holds that only the control variable and constants were statistically significant at α=0.01. In the regression with Tobin’s Q, the impact of the quota, derived from the interaction variable NORWAY2010, is small and insignificant (p = .892). In the regression with ROA, the impact of the quota is also insignificant (p = .242). From this finding, it is derived that the decrease in firm performance is not significantly stronger for Norway than for Denmark, implying that the impact of the quota was not significant. For robustness, both
regressions are also performed without incorporating the control variable firm size. The findings from this regression were not statistically significant. The results are presented in column (1b) and (2b) of Table 2.
As was concluded earlier from the descriptive statistics, an anomaly was found in the data for Norway. One firm in Norway, Royal Caribbean Cruises, had only 11.11% female board
members. This could be an indicator for the insignificance of many of the results from the first regression, since according to these results the implementation of the quota was not translated into a significant increase in the amount of female board members. However, running the regression without this firm does not result in significant findings. The results of this regression are presented in Table 8, in the Appendix.
The coefficients of determination (R2) are moderate for both regressions. This means that a relatively small proportion of the variance in the dependent variable is explained by the variables in the model. This finding is expected, since firm performance is influenced by many different factors.
To sum up, from the first regression analysis, it cannot be concluded that there is a significant impact of gender quotas on firm performance. This finding is in line with many previously published papers that find no significant effect of gender quotas on the financial performance of firms. The hypothesis that gender quotas have a positive effect on firm performance can therefore not be confirmed by this research.
Table 2: OLS Regression Model
The impact of gender quotas on firm financial performance
TOBINQ ROA (1a) (1b) (2a) (2b) Constant 10.07308*** (5.86) 2.687033*** (6.27) 39.48253*** (4.56) 8.634615*** (4.18) NORWAY -.778833 (-1.56) -.5729036 (-.99) 2.303442 (0.91) 3.16351 (1.14) 2010 -.3957755 (-0.75) -.5488206 (-.91) .1253489 (0.05) -.5138462 (-.18) NORWAY2010 -.0958507 (-0.14) -.0425978 (-.05) -4.206065 (-1.18) -3.983654 (-1.01) LNSIZE -.4144466*** (-4.40) -1.730942*** (-3.65) Number of observations 58 58 58 58 R2 0.3197 0.0708 0.2467 0.0575
Notes: t-values are given in parentheses below the estimated coefficients, *=significant at α=0.10, **=significant at α=0.05, ***=significant at α=0.01
Variables: TOBINQ (approximation of Tobin’s Q), ROA (return on assets, %), NORWAY (dummy variable for Norway), 2010 (dummy variable for 2010), NORWAY2010 (interaction between NORWAY and 2010), LNSIZE (natural logarithm of firms’ total assets), R2 (coefficient of determination: indicator of the
percentage of the total variance explained by the model)
4.2 Gender Diversity and Firm Financial Performance
This part focuses on the second research question: “Does gender diversity in corporate boards have a positive impact on firm financial performance?” To answer this question, the second OLS
regression is performed.
From the regression performed for Norway using Tobin’s Q as dependent variable, it is found that the percentage of female board members has a marginally significant positive impact on firm financial performance (p = .073). Additionally, as discussed in subsection 3.2, the issue of non-linearity has to be taken into account. By incorporating FEM2 into the regression, the optimal percentage of female board members can be derived from this regression by taking the derivative of the regression function with respect to FEM. The result for Norway is an optimal level of 32.39% female board members. This level is marginally significant at α=0.10. Thus, the relationship between the percentage of female board members and Tobin’s Q is optimal if 32.39% of the corporate board is female. This suggests that, in Norway, this relationship is non-linear, implying that some amount of diversity is desirable, but extremes are not good for firm financial
added. The relationship with ROA cannot be determined from this regression, since none of the coefficient estimates are significant.
For Denmark, the coefficients were not statistically significant. The insignificance of these results can indicate one of two things: either the statistical power of the test was too low, or there is no significant effect of gender diversity on firm performance for firms in Denmark. In the
discussion the insignificance of the results will be further elaborated upon.
For robustness, both regressions are also run without incorporating FEM2. The findings are presented in columns (1b) and (2b), and are all statistically insignificant, except the constant and firm size.
Even though the results for the regression with return on assets as a dependent variable are insignificant, it is striking that all coefficients are the opposite of the coefficients of the regression with Tobin’s Q as a dependent variable. A possible explanation for the contrasting results is that the two performance measures express different meanings. Return on assets is a measure of operational performance, whereas Tobin’s Q is used to measure growth opportunities for a company. This implies that Tobin’s Q should capture the value of intangible assets as well as operational performance.
