Evidence from California The effects of mandatory gender quota for corporate boards:

(1)

The effects of mandatory gender quota for corporate boards:

Evidence from California

Abstract: This paper investigates the effect of mandatory gender quota in California on the

stock returns of Russell 3000 firms headquartered in the state. Abnormal returns and cumulative abnormal returns on and around the announcement date of the approval are analysed. The results demonstrate that the gender quota had a negative effect on the stock returns of the affected firms which is stronger for firms with a low female presence in the boardroom. Firms that are internationalized through foreign ownership, experience a lower decline in their abnormal returns. A similar negative effect after the announcement of the gender quota on the stock return is observed for a sample of Russell 3000 firms headquartered in other U.S. democratic states.

Keywords: board gender diversity, corporate governance, gender quota, internationalization,

democratic state, event study, market reaction, abnormal returns.

University of Groningen 2018-2019

Study Program: MSc International Financial Management

(2)

1

1. Introduction

Achieving gender diversity in the boardrooms has become an important objective in the last decades. Studies show that women positively contribute to the quality of decision-making process when appointed to corporate boards (Matsa and Miller, 2013; Choudhury, 2015). Women have greater monitoring power (Adams and Ferreira, 2009), are more independent (Simpson et al., 2010), are ethically concerned (Bart and McQueen, 2013) and have higher risk aversion (Jianakoplos and Bernasek, 1998). Despite this, the relationship between board gender diversity and various measures of firm performance is not clear yet. The existing literature regarding this topic reports inconclusive results. Some authors find positive associations between female representation in the boardroom and firm performance (Carter et al., 2003; Nguyen and Faff, 2007; Dezso and Ross 2012), some find a negative association (Shrader et al., 1997; Adams and Ferreira, 2009), while others report a neutral relationship (Farrell and Hersch, 2005; Francoeur et al., 2008; Carter et al., 2010).

Country regulators dedicated significant attention to gender diversity in the boardroom and initiated diverse actions towards achieving gender parity on corporate boards. Countries such as Denmark, United Kingdom and Australia implemented voluntary disclosure on female directors following the comply or explain principle. Other countries introduced mandatory gender quotas that required firms to take actions in order to achieve a proportion of 30% to 40% female membership in the boardroom within a specific time frame. The first country to implement such a quota was Norway in 2005, followed by other countries such as Israel (2007), Iceland (2010), France (2011), Italy (2011) and Germany (2015) (see Table2, Appendix A).

(4)

3

the end of 2019. Further requirement is that by the end of 2021 firms should have at least two women in a boardroom with four members and three women in a boardroom with six or more members. Firms that do not comply with the quota requirements are subject to financial consequences.

Although gender quota is seen as an effective tool to rapidly increase the representation of women in the boardroom (Allemand et al., 2014), it is not globally acknowledged to be the right approach (Choudhry, 2015). Previous research has shown that the effect of gender quota legislation is ambiguous. Ahern and Dittmar (2012) document that gender quota had a negative effect on the stock price of affected firms in Norway. Moreover, due to the scarce supply of qualified female directors, the newly appointed women did not have sufficient experience for the directorship position. Nygaard (2011) states that cumulative abnormal returns are contingent upon the level of information asymmetry in the affected firms. Firms with low information asymmetry have significant positive cumulative abnormal returns around the announcement date of the gender quota. More recently, Eckbo et al. 2019 report a zero impact of gender quota on the abnormal returns of Norwegian firms.

(5)

post-4

legislation environment the female directors might not be entirely qualified for the position (Terjesen et al., 2015) and can be perceived as “quota-filling directors” (Labelle et al., 2015). This will lead to a sub-optimal board composition which is expected to reduce shareholders value (Nygaar, 2011). Since investors care about their wealth, they are likely to react to the new legislation (Griffin et al., 2009). Based on the abovementioned reasoning, this paper will focus on the following research question: How does the mandatory gender quota in California affect the shareholder value of public firms based in the state?

(6)

5

democratic states experienced similar effects on their shareholders value at the gender quota in California.

