Effect of a mandated boardroom gender quota on firm performance : evidence from Norway

(1)

26 June 2018

University of Amsterdam

Effect of a mandated boardroom gender

quota on firm performance

Evidence from Norway

Abstract

This paper examines the effect of a mandated gender quota in the boardroom on firm

performance. Norway was the first country who introduced such a quota. As of January 2008 all public limited liability companies (ASA) had to have 40 percent of each sex represented in their board. I compare ASA companies with limited liability companies (AS), who did not have to comply. A negative relationship is found between the introduction of the quota and firm performance, measured by both ROA and ROE. Results are based on data from 96 ASA companies and 115 AS companies for the years 2005 – 2010.

(2)

NATHALIE BORST 2018 2

Introduction

Diversity, or specifically gender diversity has gotten a lot of interest from not only academia, but also the popular press. A lot of research has already been done about how gender

diversity in the board can affect firm value. These researches have mixed findings on the effect of board female representation. Nowadays, a lot of boards are still male-dominated. Women face the so called ‘glass ceiling’, while men benefit from the ‘glass escalator’ in top management positions (Ryan & Haslam, 2005). It is interesting to look into the effect of forced gender diversity in the form of a gender quota. In this case not only the effect of a more gender diverse board could influence the performance of a firm, but also the effect of conforming to a law within a few years could have a significant impact. Results would be driven by the mandate of the quota and not solely by the effect of gender diversity.

In the past decade the introduction of laws on gender diversity has become a frequent topic on the political agenda of countries. The European Commission for example is involved in targets regarding gender diversity. They first put the issue of women in leadership

positions on the agenda in 2010. At that time, on average 23.3 percent of the board members of the European largest publicly listed companies were women. In 2011 the European Commission called for companies to self-regulate and introduced internal quotas. One year later, in November 2012, the Commission proposed a law aiming to speed up the progress towards a more balanced representation of women and men on boards of listed companies (Jourová, 2016). Norway however was ahead of regulations improving gender diversity. It was the first country to introduce a quota for women on company boards. By 2008 at least 40 percent of each gender had to be represented in the boards of public limited companies (Public Limited Liability Companies Act, Chapter 6, Section 11a). Such quotas have been discussed in several European countries, but discharged by governments.

Studying this quota is of interest, because the findings of this study could maybe guide other countries who are considering, or have considered, introducing a gender quota. Furthermore, a lot of literature is available about gender diversity in boards, but few studies exist about this introduction of a mandated quota.

The introduction of the quota

Since the introduction of the quota in 2008, the number of women on boards has reached 40 percent as required by law. The Gender Equality Act came initially into effect in 2003 and applied to public limited companies (Allmennaksjeselskap or ASA), cooperative companies and state and municipality owned companies. Before the government passed this amendment merely 6 percent of the board members of ASA’s were women (Storvik & Teigen, 2010). This Act has the particular objective of improving the position of women since men were overrepresented in boards of Norwegian firms (Gender Equality Act, Chapter 1, Section 1). Before the government passed the amendement in 2003 merely 6 percent of the board

members of ASA’s were women (Storvik & Teigen, 2010). It was required for the companies that had to comply to have a minimum of 40 percent of each gender represented in the board

(4)

NATHALIE BORST 2018 4 by the end of 2006. Initially the quota was not obligatory and no sanctions were imposed when companies did not meet the target. Expectations were companies would follow the quota of oneself after the passing of the law in 2003. In reality this was not the case. By July 2005 only 12 percent of the boards was female (Teigen, 2008). Because of the disappointing numbers the new section 6-11a of the Norwegian Public Limited Liability Companies Act entered into force on 1 January 2006, giving firms two years to comply. The exact

requirements are as follows:

Requirement regarding the representation of both sexes on the board of directors (1) On the board of directors of public limited liability companies, both sexes shall be represented in the following manner:

1. If the board of directors has two or three members, both sexes shall be represented. 2. If the board of directors has four or five members, each sex shall be represented by at least two members.

3. If the board of directors has six to eight members, each sex shall be represented by at least three members.

4. If the board of directors has nine members, each sex shall be represented by at least four members, and if the board of directors has more members, each sex shall

represent at least 40 percent of the members of the board. (Public Limited Liability Companies Act, Chapter 6, Section 11a).

This rule applies to all Norwegian public limited liability companies registered in the Norwegian Register of Business Enterprises (Brønnøysundregistrene). Since 1 January 2008 Norwegian public companies are obliged to have the required number of women in their board. Non-compliant companies will be forced dissolution unless the Ministry of Trade and Industry decides the company has a substantial public interest. In this case the company has to pay a fine until it is complied with the regulation (Storvik & Teigen, 2010). This also applies to newly established firms. If they fail to have a board in accordance with the law, then the Norwegian Business Register will refuse to register a company due to the board not meeting the statutory requirements (Teigen, 2008).

The implementation of the quota was an effort to change the low number of women on boards and to speed up the moderate increase in females on corporate boards (Sweigart, 2012). It is an attempt to increase the influence of women. The Norwegian Government (2008) argued that to remain competitive, companies should draw on roughly the same number of men and women (Seierstad & Opsahl, 2011). In Norway women are well-represented in politics and the women’s employment rate is almost the same as men.

However, female representation in boards remained a problem (Tyson, Zahidi & Hausmann, 2008), hence the introduction of the quota with the goal to distribute power in the boardroom more equally. The parliaments goal to have more women in the boardroom was eventually

(5)

NATHALIE BORST 2018 5 reached in April 2008, but there has been an heated political and public debate before the decision of the parliament had been made.

Supporters of the law claimed redistribution of resources by positive discrimination was necessary in order to achieve gender equality (Storvik & Teigen, 2010). They considered the male domination in boardrooms as a form of unfair gender discrimination. Opponents found that recruitment to corporate boards should not be based on gender, but on quality. Introducing the quota would therefore result in less competent women replacing more competent men. According to them, not enough women had the relevant experience (Storvik & Teigen, 2010). Furthermore, the quota has been democratically criticized, because of the hindering of “owners’” democratic right to self-regulate and choose their board candidates by their selves without a law restricting them. The opponents were mainly business leaders and Norwegian employers’ organizations like NHO.

This paper focusses primarily on the economic consequences of the quota. I compare the ASA companies with the privately owned limited liability companies (Aksjeselskap or AS). Private limited companies have less strict rules regarding board composition and share capital in comparison to public limited companies (Storvik & Teigen, 2010). Above all, these AS companies are not subjected to the gender quota law, which makes them a good control group for a difference-in-difference analysis (DiD). AS firms will be compared to the ASA firms in the period before and after the introduction of the quota. I show that the introduction of the boardroom gender quota has a negative impact on firm performance when examining the 3 years pre- and 3 years post-quota. I find robust evidence that the relationship with the compliable firms in the post-reform period and their return on assets (ROA) and return on equity (ROE) is negative. Despite the drastic increase in female representation board competence deteriorated as a consequence.