From the findings of the second regression, it can be concluded that for Norway a positive effect between gender diversity and firm performance was found, whereas for Denmark the evidence was inconclusive.
Table 3: OLS Regression Model
The impact of gender diversity on firm financial performance
Norway 2010 TOBINQ ROA (1a) (1b) (2a) (2b) Constant 3.379439*** (3.95) 4.070892*** (4.74) 28.67583 (1.35) 27.15787 (1.46) FEM 7.333165* (1.96) .1681171 (.21) -21.15413 (-.23) -5.424566 (-.31) FEM2 -11.32158* (-1.95) 24.85448 (.17) LNSIZE -.1643739*** (-3.66) -.1486512*** (-3.05) .9803565 (-.88) -1.014873 (-.96) Number of observations 16 16 16 16 R2 0.5604 0.4204 0.0846 0.0823 Denmark 2010 TOBINQ ROA (1a) (1b) (2a) (2b) Constant 6.423081 (1.31) 8.807898* (2.00) 29.5051 (1.41) 42.35519* (2.18) FEM 16.58453 (.96) -1.21164 (-.23) 77.25723 (1.04) -18.63366 (-.80) FEM2 -50.53271 (-1.08) -272.2851 (-1.36) LNSIZE -.2851888 (-1.11) -.3543556 (-1.41) -1.320162 (-1.21) -1.692853 (-1.53) Number of observations 13 13 13 13 R2 .2875 0.2872 .4087 0.1960
Notes: t-values are given in parentheses below the estimated coefficients, *=significant at α=0.10, **=significant at α=0.05, ***=significant at α=0.01
Variables: TOBINQ (approximation of Tobin’s Q), ROA (return on assets, %), FEM (% of female board members), FEM2 (% of female board members squared), R2 (coefficient of determination: indicator of the
5. DISCUSSION
This section will give a critical discussion of the limitations of this research and suggestions for further research will be provided.
5.1 Sample Size
The first possible limitation of this research is sample size. The OBX and OMX stock exchanges that were used in this paper both consisted of a limited number of firms, namely 25 and 20
respectively. Since the firms had to be active in 2006 as well as 2010, some firms had to be dropped from the list as well. This resulted in only 16 Norwegian, and 13 Danish firms. In further research, it will be valuable to include more firms from different stock exchanges or more (Scandinavian) countries into the regression to increase the sample size.
5.2 Board Characteristics
In the research set-up that was used, the only board member characteristic that was accounted for was gender. However, besides gender, there are many other characteristics that contribute to successful board members. In further research, characteristics such as age, education, and tenure should be incorporated.
5.3 Selection Bias
In the sample, there might have been selection bias. Selection bias occurs when the data that is gathered is non-random. As discussed in the second chapter, only a selection of the firms from the OBX and OMX indices were used in the sample, because not all firms were active in both 2006 and 2010. In line with theory, it was suspected that some ASA firms might have become private at the announcement of the mandatory quota. However, none of the firms that dropped out between 2006 and 2010 have become private. Some of them have gone bankrupt, and others switched to different stock indices. In Tables 5 and 6, some observable characteristics of the Norwegian and Danish firms are given. The firms that dropped out do not show striking differences. This suggests that the sample is a sufficient representation of the whole index. Therefore, selection bias does not seem like the most important driver of the insignificant results.
5.4 Time Frame
As elaborated upon in the subsection on the results, the time frame may have been problematic for the significance of the results. The financial crisis took place in the middle of the time frame,
strongly influencing the data and mainly the financial performance measures. It was assumed that the difference-in-differences method would correct for the effects of the financial crisis, but the non-parametric suggest that firms in Denmark responded differently to the crisis than firms in Norway. This may even be to an extent that the test is no longer valid. Additional analyses should be performed to find the exact impact of the crisis on the results presented in this paper.
Apart from the financial crisis, the time frame might have been problematic for another reason as well. Namely, the first gender quota had already been introduced in 1998 for government boards, and in 2003 for public limited liability companies. Although these quotas were not yet mandatory, it is possible that firms were already taking this into account. This might have
influenced the power of the test and hence the significance of the results. In further research, it will be valuable to use a time frame that starts before the introduction of the voluntary quota,
following research done by Bertrand et al (2017), who used a time frame that started in 1998. Also, it would have been beneficial to look at more than two years. Since it takes time to make an impact as a board member, the advice for further research is to also look at the impact on firm performance one or two years after the installation of new female board members.