The empirical analysis was conducted using a sample of 143 Russell 3000 firms headquartered in California. The results of the study confirm that the implementation of mandatory gender quota in California was perceived as value reducing by the investors, leading to negative and statistically significant abnormal returns of the sample firms. A firm included in the analysis has, on average, an abnormal return of -1.2% in the next trading day after the announcement and a cumulative abnormal return of -1.9% in the (-1;+1) event window. The shortfall of female directors to comply with the quota has a positive and significant impact on the magnitude of the negative abnormal returns. A firm that needs to appoint two women in the boardroom has, on average, 0.3% lower abnormal returns than firms that only need one additional female director. If a firm needs to appoint three women in the boardroom then its abnormal return is, on average, 0.9% and 0.6% lower than the abnormal returns of firms that need to add one and two women on board, respectively. However, the regression analysis does not show a significant linear relationship between shortfall and abnormal returns. This paper also finds a significant and positive relationship between firm’ internationalization and abnormal returns on the event date. An extra percentage in foreign ownership is likely to increase the abnormal return of the affected firm by 0.029 percentage points. However, the regression analysis fails to find a significant relationship between foreign ownership in countries with gender quota and the abnormal returns of affected firms. Finally, this study finds that firms located in other democratic states also experienced a decline in their returns after the announcement of the gender quota in California.

(7)

6

available in the market before mandating a gender quota. In addition, regulators may consider encouraging gender diversity starting from lower career levels so that the female talent is more visible, increasing the supply of qualified candidates for directorship positions. Second, investors can use the concepts and findings of this paper to assess the vulnerability of their wealth to gender quota legislation. Third, management and corporate boards of companies may use this paper to reconsider their current female representation in the boardroom and the internationalization policies in a pre-quota environment, especially in other democratic states in the U.S. If they start taking actions before the quota is implemented by increasing the number of female board members and the level of internationalization, they might cope easier with a potential implementation of a mandatory gender quota in the future. Finally, this paper can be used by academics to further develop the research regarding gender quota legislation.

This study contributes to the limited literature on gender quotas being one of the first papers that focuses on the case of California1_{. It focuses on the short-term economic effects of}

mandatory gender quota by accounting for the supply-side constraints in the female directors’ market. Moreover, this study adds to the literature by employing an analysis on two particular factors in the context of gender quota legislation. One is the role played by internationalization in the stock return reaction to the gender quota law. The second one is the effect of the gender quota legislation on shareholder value for firms located in states with similar political values. The paper is organized as follows. Section two presents a review of the extant literature on board gender diversity and gender quota and it continues with the hypotheses-development. Section three describes the methodology of the study. Section four elaborates on the results of the empirical study and section five discusses the conclusions of the paper.

1_{Next to:}

1. Hwang, S., Schivdasani, A., Simintzi, E , 2019. Mandating women on Boards: Evidence from the United

States. Unpublished working paper. University of North Carolina

2. Meyerinck, F., Niessen-Ruenzi, A., Schmid, M., Solomin, S. D., 2019. As California Goes, So Goes the

Nation? The Impact of Board Gender Quotas on Firm Performance and the Director Labor Market.

(8)

7

2.Theoretical background and hypotheses development

2.1. The gender quota in California

California is the first state in the U.S. that mandated a quota for the number of female directors on corporate boards. On September 30th, 2018, the Senate Bill (SB) 826 that sets the requirements of a more diverse board has been established. One of the reasons for implementation of the gender quota stated in the SB 826 is that “More women directors serving on boards of directors of publicly held corporations will boost the California economy, improve opportunities for women in the workplace, and protect California taxpayers, shareholders, and retirees” (Senate Bill 826, chapter 954, section 1.a). The new law has been introduced in California by two democratic senators on January 3, 2018. It was passed to the House on May 31, 2018, after being read by the Insurance, Banking and Finance and the Judiciary committees. On August 29, 2018 the bill was passed to the California Senate. Finally, the bill was signed and enacted on Sunday, September 30, 2018.

(9)

8

After the gender quota was signed into law on Sunday, September 30, 2018 an it received extensive attention from media on Monday, October 1. Hwang et al. (2019) investigate the trends on Google search for the period since the law was introduced until one week after the pass of the law. They observe that from January 1 until October 1, the search volume for the “SB 826” and “California women on boards” was low. However, on October 1 the attention for the new law increased sharply, which implies that investors awareness about possible consequences regarding the new legislation also increased.