Empirical findings

Board and firm performance

Norway distinguishes itself because of the use of a one-tier board system. This in contrast with most European countries who operate with a two-tiered board system, consisting of a supervisory and a management board (Sweigart, 2012). In the Norwegian “one-tier” system supervising is an additional task of the board. Although there is a difference in tasks,

superiority of one system over the other has not been proven (Jungmann, 2006). Furthermore, Norwegian boards comprise the owner’s and employees’ representatives. In joint stock companies with over 200 employees, one-third of the board members has to be elected by the employees while two-thirds consists of shareholder-elected members. This is ensured in the Limited Liability Act (Public Limited Liability Companies Act, Chapter 6, Section 4). In smaller sized companies with over 50 employees, at least two members will be elected by the employees (Dale-Olsen, Schøne & Verner, 2013).

(6)

NATHALIE BORST 2018 6 The board of directors is a substantial element of the corporate governance structure. The board monitors the management team and provides them with advice and guidance (Pande & Ford, 2012). Additionally they have a central role in establishing the business strategy, setting policy objectives and plan and manage resources in co-operation with the management team (Nekhili & Gatfaoui, 2013). These monitoring and decision-making roles of the board are therefore important corporate governance tools. Gender diversity can affect the quality of these tasks and thus affect financial performance. It is interesting to look into this effect since Williams and O’Reilly (1998) found evidence that diversity in visible characteristics, like gender, have a larger impact than non-visible attributes (education, experience). The effectiveness of a board depends on two group-level principles according to Forbes and Milliken (1999): task performance and the ability of working together as a group.

Wood (1987) found women where better in tasks where the quality of the solution is the dependent measure, whereas men were better at tasks where quantity was the dependent measure. This shows the way a task is measured matters and influences

performance. Also the type of task can cause difference in behaviour between men and women. Men tend to display more verbal and non-verbal power on ‘masculine’ tasks, whereas women exhibit more power for most of the verbal and non-verbal measures when assigned to do ‘feminine’ tasks (Dovidio et al., 1988). When women are told the task is particularly difficult for women they actually perform worse. Spencer et al. (1999) showed women performed as good as men on math tests when they were being told there is no difference in difficulty between genders. However, women performed worse when they thought the test was more difficult for them specifically. When looking at the total group performance, Wegge et al. (2008) found a negative relationship between gender diversity and group performance when examining data from 4538 federal tax employees in the United States working in 222 natural workgroup units. This in contrast to the findings of Bowers, Pharmer and Salas (2000) who found no significant difference between homogeneous and heterogeneous groups with respect to gender on task performance.

As far as the relationship between gender diversity and group cohesion, Rogelberg and Rumery (1996) studied five groups with different types of gender compositions (all-male, lone-female, balanced-gender, lone-male, and all-female), but found no significant

differences in interpersonal cohesion within the teams with different gender ratios. This is in line with the results of Webber and Donahue (1999) and Lee and Farh (2004). They also concluded gender diversity is unrelated to group cohesion. Based on these findings group cohesion should not be influenced by variants of gender ratios within groups.

Gender diversity and firm performance

The overall effect of gender diversity on group and firm performance has also been widely studied. Existing literature on this topic has yielded inconsistent results. A widely cited report from the Catalyst (2004) analysed data from over 500 firms in the United States from 2001 to 2004. They found companies with a higher female representation showed a significantly higher return on investment (ROI) and return on equity (ROE) than those with a

(7)

NATHALIE BORST 2018 7 lower share of women in the board. Campbell and Mínguez-Vera (2008) also found positive results. Gender diversity in boards, measured by both the Blau and Shannon indices and the percentage of women on a board, has a positive effect on firm value. Firm value is measured using an approximation of Tobin’s Q. Surprisingly, they found that the presence of one or more women on the board had no effect. Campbell and Mínguez-Vera used panel data analysis on Spanish companies which gives a different view on the topic since a lot of literature is based on U.S. data. Admittedly, multiple studies found a positive relationship between gender diversity in the board and firm performance. Yet, if women are scarce commodities, they might have the opportunity to choose to work for better performing firms and self-select these companies. Shrader et al. (2003) suggested better performance induces diversity rather than vice versa after they found a positive effect of gender diversity on the ROA and ROI. Farrel and Hersch (2005) also concluded firms with a high ROA are more likely to acquire female directors. Adams and Ferreira (2009) studied the effect of gender diversity on firm performance measured by Tobin’s Q. Information from S&P firms from 1996 till 2003 was used to conclude that a gender-diverse board has a negative effect on firm performance. They suggest gender-diverse boards tend to over-monitor. Dobbin and Jung’s (2010) findings correspond with this. They also looked at data from the United States. Based on information of America’s 500 largest companies, they concluded that an increase in gender diversity has no effect on profits, but a negative effect on stock price.

Quota and firm performance

According to Barney (1997) employee and management capabilities are the key resources to a competitive advantage. These resources are sustainable and are hard for competitors to imitate. A high quality board is thus one of the key factors to outperform the competitors. Also Rosener (1995) states human resource management is the major

determinant of global competitiveness. She argues that companies who are fully utilizing the diverse talents of female managers gain a competitive advantage over the firms who do not. Top female managers have a big impact on productivity, morale and profits according to her, because of their qualities of seeing the big picture issues. Norwegian women may have underinvested in their individual human capital if they believed no opportunities were available to them. The introduction of the quota gives an opportunity for women and can signal that a leadership position is possible. It can encourage women to invest more in

themselves as a valuable resource which eventually can improve the board quality. Women in current leadership positions could in this way inspire other women. They can act as a role model for others. Being a female in a board used to be not that feasible, because boards used to be male-dominated. The mandated quota caused a direct increase in female board

representation, which can serve as an inspiration for other women. On the other hand, could this also lead to an opposite effect where the quota can reduce a woman’s investment incentives if she thinks the way to the top has been made easier because of the quota (Pande & Ford, 2012). Coate and Loury (1993) formulate this as a “patronizing equilibrium” where it is easier for women to be assigned to a good job and harder for men because of the quota.

(8)

NATHALIE BORST 2018 8 Consequently resulting in women investing less in their career since it does not require much to get a board position.

All in all the effect of the quota is visible in Norway. The amount of women on

companies’ boards has drastically risen. However the total female representation not as much. The phenomenon “golden skirts” has emerged. The share of women serving on multiple boards has risen since the introduction of the quota. This led to a smaller than predicted increase in the total amount of women on corporate boards. Seierstad and Opsahl (2011) found in their sample that the number of directors, serving on multiple boards, rose from 91 to 224 after the introduction of the quota. More specifically, in 2002 at the beginning of their observation period only 7 of the 91 directors holding multiple board positions were women. In 2009, after the quota entered force, 107 of the 224 were women. This is 47.8 percent. When looking into directors with at least 3 directorships, the share of women was at 61.4 percent. The exogenous shock of the mandate caused by the quota could have a led to a supply shortage of qualified women, hence the increase of the share of women holding multiple board positions. This overrepresentation of female directors in multiple boards suggests there are not enough women with governance experience. If the lack of overall female representation in boardrooms is due to the shortage of suitable candidates, then mandated quotas will have a negative impact on the quality of the board. Women have to be appointed because of their gender and not necessarily because of their quality (Shrader et al., 1997). However, the multiple board membership could also be due to the fact recruiters are unwilling to hire less experienced women or women outside their existing network (Pande & Ford, 2011).