5.5 Industry
Apart from firm size, there were no firm-specific characteristics accounted for, such as industry. An additional analysis of the industries that the different firms were in is performed to find out if there is a significant change between 2006 and 2010. Firms that were dropped out of the sample for not remaining active in 2010 are also taken into account, to see if the sample became too narrow. It is found that by dropping the firms out of the sample, there was no drastic change in the different industries. For further research, it may be interesting to use a dummy variable for each industry and find out if this impact is significant. This can only be done if the sample size is increased, because in the current sample some industries only consisted of one or two firms, which inherently results in insignificant results.
6. CONCLUSION
Driven by recent discussions on female representation in leadership roles, this paper examined the impact of legislative quotas for women on firm performance. A secondary goal was to investigate the impact of gender diversity on firm performance. To answer these questions, first, a review of existing literature on the origin, advantages and disadvantages of gender quotas, and on gender diversity was carried out. This information formed the basis for the main hypotheses. The hypotheses were as follows: (1) There is a significant positive relationship between gender quotas and firm performance, and (2) There is a significant positive relationship between gender diversity and firm financial performance. To answer the research questions, a set of tests were performed, including OLS regressions and non-parametric tests. Information on board composition and firm-specific data from Norwegian and Danish firms was used. The sample consisted of the yearly data of 16 Norwegian and 13 Danish firms in 2006 and 2010.
The results do not provide significant evidence to prove that gender quotas have any impact on firm performance. These findings support the conclusion of a number of other studies that did not find a significant link either. Regarding the second research question, it was found that in Norway, the percentage of female board members has a positive impact on firm
performance. According to the second regression analysis, the optimal percentage of female board members is 32.39%, which implies some level of diversity. For Denmark, the results were
7. REFERENCES
Adams, R.B. & Ferreira, D. (2009). Women in the Boardroom and their Impact on Governance and Performance. Journal of Financial Economics, 94(2), 291-309.
Ahern, K.R. & Dittmar, A.K. (2012). The Changing of the Boards: The Impact on Firm Valuation of Mandated Female Board Representation. The Quarterly Journal of Economics, 127(1), 137-197. Bertrand, M., Black, S.E., Jensen, S. & Lleras-Muney, A. (2017). Breaking the Glass Ceiling? The
Effect of Board Quotas on Female Labor Market Outcomes in Norway. National Bureau of Economic Research Working Paper.
Bøhren, O. & Staubo, S. (2014). Does Mandatory Gender Balance Work? Changing Organizational Form to Avoid Board Upheaval. Journal of Corporate Finance, 28, 152-168.
Campbell, K. & Mínguez-Vera, A. (2008). Gender Diversity in the Boardroom and Firm Financial Performance. Journal of Business Ethics, 83(3), 435-451.
Carter, D., Simkins, B. & Simpson, W. (2003). Corporate Governance, Board Diversity, and Firm Value. Financial Review, 38(1), 33-53.
Comi, S., Grasseni, M., Origo, F. & Pagani, L. (2016). Where Women Make the Difference. The Effects of Corporate Board Gender-Quotas on Firms’ Performance Across Europe. Unpublished mimeo.
European Union, 2016. Gender Balance on Corporate Boards. Europe Is Cracking the Glass Ceiling. Directorate-General for Justice and Consumers, Fact sheet, July 2016.
Ferrari, G., Ferraro, V., Profeta, P. & Pronzato, C. (2016). Gender Quotas: Challenging the Boards, Performance, and the Stock Market. CESifo Working Paper.
Harjoto, M., Laksmana, I. & Lee, R. (2015). Board Diversity and Corporate Social Responsibility. Journal of Business Ethics, 132(4), 641-660.
Hoogendoorn, S., Oosterbeek, H. & van Praag, M. (2013). The Impact of Gender Diversity on the Performance of Business Teams: Evidence from a Field Experiment. Management Science, 59(7), 1514-1528.
Kogut, B., Colomer, J. & Belinky, M. (2014). Structural Equality at the Top of the Corporation: Mandated Quotas for Women Directors. Strategic Management Journal, 35(6), 891-902. Lehmann, J. (2013). Job Assignment and Promotion Under Statistical Discrimination: Evidence
Matsa, D., & Miller, A. (2013). A Female Style in Corporate Leadership Evidence from Quotas. American Economic Journal: Applied Economics, 5(3), 136-169.