The SB 826 in California was first anticipated by a non-binding resolution in 2013. The Senate Concurrent Resolution 62 encouraged a diverse gender representation on the corporate boards of publicly traded companies by increasing the number of female directors by one to three, depending on the board size by 2016 year-end (Senate Concurrent Resolution 62, Chapter 127, 2013). Since firms were not exposed to a legal duty of increasing their board gender diversity, the compliance with the resolutions achieved only 20 percent by the end of the period (Board Governance Research LLC Report, 2017). Related non-binding resolutions were introduced in other four states in the U.S., which are, similar to California, controlled by democratic parties. Table 4, Appendix A offers an overview about the timing and the requirements of the non-binding quotas imposed in those states.

2.2. Board gender diversity

(10)

9

2.1.1. Why does board gender diversity matter?

Gender diversity is defined as the heterogeneity of gender in the workplace. When the share of females is equal to the share of males the gender parity is achieved (Nielsen and Masden, 2017). Similarly, board gender diversity is defined as the heterogeneity of gender in the boardroom (Barako and Brown, 2008). This topic has received significant attention from the academic researchers. Most literature is focused on the effect of appointing women on corporate boards on decision-making process in the boardroom. This section elaborates on the female characteristics that are enhancing the quality of corporate governance when women are appointed to the board of directors.

First, women tend to have better attendance in the meetings than men and they are more likely to be elected for nominating, audit and governance committees which grant them more monitoring power (Adams and Ferreira, 2009). Likewise, firms with more gender diverse boards tend to have more meetings (Adams and Ferreira, 2004). In addition, Simpson et al. (2010) report that female directors are more independent than male directors since women that become a board member are more likely to come from outside the firm. Thus, being independent and having greater monitoring power allows female directors to keep management more accountable and alleviate agency problems.

(11)

10

Sobral, 2015) and environmental reporting (Rao et al., 2012) and dedication to improve the quality of working environment. As CSR represents a mean to enhance firms’ legitimacy (Panwar et al., 2014), firms with more gender diverse boards will benefit from more legitimacy. Finally, gender differences in risk taking influence the decision-making process in the boardroom. An experimental study shows that women are more risk averse than men (Levin et al. 1988), especially when it comes to financial decisions (Jianakoplos and Bernasek,1998). A larger proportion of female directors in the board is also associated with less volatility of corporate performance (Lenard et al., 2014). However, this view is challenged by Adams and Funk (2011) stating that once women succeed in breaking the glass ceiling, they are more likely to have similar characteristics as their male counterparts even with regards to risk-taking attitudes. Thus, the empirical evidence regarding risk aversion among female directors is mixed.

2.2.1 Board gender diversity and firm performance

Given that women contribute to the quality of decision-making process in corporate boards, a large academic literature is focused on the economic consequences of adding female directors in the board on various firm level performance measures such as market value, accounting returns and Tobin’s Q. (Catalyst, 2018). However, the findings of these studies are rather inconclusive and report contradicting results. This section introduces the existing evidence of empirical research regarding the relationship between board gender diversity and firm performance.

(12)

11

Dezso and Ross (2012) document that women in the board are value enhancing for firms that have strategies focused on innovation. However, since women that are qualified for directorship positions are scarce resources, they are privileged to choose which firm they want to work for (Farrell and Hersch, 2005). As a result, women directors turn out to be more inclined to serve better performing firms. Therefore, one should interpret the causal relationship between women in the boardroom and firm performance with caution (Lückerath-Rovers, 2013).

Other studies reveal a negative relationship between women in the boardroom and firm performance. Shrader et al. (1997) report that a higher fraction of women directors is likely to negatively affect the firm’s financial performance measured by accounting data such as returns to shareholders and firm’s earnings. Ryan and Haslam (2005) compare the firm performance before and after male and female directors are appointed. The results reveal that women are more likely to be elected by firms that have a lower performance when the stock-market is in a declining state, which confirms the hypothesis that female directors face a so called “glass cliff”, meaning that the roles they have to fulfil are riskier. In this sense, Francoeur et al. (2008) argue that women directors might outperform their male counterparts because the positions they are given are less promising from the moment of the appointment. Adams and Ferreira (2009) suggest that the impact of female board membership on firm value is dependent upon the nature of the firm’s corporate governance quality. A gender diverse board is more likely to result into negative firm performance for well-governed firms due to excessive monitoring.