Another possible negative effect of the quota could be that once firms satisfy the minimal requirement, the demand for women will vanish. Recent data from the European Institute for Gender Equality (EIGE) supports this. From 2008, the percentage of women on the boards of Norwegians largest listed companies has been close to 40. Female board representation is on average 40.1 percent for the years 2008 till 2017. Seierstad and Opsahl (2011) also address this. According to them this signifies companies simply comply with the law and are not moving towards a more gender equal setting in the boardroom. That being the case, more women have acquired board member positions as a result of the quota. However, more women on boards does not likewise lead to more female influence. According to Adams and Ferreira (2009) female board members are more likely to sit on monitoring-related committees and less likely on the compensation committees. They tend to occupy supporting management roles such as personnel or marketing within an organization, rather than operating or commercial roles. Independent directorship and committee membership are considered as senior positions and are obtained by means of exceptional expertise, skills and handling responsibilities (Vinnicombe, 2000). Wearing and Wearing (2004) concluded women in non-executive director positions are at disadvantage in getting promoted to positions as chair of important sub-committees or as chairman of the board itself. Their findings are based on data from UK companies. They formulate this as the ‘second glass ceiling’ where women get excluded from powerful positions once they have a board position. This could be another possible drawback from the quota. Bertrand et al. (2017) surmised

(9)

NATHALIE BORST 2018 9 companies strategically hiring sub-par women, expecting minimal participation from them when board decisions have to be made. Leading to women getting appointed to board positions, but in reality having no executive power.

Standard economic theory states firms maximize their profits. This indicates boards were already at a point where the boards were optimally constructed before the reform. Boards would have been more gender diverse if there were more adequate women. A potential lack of women in boards could have been because a lot of women do not aspire a position in the boardroom. This could disrupt the work and personal life balance, the high work pressure or this could be driven by another internal motivation. The constraint of more female representation in boardrooms would therefore lead to a reduction in the companies’ profits (Demsetz & Lehn, 1985). If no available candidates are suitable for the position then this will lead to a deterioration of quality in the board, due to hiring less adequate people for board positions. Welch (1976) who examined employment quotas concluded that an

affirmative action policy may result in unskilled people getting assigned to skilled jobs. As a consequence of the “patronizing equilibrium” (Coate and Loury, 1993) employers have to lower their standards in order to hire the minority workers, which are the newly appointed women after the intervention of the Norwegian government.

Though multiple studies found a positive effect of gender diversity on team and firm performance, a mandated quota could affect those benefits of diversity because of the

possible appearance of tokenism. Meaning women on corporate boards get judged and treated differently, because they got their position due to legislation and not skills. Kanter (1977) argued being a token leads to three behavioural consequences. Visibility, polarization and assimilation are the three perceptual phenomena associated with tokenism. Visibility could force performance pressure, because the minority has the impression being watched all the time. Women feel they have to work harder to prove themselves. Working harder should not affect the quality and efficiency of a board. However, performance pressure could negatively affect actual output. Also the fear of out-perform their male colleagues might make them feel uncomfortable. Polarization occurs when men, who were formerly the dominant group, feel threatened by women. Those men tend to exaggerate the difference between the two genders, which makes it hard for the tokens to integrate. The women are also likely to get excluded from the informal social networks of the men. Assimilation leads to stereotyping, where women get forced into stereotypical categories (Elstad & Ladegard, 2012). This leads to the women being treated based on their stereotype rather than the individual. Kanter (1997) considered a minority as a group that constitutes less than 15 percent of the total employee population of the workplace. Still, women have always been the minority and if men continue to hold onto the negative views then the stereotyping still remains or even gets worse by the affirmative action. This tokenism could distort the relationships among board members leading to a less efficient board.

Conforming to the quota can be difficult for some companies. Ahern and Dittmar (2010) show evidence of companies avoiding the quota. After passing the quota law, more companies chose to change their legal status. A lot of public firms converted to private firms. In 2002-2003 and 2007-2008 the conversion rates were around 6 and 8 percent, but in the

(10)

NATHALIE BORST 2018 10 years leading up to the mandate of the quota conversion rates increased. In 2005 and 2006 rates were at 11 and 16 percent and the conversion rate was 25 percent in 2007 for companies who had to adjust. Firms that already satisfied the quota had a conversion rate of 6 percent (Dale-Olsen, Schøne & Verner, 2013). Furthermore, ASA firms with a smaller share of women in their board in the pre-quota period were less likely to delist in the years 2003-2009. Bøhren and Staubo revised this question in 2014 and found consistent results.

Hypothesis

Based on all revised literature mentioned in this thesis, expectations are that the introduction of the quota has a negative impact. ASA firms will perform worse than AS firms over the observation period since those companies already had to comply in the years leading up to the quota, which could have already affected their ROA and ROE negatively. Furthermore, predictions are firms perform worse in the post-reform period compared to the pre-reform period. However, this expected effect is not caused by the introduction of the quota, but because of the start of the financial crisis in 2008. The beginning of this crisis is in line with the start of the post-reform period.

The hypothesis is as follows:

A mandated boardroom gender quota will negatively affect firm performance.

The coefficient used to measure the effect of the mandate will therefore be significantly negative according to my expectations, taking into account control variables that will be added.

Data and variables

Sample selection

The implementation of the reform in Norway provides a “natural experiment”. The data for the AS companies is mainly selected based on data availability with number of employees as starting point, because the number of employees is a good indication of company size. Information about those companies is limited, because the private limited companies are less restricted when it comes to reporting. The company data was obtained from Bureau van Dijk’s Orbis historical database. Database snapshots from years 2005 till 2010 were used. All companies who did not have available data for those years were excluded from the selection. This resulted in 115 AS companies

The same search strategy is used for all ASA companies. The Orbis historical database was consulted and all ASA companies with a known ROA and number of

employees were included in the sample. During data collection I observed the same problem Ahern and Dittmar (2010) addressed. A lot of public companies were dissolved or converted

(11)

NATHALIE BORST 2018 11 to an AS in the transition years prior to the introduction of the quota. Those companies were eliminated from the sample as well. To reduce major differences in number of employees, all ASA companies with less than 3 employees were excluded. This left a remaining sample of 96 ASA companies. For both groups most of the board members’ name and function are available. Board size is calculated manually based on that information. Characteristics like gender and age were not available for all years of the observation period. A total of 211 companies with data from 6 years is used in the analysis.

For six companies, data about their return on equity was not available for all years of the observation period. These companies are removed from the dataset when the ROE is used as the dependent variable. This dataset consists of 205 companies.