Reguera-Alvarado, N., Fuentes, P. & Laffarga, J. (2017). Does Board Gender Diversity Influence Financial Performance? Evidence from Spain. Journal of Business Ethics, 141(2), 337-350. Seierstad, C., Gabaldon, P. & Mensi-Klarbach, H. (2017). Gender Diversity in the Boardroom. Volume 2:
Multiple Approaches Beyond Quotas. London: Palgrave Macmillan.
8. APPENDIX
Table 4: List of companies used
Norway Denmark
1. Aker Solutions 1. AP Moller
2. DNB 2. Carlsberg
3. DNO 3. Danisco
4. Fred Olsen Energy 4. Danske Bank
5. Frontline 5. DSV
6. Norsk Hydro 6. Lundbeck
7. Orkla 7. Nordea Bank
8. Petroleum Geo Services 8. Novo Nordisk
9. Prosafe 9. Novozymes
10. Royal Caribbean Cruises 10. Topdanmark
11. Schibsted 11. Tryg
12. Statoil 12. Vestas Windsystems
13. Storebrand 13. William Demant Holding
14. Telenor 15. TGS
Table 5: Observable firm characteristics Norway
Firm Tobin’s Q
2006 Tobin’s Q 2010 Total assets 2006 Total assets 2010 Industry Aker Solutions 2.13 1.42 30844000 39534000 Manufacturing DNB 1.04 1.01 1320204000 1860705000 Finance, insurance, and real estate DNO 4.15 2.21 2980300 5392700 Mining Fast Search
and Transfer 1.94 - 2760482 - Services
Fred Olsen
Energy 2.86 1.75 8219482 13486915 Mining
Frontline 1.43 1.33 4580757 3797920 Transportation,
communications, electric, gas, and sanitary services Golden Ocean
Group - 1.05 - 1227218
Transportation, communications, electric, gas, and sanitary services Marine
Harvest - 1.41 - 23410200 Manufacturing
Norsk Hydro 1.61 1.13 232754000 87107000 Manufacturing
Norske Skog 1.05 - 45014000 - Manufacturing
Norwegian
Property - 1.01 - 15806813
Finance,
insurance & real estate
Orkla 1.32 1.13 79411000 86918000 Manufacturing
Petroleum
Geo Services 4.29 1.60 7181884 16253583 Services
Prosafe 2.01 2.08 13392626 7375653 Mining
Rec Silicon - 0.88 - 36529000 Manufacturing
Royal Caribbean Cruises
1.20 1.11 13393088 19694904 Transportation,
communications, electric, gas, and sanitary services
Schibsted 1.63 1.69 16260000 16407000 Manufacturing
Seadrill - 1.56 - 17396700 Mining
Sevan Marine - 0.91 - 14346711 Construction
Songa Offshore - 0.90 - 8762199 Mining Statoil 1.74 1.35 315093000 641130000 Manufacturing Stolt-Nielsen 1.25 - 1892713 - Transportation, communications, electric, gas, and sanitary services
Storebrand 1.05 1.00 220751300 390097000 Finance,
real estate Tandberg 4.78 - 2332897 - Manufacturing Tandberg Television 1.91 - 3392799 - Transportation, communications, electric, gas, and sanitary services
Telenor 1.98 1.41 146528000 170725000 Transportation,
communications, electric, gas, and sanitary services Tomra
Systems 2.60 - 3247100 -
Transportation, communications, electric, gas, and sanitary services
TGS 3.55 2.16 4180460 7017260 Services
Yara
International 1.84 1.98 32088000 63814000 Manufacturing
Note: the firms that are indicated in red were only active in one of the two investigated years, and were therefore not included in the sample.