(13)

12

board that generally derives from both external and internal forces allows female directors to choose the better performing firms. Likewise, Francoeur et al. (2008) and Carter et al. (2010) find that the participation of female directors does not have an impact on the firm stock-returns. In the attempt of reconciling the conflicting findings of the effect of board gender diversity on firm value Post and Byron (2014) combine 140 papers and analyse whether the legal and socio-cultural background of the sample firms affect the results. They conclude that female board participation has a positive impact on the accounting returns and this impact is amplified if the country-level shareholder protection is stronger. According to La Porta et al. (1998) Common Law countries offer stronger protection to their shareholder. Given that United States identifies itself as a country that uses the principles of Anglo-Saxon Law which has its origins in Common Law (Braendle 2006), a greater female participation in the board of directors is likely to result into positive effects on firm value for U.S. based firms. However, this effect can be contingent upon the selected approach to promote board gender diversity, which will be elaborated on in the next paragraphs.

2.3. The implications of a mandatory approach to achieve board gender diversity

The following paragraphs describe the potential disadvantages of a mandatory gender quota and what are the costs incurred to the affected firms.

2.3.1. Mandatory gender quota

(14)

13

(Ben-Amar et al., 2017). The enabling approach refers to the enactment of a set of rules and guidelines that are incorporated in the codes of corporate governance and fall under the “comply or explain” principle (Seidl et al., 2013). Finally, the coercive approach is seen as the most radical because it refers to laws that are mandated by the government, for which penalties are set in case of non-compliance. Since the progress towards a more gender diverse board is slow (Ross-Smith and Bridge, 2008), governments make use of legal means to stimulate gender diversity in corporate boardrooms. Still, the question raised by literature is whether this issue should be regulated, or it is better to let the market forces naturally administer the adjustment of female representation in the boardrooms to its optimal level (Labelle et al., 2015). This will, in turn, bypass the costs of such a legislation.

(15)

14

In the context of optimal board composition, Hermelin and Weinsback (2006) state that corporate governance represents a set of contracts between the major participants, namely the management, board of directors and the shareholders. The ‘optimal’ contract is designed to facilitate the maximization of firm value and avoidance of management opportunism. Subsequently, the board composition comes from optimal appointment decisions or contracts among shareholders, management and directors. From this view, any regulatory constraint that is introduced, such as obliging firms to appoint more female directors, will damage the contract and reduce firm value (Nygaar, 2011). In line with this arguments Labelle et al. (2015) document that firms faced with a regulatory approach to increase gender diversity in the board, experience a negative relationship between board gender diversity and firm performance as opposed to firms that use a voluntary approach, which experience a positive relationship. Consistent with this evidence the California Chamber of Commerce states that the corporate board composition has to be decided by the companies themselves and should not be mandated by the government (Grundfest, 2018).

2.3.1. Potential concerns

(16)

15

affected firms encountered a significant increase in their labour costs due to a lower number of layoffs and increased employment compared to the firms that were not affected by the quota.

Second, the limited supply of qualified female directors could lead to the appointment of unqualified female directors in the board to simply comply with the quota requirements. For example, Ahern and Dittmar (2012) find that the demand for female directors outweighed the supply of experienced and qualified female directors in the Norwegian market. This led to a sub-optimal board composition and a decrease in firm value.

Third, firms that are forced to increase the female representation in their boardroom might need to expand their boards (Hwang et al. 2019) instead of laying off male directors in order to make space for the new female members. Yermack (1996) finds that a larger board size leads to weak corporate governance. This can lead to higher expropriation by managers and a decline in firm value. On the other hand, qualified women that have the opportunity to sit on multiple boards will be less likely to sit on boards of poorly governed firms. Thus, the search and appointment costs for qualified women directors to comply with the gender quota will further increase.