Dependent variables

Economic performance can be measured in multiple ways. Two types of measures can be used according to Bhagat and Jefferis (2002): market-based and accounting-based measures. Most of the commonly used market measures is Tobin’s Q. Defined as the ratio of the market value of a firm to the replacement cost of its assets (Chung & Pruitt, 1994). Several

researches about the effect of gender diversity on firm performance also used Tobin’s Q as a measure of firm performance. Studies using market-based measures for firm performance could be subjected to investor bias. Acquiring women may influence stock performance because of the investors’ reaction to this appointment. Institutional investors may react either negatively or positively to the announcement of the appointment of a female board member. I will use accounting-based measures, considering this research uses financial data from both public and private companies and market data will not be available for private companies. As a measure of firm performance I will therefore use the return on assets (ROA) and return on equity (ROE) for company i at time t. The values used to calculate these ratios are in U.S. dollars. A higher ROA indicates a higher asset efficiency. The ROE focusses on the return to the shareholders. A higher ROE implies a higher profitability on the shareholders’

investments. This thesis will focus on the ROA as the main measure of firm performance. In my dataset, the mean ROA for ASA companies is 5.627 (21.766) and 7.1 (10.851) for AS companies observable from Table 1. The average ROE is 29.462 (119.628) for ASA companies and 30.627 (96.556) for AS companies (Table 2). This data shows AS companies have a higher average ROA and ROE during the observation period.

Independent variables.

To compare the two types of legal company forms for both periods, a difference-in-difference analysis (DiD) is used. The DiD approach is a research design for estimating causal effects (Lechner, 2010). This method eliminates possible fixed effects. For example, the year of the introduction of the quota, in 2008, was also the year the financial crisis broke out. Firm

(12)

NATHALIE BORST 2018 12 performance was most likely affected by this crisis. Since both private limited as public limited companies were affected by the impact of the crisis, this effect will cancel out.

There are four groups for the DiD approach: treatment group in the pre-reform period, control group in the pre-reform period, treatment group in the post-reform period and control group in the post-reform period. The treatment group is the group of all ASA companies and thus, the group who had to comply. The control group is the group of all AS firms who are not subjected to the quota. Three OLS-regressions are done with ROA as dependent variable. The independent variables will be examined in each regression. In the first regression the dependent variable asa will be used. This regression is as follows:

(1) 𝑟𝑜𝑎𝑖𝑡 = 𝛽0 + 𝛽1∙ 𝑎𝑠𝑎𝑖 + 𝛽2∙ 𝑋′𝑖𝑡 + 𝜀𝑖𝑡

In this first analysis the company groups will be compared for the years 2005-2010. The asa variable is a dummy variable and gets the value of 1 if the firm is an ASA firm and 0 if the company is an AS firm. The β1 coefficient will estimate if firms who are legally established as a public limited liability company (ASA) have on average a lower or higher ROA than limited liability companies (AS). A negative coefficient indicates firms constructed as ASA perform on average worse than AS companies. X’ is a vector for all control variables in the regression which will be defined in the next section.

For the second regression both periods will be compared. The dependent variable post is a dummy variable which gets the value of 1 if the observation is in the post quota

introduction period and a value of 0 if the observation is in the pre-quota period. This regression is as follows:

(2) 𝑟𝑜𝑎𝑖𝑡 = 𝛽0 + 𝛽1∙ 𝑝𝑜𝑠𝑡𝑡 + 𝛽2∙ 𝑋′𝑖𝑡 + 𝜀𝑖𝑡

The years 2005-2007 count as the pre-quota period and the years 2008-2010 as the post-quota period. The β1 estimator signifies the difference in ROA for all firms between the two

periods. This coefficient will most likely be negative since the first year of the second period is the start of the financial crisis as well. The control variables (X’) will be explained in the next section.

The first two regressions break down the final regression, which will be the actual DiD-analysis to compare both groups for both periods. This regression is as follows:

(3) 𝑟𝑜𝑎𝑖𝑡 = 𝛽0 + 𝛽1∙ 𝑎𝑠𝑎𝑖 + 𝛽2∙ 𝑝𝑜𝑠𝑡𝑡 + 𝛽3∙ 𝑑𝑖𝑑𝑖𝑡 + 𝛽4∙ 𝑋′𝑖𝑡 + 𝜀𝑖𝑡 Both dummies, asa and post, are added to the regression, which will be estimated by respectively β1 and β2. To estimate the effect of the mandated quota an interaction term is introduced: did. This interaction term is also a dummy variable and is the product of the asa and post dummy. Therefore, the did variable will get a value of 1 if the observation is an ASA company in the post-reform period. All other observations will get a value of 0. Expectations are β3 will be negative, which signifies the introduction of the quota having a negative effect on firm performance. The control variables which are collected in vector X’ are explained in the next section.

(13)

NATHALIE BORST 2018 13 The same three regressions will also be performed with the ROE as dependent

variable. These regressions are:

(4) 𝑟𝑜𝑒𝑖𝑡 = 𝛽0 + 𝛽1∙ 𝑎𝑠𝑎𝑖 + 𝛽2∙ 𝑋′𝑖𝑡 + 𝜀𝑖𝑡 (5) 𝑟𝑜𝑒𝑖𝑡 = 𝛽0 + 𝛽1∙ 𝑝𝑜𝑠𝑡𝑡 + 𝛽2∙ 𝑋′𝑖𝑡 + 𝜀𝑖𝑡

(6) 𝑟𝑜𝑒𝑖𝑡 = 𝛽0 + 𝛽1∙ 𝑎𝑠𝑎𝑖 + 𝛽2∙ 𝑝𝑜𝑠𝑡𝑡 + 𝛽3∙ 𝑑𝑖𝑑𝑖𝑡 + 𝛽4∙ 𝑋′𝑖𝑡 + 𝜀𝑖𝑡

The predictions will be the same as for the ROA-analyses, since the ROE is another measure of firm performance, whose values do not differ that much from the ROA.

Table 1 shows the mean and standard deviation of the ROA for the four groups. As observable from the table, the average ROA for ASA firms in the pre-reform period is 8.785 (21.36). The average ROA for ASA firms in the post-reform period is with a value of 2.47 (21.748) more than 3 times as small as the ROA in the pre-reform period. The standard deviation remained almost the same, but the average ROA has shrunken drastically in the post-quota period. The average ROA for the AS companies in the pre-quota period is 7.35 (10.519) and has decreased a little in the post-quota period to a value of 6.848 (11.183). This decline in the ROA is not nearly as much as the decline for the ASA companies.

The first regression uses asa as the independent variable. Table 1 shows that the average ROA for the treatment group is 5.627 (21.766) and for the control group 7.1

(10.851). These values are for both periods. As observable from the table is the average ROA for the AS companies a little bit higher. The values for the ROA are less dispersed for the AS firms.

The dummy variable post is used in the second regression. The mean ROA in the pre-reform period is 8.003 (16.368), which is almost double the value of the average ROA in the post-reform period: 4.856 (16.959) (Table 1). These averages are for all companies in the sample.