Table 6: Observable firm characteristics Denmark
AP Moller 1.30 1.11 309635000 369589000 Transportation,
communications, electric, gas, and sanitary services Carlsberg 1.45 1.15 57629000 142931000 Manufacturing Coloplast 3.42 - 7854000 - Manufacturing Dampskibssel skabet Norden - 0.79 - 12539590 Transportation, communications, electric, gas, and sanitary services
Danisco 1.40 1.58 32020000 20375000 Manufacturing
Danske Bank 1.03 1.00 2738921000 3212193000 Finance,
insurance, and real estate
DSV 2.07 1.84 15800000 22636000 Transportation,
communications, electric, gas, and sanitary services FLSmidth & Co - 1.95 - 21871000 Manufacturing G4S 1.48 - 3055900 - Services Genmab - 1.75 - 2468336 Services GN Store Nord 2.50 - 8031000 - Manufacturing Jyske Bank - 1.01 - 244112717 Finance,
insurance & real estate
Lundbeck 3.26 1.54 11351900 17892000 Manufacturing
NKT - 1.24 - 12291100 Manufacturing
Nordea Bank 1.04 1.01 3129556304 5218848609 Finance,
insurance, and real estate
Novo Nordisk 3.78 6.42 42781000 59555000 Manufacturing
Novozymes 4.38 4.28 7920000 12522000 Manufacturing
Santa Fe
Group 2.12 - 2748000 - Services
Sydbank - 1.01 - 150832000
Finance,
insurance & real estate
TDC 1.49 - 80235000 -
Transportation, communications, electric, gas, and sanitary services Topdanmark 1.30 1.11 42868000 56755000 Finance, insurance, and real estate Torm 1.47 - 11808929 - Transportation, communications, electric, gas, and
sanitary services Tryg 1.47 1.15 41222000 48899000 Finance, insurance, and real estate Vestas Windsystems 2.34 1.30 26045132 51035367 Manufacturing William Demant Holding 10.11 4.31 3021843 6520200 Manufacturing
Note: the firms that are indicated in red were only active in one of the two investigated years, and were therefore not included in the sample.
Table 7: Descriptive statistics
Norway 2006 Variable Number of
observations Mean Standard deviation Min Max
Tobin’s Q 16 2.11 1.048096 1.040862 4.285167 Total number of board members 16 7.50 2.556039 5 11 Number of female board members 16 2.44 1.314978 0 5 % female 16 .33 .1530665 0 .5 LnSize 16 17.32 1.815834 14.90753 21.00105 Norway 2010 Tobin’s Q 16 1.52 .4136068 1.003144 2.206465 Total number of board members 16 8.25 2.32379 5 11 Number of female board members 16 3.06 1.181454 1 5 % female 16 .37 .1098325 .1111111 .5 LnSize 16 17.57 1.822759 15.14996 21.34422 Denmark 2006 Tobin’s Q 13 2.69 2.484701 1.027979 10.11491 Total number of board members 13 10.23 2.891189 5 15 Number of female board members 13 1.77 1.640825 0 5 % female 13 .16 .109569 0 .3333333 LnSize 13 17.82 2.079114 14.92138 21.86416 Denmark 2010 Tobin’s Q 13 2.14 1.726213 .998194 6.418402 Total number of board members 13 10.62 2.467741 7 15 Number of female board members 13 2.00 1.354006 0 5 % female 13 .18 .0967382 0 .3333333 LnSize 13 18.19 2.038212 15.69042 22.37554
Table 8: OLS Regression Model without Royal Caribbean Cruises The impact of gender quotas on firm financial performance
TOBINQ ROA (1a) (1b) (2a) (2b) Constant 10.23936*** (5.90) 2.687033*** (6.18) 40.39961*** (4.63) 8.634615*** (4.13) NORWAY -.696605 (-1.36) -.5118712 (-.86) 2.73573 (1.07) 3.512718 (1.23) 2010 -.39233 (-0.74) -.5488206 (-.89) .1443515 (.05) -.5138462 (-.17) NORWAY2010 -.1347901 (-.19) -.0762465 (-.09) -4.374388 (-1.21) -4.128154 (-1.02) LNSIZE -.4237769*** (-4.46) -1.782401*** (-3.73) Number of observations 56 56 56 56 R2 0.3282 0.0665 0.2620 0.0610
Notes: t-values are given in parentheses below the estimated coefficients, *=significant at α=0.10, **=significant at α=0.05, ***=significant at α=0.01
Variables: TOBINQ (approximation of Tobin’s Q), ROA (return on assets, %), NORWAY (dummy variable for Norway), 2010 (dummy variable for 2010), NORWAY2010 (interaction between NORWAY and 2010), LNSIZE (natural logarithm of firms’ total assets), R2 (coefficient of determination: indicator of the
Table 9: P-values for non-parametric tests (1) Norway 2010 = 2006 (2) Denmark 2010 = 2006 (3) Norway = Denmark Tobin’s Q .0052 .0392 .7257 ROA .0131 .8613 .0226
Notes: In column (1), the results of the Wilcoxon matched-pairs signed-rank test are given, testing whether the financial performance of firms in Norway differed significantly between 2010 and 2006. In column (2), the results of the Wilcoxon matched-pairs signed-rank test are given, testing whether the financial performance of firms in Denmark differed significantly between 2010 and 2006. In column (3), the results of the Wilcoxon rank-sum test are given, testing whether the difference in firm financial performance was significantly different for Norway than for Denmark.