2.4. Hypotheses development

The hypotheses are developed based on the literature review and empirical findings. They refer to the market reaction to the announcement of the gender quota, the effect of shortfall of female directors and firm internationalization level. In addition, the market reaction of the public firms headquartered in other democratic states is considered.

2.4.1. Market reaction to quota announcement

(17)

16

shareholders wealth is under pressure in the context of a sub-optimal board composition created by mandated quotas to increase the number of female directors. Thus, since rational investors care about the risk and the expected return on their portfolios (Modigliani and Miller, 1961), they will most likely react to the new regulation. Griffin et al. (2009) argue that the days in which the news is made public are of great importance since they make the investors aware of the expected changes in their wealth leading to a greater market reaction.

The efficient market hypothesis states that stock prices fully reflect the available information (Fama, 1970). Given the fact that informed investors gain information simultaneously (Hirshleifer and Subrahmanyam, 1994) and interpret the news in a rational manner (Myers and Majluf, 1984), the introduction of new regulation is likely to affect firms’ stock returns. Thus, a positive abnormal return after signing the quota legislation will signal that the new law is perceived as good news. This means that investors expect to gain value after firms comply with the requirements of the quota. However, negative abnormal returns can indicate that the quota is deemed as negative news and investors expect to lose value.

(18)

17

In contrast to the European countries, the selection of board members in United States is heavily based on social networks (Wiersema and Mors, 2016). Since the members of these networks are predominantly men (Boyallian et al., 2018), women face difficulties in being visible to men-dominated networks. Thus, in the context of a limited supply of female directors, the affected firms in California are expected to experience even higher costs for the search and employment of new female directors to comply with the gender quota. Moreover, the election committees might need to reinvent their whole recruitment process (Wiersema ad Mors, 2016), which will incur additional short-term costs.

Considering the limited supply of qualified female directors in the market at the enactment of the quota law, there are two aspects that might trigger the investors awareness concerning the vulnerability of their wealth. One is the anticipation that the new appointed female directors will not be sufficiently qualified for a directorship position which will lead to a sub-optimal board structure. The second aspect is the high search and recruitment costs incurred by the new law that are ultimately payed by shareholders. Hence, the law can signal bad news to the shareholders of the affected firms. Nofsinger (2001) indicates that investors react quickly to bad news, by selling their equity. Based on the arguments provide above, the following hypothesis is formulated:

Hypothesis 1:

The announcement of the gender quota law is likely to negatively influence abnormal returns of the public firms headquartered in California.

2.4.2. The effect of shortfall

(19)

18

percent have three or more female directors in the boardroom (Senate Bill, Chapter 954, section 1.e). According to Equilar Inc. data, among California headquartered firms that are part of the Russel 3000 Index as of September 2018, 377 firms need to appoint female directors on their board by 2021.2 Thus, it is interesting to see whether a different demand for female directors to comply with the quota is expected to incur different costs for the affected firms.

The “Shortfall”3_{of female directors to comply with the requirements of the quota by}

2021 can capture the different impact of the quota constraint across the affected companies. Taking into account the optimal board composition (Hermelin and Weinsback, 2006) and the limited supply of qualified female directors (Hwang et al., 2018), a higher need for female directors will signal higher searching and recruitment costs. In public firms, these costs are normally payed by the shareholders (Adams et al., 2011). Accordingly, assuming the acceptance of hypothesis 1, one could expect that the magnitude of the negative abnormal return will increase with the increase shortfall, leading to the following hypothesis:

Hypothesis 2

A higher Shortfall of female directors to comply with the gender quota by 2021 will positively influence the magnitude of negative abnormal returns of public firms headquartered in California as a reaction to the announcement of the gender quota law.

2.4.3. The effect of internationalization

The extant literature argues that internationalization might grant potential benefits to increase the ability of firms to overcome economic constraints. Johanson and Vahlne (1977) define internationalization as the process of integration of a firm’s operations in foreign

2_{According to Annalisa Barrett, Corporate Governance and Board of Directors Research Expert, September 1,}

2018: 66 companies have to add three women to their boards, 175 companies have to add two women to their board and 136 companies have to add one woman to their boards.