The average ROA in the post-reform period for the treatment group is 2.47 (21.748) (Table 1). Not surprisingly is this the lowest average value when comparing the asa and post dummies. The hypothesis is the mandate has a negative effect on firm performance, so a low ROA when post=1 and asa=1 is in line with the expectations. The average ROA for all companies for the total observation period is 6.43 (16.733), which is an overall positive value.

Table 1: Mean and standard deviation dependent variable (ROA)

Obs Pre-reform post=0 Obs Post-reform post=1 ASA asa=1 288 8.785 (21.36) 288 2.47 (21.748) 5.627 (21.766) AS asa=0 345 7.35 (10.519) 345 6.848 (11.183) 7.1 (10.851) 8.003 (16.368) 4.856 (16.959) 6.43 (16.733)

(14)

NATHALIE BORST 2018 14 The differences in ROE among the four groups are proportionally similar as for the ROA. As observable from Table 2 The ROE for the ASA companies decreases from 46.353 (116.579) to 12.572 (120.452). The ROE in het pre-quota period is almost 3.5 times as big as in the post-reform period. This ratio is the same for the ROA of the ASA companies. The ROE of the AS companies also slightly decreases from 35.923 (69.173) to 27.341 (72.46), but this deterioration is not as much as for the ASA companies (Table 2). The overall

difference in ROE between ASA and AS companies is not as much as the difference in ROA between those groups. The average ROE for ASA companies is 29.463 (119.628) and 31.632 (70.912) for AS firms. Furthermore, in the post-period is the average ROE approximately halved compared to the pre-period. The ROA shrunk from a value of 40.757 (94.219) to a value of 20.497 (97.87). This is again in line with the average ROA for those periods where the average ROA is also almost double the value in the pre-reform period compared to the post-reform period. The average ROE for all observations is 30.627 (96.556) as shown in Table 2.

Table 2: Mean and standard deviation dependent variable (ROE)

Obs Pre-reform _post=0 Obs Post-reform _post=1 ASA asa=1 285 46.353 (116.579) 285 12.572 (120.452) 29.463 (119.628) AS asa=0 330 35.923 (69.173) 330 27.341 (72.46) 31.632 (70.912) 40.757 (94.219) 20.497 (97.87) 30.627 (96.556) Control variables

To control for firm size the variable employees is used. Numerous papers use the natural logarithm for this variable to add company size to the regression. I do this as well to reduce the impact of outliers and also because the values are dispersed. The minimum measured employees for the ASA companies is 3 and the maximum is 41,428. These values are more scattered than those of the AS companies with a standard deviation of 6,625.504 compared to 1,049.656. For the AS firms the minimum employees is 8 and the maximum is 12,428. This with a mean of nearly 955 employees compared to a mean of 2,604.889 employees for ASA firms (Table 3; Table 4). Therefore, ASA companies are bigger on average.

Existing research states that board size could be an important factor regarding firm performance. Conyon and Peck (1998) as well as Eisenberg, Sundgren, and Wells (1998) found a negative relationship between board size and firm performance. The board

characteristic board size is therefore added as a control variable and measured by the number of members in the board. Besides the gender ratio in the board, board size could also be an influence on the ROA and ROE, hence this variable is added to the regression. The AS

(15)

NATHALIE BORST 2018 15 with a higher standard deviation, 3.212 and 2.655 respectively, are the values more dispersed than ASA companies (Table 3; Table 4).

The age of a firm could potentially affect their performance. The minimum age for both AS and ASA companies is 1, which tells those companies were only active one year before the observation period. Firms usually have a hard time being profitable in the beginning years of their existence. A younger firm could therefore have a lower ROA and ROE than older, well-established companies. The AS companies have an average age of 13.422 (10.219) years, while ASA are almost two times as old, with an average value of 31.229 (43.692) years (Table 3; Table 4).

As another control, industry dummies have been added to the regression. There are economical differences among industries. Each industry has their own characteristics. This is why it is important to take industry into account as a factor that could influence a firm’s performance. A total of 17 dummies is added to the regression with the 18th industry, category A, as reference category. The industries are categorized using the North American Industry Classification System (NAICS) 2017 revision(see Appendix for the industry categories).

Total assets in U.S. dollars is shown in the descriptive tables to give an indication of the assets for both groups. This variable is however not added to the regression, because total assets is an indicator for firm size, which has already been added in the form of total

employees. The average total assets for ASA firms is 2,261,278 (9,205,519) USD, which is more than the average total assets of the AS companies: 64,824.61 (373,312.4) USD (Table 3; Table 4).

Table 3: ASA companies descriptive statistics

Variable Obs Mean Std. Dev. Min Max employees 576 2,604.889 6,625.504 3 41,428 boardsize 576 8.55 2.655 3 17 firm age 576 31.229 43.692 1 243 totalassets 576 2,261,278 9,205,519 1.452 1.10e+08

Table 4: AS companies descriptive statistics

Variable Obs Mean Std. Dev. Min Max employees 690 944.951 1,049.656 8 12,428 boardsize 690 9.473 3.212 2 18 firm age 690 13.422 10.219 1 110 totalassets 690 64,824.61 373,312.4 2.616 4,300,00

(16)

NATHALIE BORST 2018 16 As shown in Table 5, there are in total 1,266 observations for 211 companies for the years 2005 till 2008. Those companies have 1,700.183 (4,608.386) employees on average. The average board size is 9.054 (3.006) board members. The age of the companies has a mean of 21.524 (31.675) years. The average total assets is 1,064,159 USD. The values for total assets are widely dispersed as the minimum is 1.452 and the maximum 1.10*108 US dollars with a standard deviation of 6,099,319 USD.

Table 5: All companies descriptive statistics

Variable Obs Mean Std. Dev. Min Max employees 1,266 1,700.183 4,608.386 3 41,428 boardsize 1,266 9.054 3.006 2 18 firm age 1,266 21.524 31.675 1 243 totalassets 1,266 1,064,159 6,308,096 1.452 1.10e+08

As shown in Table 6 and 7, for both ASA and AS companies have the number of employees gone up in the post-quota period. For the ASA companies average changed from 2,591.92 (6,968.512) to 2,617.858 (6,275.91) workers and for the AS companies from 922.638 (954.042) employees on average to 967.264 (1,138.229) employees. The number of employees is more widespread after the quota since the standard deviation increased with almost 200 employees.

A slight decrease in board size is observed in Table 6 for ASA companies. The mean went from 8.701 (2.438) to 8.4 (2.852) board members. This is the same for AS companies were board size slightly declined from 10.278 (3.3) members on average to 8.7 (2.915) (Table 6; Table 7).

Total assets have on average increased for both groups. The average total assets for ASA companies is 2,060,415 (7,740,508) USD in the pre-quota period and 2,462,142 (1.05e+07) USD in the post-quota period. The average total assets for AS companies almost doubled in the post-reform period. In the pre-quota period total assets are 42,213.36

(290,925) USD and 87,435.85 (439,848.3) USD in the post-quota period (Table 6; Table 7).