3_{This variable is also used by Hwang et al., 2019 and Eckbo et al., 2016. The shortfall of female directors is the}

(20)

19

markets. However, a company can be internationalized in two ways. One is through its production activities located abroad and the second is through its corporate governance practices (Hassel et al., 2003). This part is focused on the potential advantages of multinational firms in attracting female directors into their boards.

One argument suggests that the level of foreign involvement seems to play an important role in the way a firm is acquiring its human resources. Internationalized firms have greater access to global markets which gives multinational firms the advantage of accessing various resources at a lower price compared to domestic firms (Desai et al., 2007). Gao and Chou (2015, pp.280) argue that multinational firms have “better access to international human resources and knowledge stock”. Thus, in the context of a limited pool of qualified female directors’ in California labour market, multinational companies will be able to acquire female directors from the international labour market easier and at a lower cost than domestic firms.

(21)

20

Hassel et al. (2003) use the foreign ownership percentage to measure the level of firm internationalization. This shows the openness of a firms to the international capital markets. The higher the rate of foreign investors, the higher the firm’s openness and proximity to international financial markets, which in turn will increase the firm value. In this sense, Mishra (2014) provides evidence of a positive and significant impact of foreign ownership on firm value for a sample of 126 Australian firms. Sytse et al. (2006) find a positive relationship between foreign ownership and firms value measured by return on assets and Tobin’s Q.

Therefore, assuming that investors are aware of the advantages that multinational firms have in attracting female directors from the international market to comply with the gender quota, these firms are likely to experience less radical declines in their value. Taking into account the previous discussion, the following hypotheses is stated:

Hypothesis 3

A higher degree of foreign ownership in the public firms headquartered in California is likely to positively influence the abnormal returns reaction to the announcement of gender quota law.

(22)

21

recruitment processes that were developed in their countries to cope with the constraints imposed by the quota. Hence, investors might acknowledge that even though in the short term the company may experience high costs in their attempt to comply with the new law, the long-term effects might be value enhancing for the firm. Thus, the following hypothesis is stated:

Hypothesis 3.a.

A higher degree of foreign ownership concentrated in countries that already adopted a gender quota in public firms headquartered in California is likely to positively influence the abnormal returns reaction to the announcement of gender quota law.

2.4.4. Market reaction in other democratic states

On the premise that gender corporate policies harmonize with the institutional environment in a specific country, Terjersen et al. (2015) analyse the institutional factors that explain the implementation of the gender quota legislation. The authors conclude that left-leaning political parties are more likely to establish gender quotas legislation for boards of directors. The power resource theory suggests that left wing governments are more egalitarian (Rueschemeyer et al., 1992). Borre and Scarbrough, (1995) sustain that the equality of opportunity is associated with left-wing parties. Additionally, the gender equality figure as a ‘new value’ of left-leaning political parties’ agendas since 1980 (Inglehart, 1997). These arguments promote the establishment of gender quota in states were left-wing parties govern.

The most left-wing political party in the United States is the Democratic party and California has gained the reputation of a strongly democratic state4. Hence, the implementation of corporate gender quotas comes as a confirmation of the influence that the political parties

(23)

22

have on gender corporate policies. As reported by Gallup global analytics (2018), there are 22 democratic states in the United States that share the same political values in 2018. As a result of the pressing need for legitimacy (Aguilera and Cuervo-Cazura, 2004) the other democratic states can follow suit and implement the same regulation. This possibility in even higher given the fact that four democratic states already adopted non-binding resolutions to increase gender diversity in the boardroom (see Table 4, Appendix A). These resolutions are similar to the resolution that preceded the quota legislation in California with only of two years-timeframe difference.

This paper aims to examine the likelihood of the other democratic states to enact similar gender quotas in their corporate world and assess the stock price reaction of the firms headquartered in these states after the announcement of the gender quota law in California. When investors expect that other democratic states in U.S. will implement the same law as California did, the market will follow the same pattern and the abnormal return of the firms headquartered in those states will also experience a negative tendency. This leads to the following hypothesis:

Hypothesis 4

(24)

23

3. Methodology

This section describes the collection of data and the research methods used to perform the empirical analysis in this paper.

3.1. Data

The stock price data is collected from Compustat Daily via Wharton Research Data Services (WRDS) for a period of 13 months, from October 1, 2017, which is one year before the event until October 31, 2019, which accounts for a period of one month after the event. The raw sample of data contains 2570 firms that are spread across all states in U.S., 776 of which are firms from California. The gender quota law targets the public companies headquartered in California and most of the public firms in California are components of the Russell 3000 Index. Thus, firms that are constituents of the Russell 3000 Index are selected as the main sample of this research paper. Russell 3000 Index is a market-capitalization -weighted equity index which records the performance of the 3000 largest trading stocks in the U.S. (FTSE Russell v3.8). Every year, the index is reconstituted on the last Friday of June. By using the latest list of the Russell 3000 constituent firms, from June 25, 2018, the Californian firms that are part of the index are chosen using the Ticker code as a merging variable. The final sample of Russel 3000 Californian firms to be used in the event study consists of 143 firms.

For the event study methodology, a risk-free rate and a market portfolio return are required to be able to calculate the expected security returns using the Capital Asset Pricing Model (CAPM). A set of one-year (daily) treasury bill rates that measure the risk-free rate of U.S. treasury bonds are obtained from the official site of the United States government called U.S. Department of the Treasury. The market portfolio return is measured by the CRSP5_U.S.

Total Market Index and it is obtained from CRSP database via Wharton Research Data Services

(25)

24

(WRDS). A second proxy for the market portfolio return is the Russell 3000 Index return, which is later used to compute a robustness test. The daily data for this variable were obtained from Yahoo Finance database.

With regards to the second event study that is conducted for the sample of companies that are headquartered in the other democratic U.S. states, the same method is used as described earlier. Namely, the data was obtained from Compustat Daily and then the firms that are components of Russell 3000 Index were selected using the Ticker symbol a as merging element. According to Gallup6_{global analytics there are 22 democratic states in the United}

States, 14 of which are solid democratic and 8 are lean democratic. This classification is based on the political affiliation of the U.S. states residents in 2018 in terms of voters. A list of each group of states and the corresponding number of Russell 3000 firms which are contained in the final sample is presented in Table 3, Appendix A. Vermont state is excluded from the sample because of lack of data. The final sample consists of 453 Russell 3000 firms headquartered in other 20 U.S. democratic states.

The firm level data required to construct the Shortfall variable to test the second hypothesis for the sample of Russell 3000 Californian firms were collected from Orbis database. The number of female directors in the boardroom was obtained by summing all female directors while the board size is given by the database at 2017 year-end. After eliminating the companies that didn’t have the necessary data on the board of director to calculate the shortfall variable, a sample of 88 firms out of the initial sample of 143 firms was gathered.

To account for the high concentration of technology firms in California (Hwang et al., 2019) the Fama-French 12 industries classification is used. A description of how the sample

(26)

25

firms included in the regression analysis are spread across the 12 industries is presented in Table1, Appendix A. Most of companies in the sample come from business equipment (44%) and healthcare (23%) sectors.

With regards to the third hypothesis, data that describes the foreign owners, namely the percentage of shares held by each owner and accounting data necessary for the calculation of firm level control variables (Total Assets, Intangible assets, Debt, Earnings Before Interest and Taxes) was also obtained from Orbis Database. The set of variables is available for 83 firms of the initial sample and is collected for the end of 2017 fiscal-year. In this part of the study the financial services firms are excluded because they are highly regulated and they disclose different accounting data compared to the rest of the sample firms (Badenhorst et al., 2015). After excluding the financial firms and the firms that do not have the available information, a final sample of 70 firms is obtained to test the third hypothesis.

The information about countries that already mandated gender quotas is obtained from the European Commission report about gender balance on corporate boards. There are 9 countries that adopted binding gender quotas on corporate boards which are included in the construction of foreign ownership from countries with gender quota variable. These countries are Norway, Israel, Iceland, France, Italy, Belgium, Malaysia, India, Germany (see Table 2, Appendix A for more information about the quotas).

3.2. Research methods

(27)

26

value in this paper, the event study methodology is used. This is an important tool in economics and finance to determine the reaction of stock price to news (Douglas, 2001) around the time when the event occurred.It is a widely used method because it allows to investigate the impact of a “change in the regulatory environment” on firm value (MacKinlay, 1997; Schweitzer, 1989) or to evaluate the damages. At the foundation of an event study method stays the efficient capital market hypothesis which argues that security prices fully reflect the information available in the market on the (Fama, 1970).