Table 6: ASA companies pre and post quota descriptive statistics

Variable Obs Mean pre Std. Dev pre Mean post Std. Dev post employees 288 2,591.92 6,968.512 2,617.858 6,275.91

boardsize 288 8.701 2.438 8.4 2.852

(17)

Table 7: AS companies pre and post quota descriptive statistics

Variable Obs Mean pre Std. Dev pre Mean post Std. Dev post employees 345 922.638 954.042 967.264 1,138.229

boardsize 345 10.278 3.3 8.7 2.915

totalassets 345 42,213.36 290,925 87,435.85 439,848.3

Results

For the first OLS-regression the asa variable is used as independent variable. Table 8 (1) shows the results of this regression. The asa coefficient has a value of -4.566 and is with a p-value of 0.00 significant with a 1 percent level of significance. This finding implies that ASA firms perform worse than AS firms. The ASA firms have on average a 4.566 lower ROA than the AS assuming ceteris paribus. Table 8 (4) shows ASA firms have on average a 21.688 lower ROE than AS firms when leaving everything else constant. These results show firms who are established as an ASA perform worse than firms who are established as an AS. Surprisingly, firm size measured by the logarithm of employees is significantly positively related with the ROA while it is negatively, yet not significantly, correlated with the ROE. This with the coefficients 1.378 (p=0.00) and -2.424, respectively (Table 8; Table 9).

Table 8 (2) shows the results of the second OLS-regression where roa is the dependent and post the independent variable. With a coefficient of -3.368 and a p-value of 0.00 is this positively related with the ROA, indicating firm performance is worse in the post-reform period than in the pre-post-reform period. The same result is found when the ROE is used as the dependent variable. With the coefficient -20.94 for the post variable, a significantly negative relationship is found between the post-quota period and firms’ ROE (Table 9 (5)).

The results of the third OLS-regression and thus the effect of the interaction variable are observable from Table 8 (3). The effect of the post dummy depends on the asa dummy, since the DiD-analysis requires an interaction term between those two variables. The effect of the post-quota period in comparison with the pre-quota period is measured by the post

coefficient plus the did coefficient. Where the did coefficient depends on the firm being an ASA or not. If a firm is an AS and the observation lies in the post-reform period, then the effect of post on the ROA is -0.623. Despite the negative correlation is this effect nonsignificant (p=0.47). The effect of the post variable on the ROA when a firm is an ASA is -6.846. This is the sum of the post and did coefficient. This signifies ASA firms operating in the post-quota period have a significant lower ROA than AS firms in this period. Both groups therefore perform worse in the post-quota period compared to the pre-quota period. However, this is only significant for ASA firms.

The effect of the asa dummy is in turn dependent on the post dummy. The effect of a firm being an ASA is estimated by the sum of the asa coefficient and the did coefficient.

(18)

NATHALIE BORST 2018 18 Where the did coefficient depends on the period. The effect of an ASA firm in the pre-reform period is indicated by the asa coefficient with a value of -1.543. This relationship is negative, however not significant (p=0.29) (Table 8 (3)). The effect of the asa dummy on the ROA in the post-reform period is significant. The effect of being an ASA firm on the ROA in the post period is -7.766, which is the sum of the asa and did coefficient. Concluding ASA firms perform worse than AS firms in both period. This effect however, is only significant for the post-quota period.

The did coefficient is the estimator of the effect of the mandated quota on firm performance. As shown in Table 8 (3), this effect is significantly negative (p=0.00). This signifies that, with a coefficient of -6.223, ASA firms in the post-reform period have on average a -8.389 (sum of β1, β2 and β3) lower ROA than those firms had in the pre-reform period and than the AS firms in both periods, leaving everything else constant. The last regression with the ROE as dependent variable, shown in Table 9 (6), supports this. The did coefficient is with a value of -23.284 significant and negative. Concluding, the introduction of the quota in 2008 has had a negative impact on firm performance for firms who had to comply with this regulation. This confirms my hypothesis that a mandated quota regarding gender diversity has a negative effect on firm performance.

In compliance with Dale-Olsen et al.’s (2013) results, board size has no effect on firm performance. A negative correlation exists between board size and the ROA with a coefficient of -1.119, yet this relationship is not significant (Table 8 (3)). The size of a board does also have no significant effect on the ROE (Table 9 (6)). Surprisingly, the size of a firm does have a negative effect on firm performance when it is measured by the ROA, but is not negatively correlated with firm performance measured by the ROE. With a coefficient of 1.537 and a p-value of 0.00 the variable Log(employees) has a positive effect on a firm’s ROA (Table 8 (3)). This relationship is negative for employees and the ROE with a

coefficient of -1.494 as shown in Table 9 (6). This means no conclusions can be made about the size of the firm and the firm performance. The effect of a companies’ age and their firm performance is also not elucidated. There is a positive correlation between age and ROA (0.02), but this is not significant. This in contrast to the significantly positive relationship between a firm age and ROE. Older firms perform slightly better on average (0.108) when measured by the ROE (Table 9 (6)). Companies operating in the industries: mining and quarrying, manufacturing, water supply: sewerage, waste management and remediation activities, wholesale and retail trade, accommodation and food service activities, information and communication, financial and insurance activities, real estate activities, professional, scientific and technical activities, administrative and support service activities, arts, entertainment and recreation and other service activities have on average a significantly higher ROA than the reference category agriculture, forestry and fishing. This is shown in Table 8 (3).

(19)

Conclusion and discussion

This study examines the effect of a mandated gender quota in the boardroom on firm

performance. Norway was the first country who introduced such a quota. As of January 2008 all public limited liability companies (ASA) had to have 40 percent of each sex represented in their board. I compare ASA companies with limited liability companies (AS), who did not have to comply, using a difference-in-difference (DiD) analysis. A lot of research has already been done about the effect of gender diversity in the board on firm performance, but this paper examines the effect of a forced gender diversity. All in all does the introduction of a gender quota have a negative effect on firm performance, based on the Norwegian data. When comparing companies that had to comply with those who did not in the periods before and after the introduction in 2008, a negative impact of this regulation was found. Results are based on data from 96 ASA companies and 115 AS companies for the years 2005 – 2010.

A negative relationship is found between the introduction of the quota and firm performance, measured by both ROA and ROE. ASA firms perform on average worse than AS firms over the observation period. No significant relationship is found between board size and firm performance. Furthermore, firm size measured by the number of employees is positively related to the ROA, but not to the ROE. Firm age has a positive effect on the ROE, but there is no effect on the ROA.

This conclusion rests on existing literature stating gender diversity does have an impact on firm performance. Furthermore, economic theory states that firms are driven by profit maximization. This signifies firms will optimize their organization in order to get maximum profits. According to this theory Norwegians board were already efficient before. The imposed mandatory male-female ratio has a negative effect on their efficiency.

Additionally, women in board will be treated differently based on “tokenism”. This could create obstacles to individual capacities and change the group interaction. Outsiders will think women got their position solely, because of their gender and not because of their quality. This could also drive women’s incentives as acquiring a board position would have been made easier for them. The introduction of the quota, which led to an exogenous shock in the market, could have affected board quality, because of a supply shortage. The amount of women with multiple board positions drastically increased, possibly because there were not enough qualified women with the needed governance experience after the quota entered force. This decrease in board quality and efficiency resulted in a decrease in firm performance.

With regard to this findings, there are some limitations to this research. First of all, this study has a sample size of 1, because only Norwegian companies were examined. To increase external validity a bigger sample with multiple countries is needed. This is

unfortunately hard to collect, since Norway has been the only country with such a quota so far. Additionally, a lot of public limited companies changed legal status in the transition years leading up to the quota. All these companies have been left out of the research. Also

companies who dissolved in the examined post-quota period (2008-2010) were left out of the sample. These implications could have led to a bigger impact on the ROA if this was taken

(20)

NATHALIE BORST 2018 20 into account. However, this is hard to measure. Like Dale-Olsen, Schøne &Verner (2013) already pointed out in their research, Norway transitioned from the Norwegian Genderally Accepted Accountring Principles (NGAAP) to the International Financial Reporting

Standards (IFRS) in 2005. Only listed companies have to follow the newly introduced IFRS and those companies are constructed as an ASA. All other companies could still follow the NGAAP. The transition to IFRS led to an increased capitalization of intangible assets and a removal of goodwill amortizations (Gjerde, Knivsflå, & Sættem, 2008). This regulation could have affected the reliability of the results, since not all companies use the same accounting principles and because in this study the measure of firm performance is accounting-based. For future research, suggestions are to increase the sample size. This research uses a dataset of 211 companies in total, which is not that big. To increase validity a bigger sample of both ASA and AS firms is desired. Furthermore, more control variables regarding board characteristics should be added to the analysis. Board size is the only board characteristic used as control variable in this thesis due to data unavailability. However, in exchange for money access can be granted to the Norwegian Register of Business Enterprises. This database has more information regarding Norwegian companies including more information about board characteristics. At the moment, the introduction of the quota is a fairly recent legislative measure. Therefore long-term effects are not yet visible. In the future, the impact of the quota can be re-examined to find out what the impact is on the long-term.

(21)

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. (2010). The changing of the boards: The value effect of a

massive exogenous shock. Unpublished manuscript. University of Michigan. Barney, J. B. (2001). Is the resource-based “view” a useful perspective for strategic

management research? Yes. Academy of management review, 26(1), 41-56. Bertrand, M., Black, S. E., Jensen, S., & Lleras-Muney, A. (2014). Breaking the glass

ceiling? The effect of board quotas on female labor market outcomes in Norway (No.

w20256). National Bureau of Economic Research.

Bhagat, Sanjai and Richard H. Jefferis, Jr. 2002. The Econometrics of Corporate Governance

Studies. Cambridge, MA: MIT Press.

Bowers, C. A., Pharmer, J. A., & Salas, E. (2000). When member homogeneity is needed in work teams: A meta-analysis. Small group research, 31(3), 305-327.

Bøhren, Ø., & 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.

Catalyst, 2004. The Bottom Line: Connecting Corporate Performance and Gender Diversity (Catalyst, New York).

Chung, K. H., & Pruitt, S. W. (1994). A simple approximation of Tobin's q. Financial

management, 70-74.

Coate, S., & Loury, G. C. (1993). Will affirmative-action policies eliminate negative stereotypes?. The American Economic Review, 1220-1240.

Dale-Olsen, H., Schøne, P., & Verner, M. (2013). Diversity among Norwegian boards of directors: Does a quota for women improve firm performance?. Feminist Economics,

19(4), 110-135.

Demsetz, H., & Lehn, K. (1985). The structure of corporate ownership: Causes and consequences. Journal of political economy, 93(6), 1155-1177.

Dobbin, F., & Jung, J. (2010). Corporate board gender diversity and stock performance: The competence gap or institutional investor bias. NCL Rev., 89, 809.

Dovidio, J. F., Brown, C. E., Heltman, K., Ellyson, S. L., & Keating, C. F. (1988). Power displays between women and men in discussions of gender-linked tasks: A

multichannel study. Journal of personality and Social Psychology, 55(4), 580. Elstad, B., & Ladegard, G. (2012). Women on corporate boards: key influencers or tokens?.

Journal of Management & Governance, 16(4), 595-615.

Erhardt, N. L., Werbel, J. D., & Shrader, C. B. (2003). Board of director diversity and firm financial performance. Corporate governance: An international review, 11(2), 102 111.

Farrell, K. A., & Hersch, P. L. (2005). Additions to corporate boards: the effect of gender.

(22)

NATHALIE BORST 2018 22 Forbes, D. P., & Milliken, F. J. (1999). Cognition and corporate governance: Understanding

boards of directors as strategic decision-making groups. Academy of management

review, 24(3), 489-505.

Gender Equality Act § 1-1 (Nor.). Retreived from

https://www.regjeringen.no/en/dokumenter/the-act-relating-to-gender-equality-the /id454568/

Gjerde, Ø., Knivsflå, K., & Saettem, F. (2008). The value-relevance of adopting IFRS: Evidence from 145 NGAAP restatements. Journal of International Accounting,

Auditing and Taxation, 17(2), 92-112.

Jourová, V. (2016). Women on corporate boards: Europe is cracking the glass ceiling. European Commission. Retrieved from

http://ec.europa.eu/newsroom/document.cfm?doc_id=46280

Jungmann, C. (2006). The effectiveness of corporate governance in one-tier and two-tier board systems–Evidence from the UK and Germany–. European Company and

Financial Law Review, 3(4), 426-474.

Kanter, R. M. (1977). Some effects of proportions on group life: Skewed sex ratios and responses to token women. American journal of Sociology, 82(5), 965-990.

Lechner, M. (2011). The estimation of causal effects by difference-in-difference methods.

Foundations and Trends in Econometrics, 4(3), 165-224.

Lee, C., & Farh, J. L. (2004). Joint effects of group efficacy and gender diversity on group cohesion and performance. Applied Psychology, 53(1), 136-154.

Nekhili, M., & Gatfaoui, H. (2013). Are demographic attributes and firm characteristics drivers of gender diversity? Investigating women’s positions on French boards of directors. Journal of business ethics, 118(2), 227-249.

Pande, R., & Ford, D. (2012). Gender quotas and female leadership. Public Limited Liability Companies Act § 6-11a (Nor.). Retreived from

https://www.oslobors.no/ob_eng/Oslo-Boers/Regulations/Acts-and-regulations Public Limited Liability Companies Act § 6-4 (Nor.). Retreived from

https://www.oslobors.no/ob_eng/Oslo-Boers/Regulations/Acts-and-regulations Rogelberg, S. G., & Rumery, S. M. (1996). Gender diversity, team decision quality, time on

task, and interpersonal cohesion. Small group research, 27(1), 79-90.

Rosener, J.: 1995, America's Competitive Secret: Utilizing Women as a Management Strategy (Oxford University Press, New York).

Ryan, M. K., & Haslam, S. A. (2005). The glass cliff: Evidence that women are over

represented in precarious leadership positions. British Journal of management, 16(2), 81-90.

Seierstad, C., & Opsahl, T. (2011). For the few not the many? The effects of affirmative action on presence, prominence, and social capital of women directors in Norway.

Scandinavian Journal of Management, 27(1), 44-54.

Spencer, S. J., Steele, C. M., & Quinn, D. M. (1999). Stereotype threat and women's math performance. Journal of experimental social psychology, 35(1), 4-28.

(23)

NATHALIE BORST 2018 23 Ebert-Stiftung, Internat. Policy Analysis.

Sweigart, A. (2012). Women on board for change: The Norway model of boardroom quotas as a tool for progress in the United States and Canada. Northwestern Journal of

international law & business, 32(4), 81A.

Teigen, M. (2008): Norwegian quota policies, paper presented at the Nineteenth Meeting of

the Helsinki group on Women and Science, Brussels, 4–5 November, ISF paper

2008:12.

Tyson, L., Zahidi, S., & Hausmann, R. (2008). The Global Gender Gap Report 2007. In

World Economic Forum.

Vinnicombe, S. (2000). The position of women in management in Europe. Women in

management: Current research issues, 2, 9-25.

Wearing, C. A., & Wearing, B. (2004). Between glass ceilings: Female non-executive directors in UK quoted companies. International Journal of Disclosure and

Governance, 1(4), 355-371.

Webber, S. S., & Donahue, L. M. (1999). Examining the “double-edged sword” of diversity in work groups: A meta-analysis. Unpublished manuscript.

Wegge, J., Roth, C., Neubach, B., Schmidt, K. H., & Kanfer, R. (2008). Age and gender diversity as determinants of performance and health in a public organization: the role of task complexity and group size. Journal of Applied Psychology, 93(6), 1301. Welch, F. (1976). Employment quotas for minorities. Journal of Political Economy, 84(4,

Part 2), S105-S141.

Williams, K. Y., & O'Reilly III, C. A. (1998). DEMOGRAPHY AND. Research in

organizational behavior, 20, 77-140.

Wood, R. E. (1986). Task complexity: Definition of the construct. Organizational behaviour

(24)

Appendix

NAICS rev.2017 industries:

A - Agriculture, forestry and fishing B - Mining and quarrying

C – Manufacturing

D - Electricity, gas, steam and air conditioning supply

E - Water supply; sewerage, waste management and remediation activities F – Construction

G - Wholesale and retail trade; repair of motor vehicles and motorcycles H - Transportation and storage

I - Accommodation and food service activities J - Information and communication

K - Financial and insurance activities L - Real estate activities

M - Professional, scientific and technical activities N - Administrative and support service activities P – Education

Q - Human health and social work activities R - Arts, entertainment and recreation S - Other service activities

(25)

Table 8: The introduction of the quota and firm performance

OLS regression on return on assets

(1) (2) (3)

asa dummy -4.566** (1.156) -1.543 (1.46)

post reform dummy -3.368** (0.895) -0.623 (0.857)

did (asa*post) -6.223** (1.849) Log(Employees) 1.378** (0.431) 1.69** (0.453) 1.537** (0.43) Board size 0.028 (0.176) -0.021 (0.174) -0.056 (0.177) Firm age 0.018 (0.011) -0.003 (0.009) 0.02 (0.011) Industry dummies B 8.66* (4.232) 10.424* (4.02) 8.528* (4.012) C 8.571* (4.291) 11.091** (4.031) 8.514* (4.079) D 9.822 (5.295) 11.597* (5.073) 9.877 (5.029) E 10.133* (4.414) 14.593** (4.188) 9.996* (4.203) F 8.37 (4.341) 11.767** (4.03) 8.104 (4.143) G 11.584** (4.291) 15.491** (3.943) 11.451** (4.082) H 7.797 (4.217) 10.512** (3.952) 7.595 (3.997) I 9.66* (4.441) 14.092** (4.092) 9.464* (4.229) J 13.034** (4.494) 14.226** (4.334) 13.087** (4.306) K 32.062** (4.718) 33.198** (4.562) 32.4** (4.487) L 13.11** (4.731) 15.522** (4.597) 13.155** (4.423) M 11.308* (4.735) 14.71** (4.363) 11.245* (4.498) N 13.46** (4.413) 17.399** (4.087) 13.23** (4.207) P 3.992 (4.533) 8.399 (4.293) 3.919 (4.349) Q 3.41 (5.596) 6.879 (5.546) 3.647 (5.575) R 23.963** (7.098) 28.029** (6.633) 23.352** (6.924) S 21.041** (4.981) 25.541** (4.931) 20.875** (4.81) Constant -12.943* (5.087) -17.006** (4.94) -12.793** (4.91) R2 0.1458 0.1452 0.1648 Observations 1,266 1,266 1,266

Notes: robust standard errors are reported in parentheses. ** and * denote 5 and 1 percent level of significance, respectively based on a two-way t-test.

(26)

Table 9: The introduction of the quota and firm performance

OLS regression on return on equity

(4) (5) (6)

asa dummy -21.688** (5.947) -10.809 (7.417)

post reform dummy -20.94** (4.965) -10.637* (5.379)

did (asa*post) -23.284* (10.659) Log (Employees) -2.424 (1.927) -0.721 (1.897) -1.494 (1.873) Board size -0.496 (0.942) -0.841 (0.92) -1.119 (0.941) Firm age 0.094 (0.053) -0.004 (0.037) 0.108* (0.053) Industry dummies B 14.105 (11.242) 22.353* (10.342) 13.289 (10.808) C 15.685 (10.577) 27.373** (9.308) 15.307 (10.105) D 2.844 (14.339) 11.374 (13.435) 3.005 (13.763) E 13.393 (11.894) 34.509** (9.886) 12.246 (11.395) F 20.297 (12.93) 36.157** (11.544) 18.566 (12.646) G 32.497** (11.972) 50.975** (10.376) 31.472** (11.56) H 7.235 (13.205) 19.241 (11.079) 5.753 (12.76) I 19.628 (14.268) 40.479** (12.402) 17.864 (13.956) J 20.139 (11.47) 25.894* (10.798) 20.357 (11.162) K 121.614** (17.37) 127.419** (16.732) 123.22** (16.865) L 16.857 (15.776) 28.401 (15.161) 16.8 (14.589) M 9.039 (20.022) 25.189 (17.632) 8.366 (19.474) N 58.277** (14.582) 76.562** (13.221) 56.651** (14.168) P -12.158 (19.594) 8.807 (19.424) -12.651 (19.844) Q -51.524 (30.96) -34.634 (30.931) -50.218 (31.259) R 162.618** (43.259) 181.205** (39.533) 157.879** (41.563) S 122.679** (5.592) 143.95** (27.351) 121.256** (26.454) Constant 26.347 (16.854) 9.198 (15.252) 32.263 (16.874) R2 0.1648 0.1690 0.1804 Observations 1,230 1,230 1,230

Notes: robust standard errors are reported in parentheses. ** and * denote 5 and 1 percent level of significance, respectively based on a two-way t-test.

(27)

Effect of a mandated boardroom gender quota on firm performance : evidence from Norway