3.2.1. Market reaction

The event study is performed using the statistical software Stata. Since the law was signed on a Sunday when the stock exchanges are closed, the next trading day is considered to be the event day which is Monday, October 1. The estimation window consists of 252 days, starting one year before the event, on October 1, 2017. To prevent the inclusion of information that might have been leaked to the market before the event, the estimation window ends one trading day before the signing of the law, on September 28. The methodology used to calculate abnormal returns in the event day is presented in the following paragraphs.

The daily stock return is calculated as follows:

Rit= (Stock price i,t – Stock price i, t-1)/Stock price i, t-1 (1)

where Rit is the return on security i on day t and Stock price i,t; i,t-1 is the closing price of the

security i on day t and previous day t-1 subsequently.

The normal ex post returns of securities in the event window that would be expected if the law announcement didn’t take place is calculated using the CAPM model as follows:

(28)

27

where NRit is the normal stock return of company i on day t, Rft is the one year treasury bond

rate on day t, βi is the non-systematic risk of the security i and (Rmt – Rft)is the market risk

premium with a market portfolio return Rmt equal to the CRSP value-weighted

daily market return.

The abnormal returns (AR) are calculated using the following formula: ARit= Rit - E(NRit) (3)

where ARit is the abnormal return of firm i on day t, Rit is the return of the stock i calculated

with formula (1) and E(NRit) is the expected return that would have occurred if announcement

on gender quota didn’t take place calculated with the formula (2).

Finally, the aggregated abnormal returns for the event window (AAR) are calculated by computing the mean of abnormal returns across the sample of observations as follows:

E(ARt)=1/N ∑𝑁_𝑖=1𝐴𝑅𝑖𝑡 (4)

To account for the information spills before the announcement and a possible underreaction of the stock price on the event day Nygaar (2011) calculates the cumulative abnormal returns for few return windows. Accordingly, the cumulative abnormal returns for two return windows (-1; -1) and (-2; +2) respectively are computed. The variables are both centralized on October 1, 2018.

The following formula is used to calculate cumulative abnormal return (CAR): CARi (-k; +k) = ∑𝑘_𝑡=−𝑘𝐴𝑅𝑖𝑡 (5)

where k is the day relative to the event day and ARit is the abnormal return of firm i on day t calculated in the formula (3).

(29)

28

different from 0 in a pre-specified interval that is 95% as well as normally and independently distributed. Taking into consideration that all the firms are affected by the law in the same time, the returns of the sample firms might be cross-correlated which reduces the accuracy of Student’s t-test. The portfolio approach (Jaffe, 1974) is used to control for the independence on the returns by calculating Patell (1976) t-test.

3.2.2. Robustness checks

In order to check the robustness of AR and CAR results obtained using the methodology described in 3.1.1, alternative estimation period was employed as well as different methods to calculate the expected normal returns. First, the estimation window was changed from 252 days to 120 days, as recommended by MacKinlay (1997). Next, the ex post return is determined using two alternative methods to CAPM. These are the mean adjusted return method (MARM) and the market model method (MMM). Lately, the return on Russell 3000 index is used as the market portfolio return proxy. Further, two non-parametric tests, namely the sign and the rank test, are performed to test the results. Additionally, the event study is performed for the alternate event dates before the passage of the law (introduction, passage to Senate and passage to the House) to check for spurious association and anticipation of the gender quota effect on firm value.

3.2.3. The effect of the shortfall

(30)

29

(fraction) of female directors in the board. A cross-sectional OLS regression at company level is employed7, as follows:

AR i = β0 + β1 Shortfall i, + β2Firm size i + β3 Board size i + Industry effect i + u i (1)

where AR i is the abnormal return of firm i on the event day, Shortfalli is the difference between

the number (fraction) of female directors required by the law and the current number (fraction) of women on board of company i, Firm size i and Board size i are firm level control variables.

The v i is a vector that accounts for the industry-fixed effects. The California firms are classified

across the Fama-French 12 industries. The industry level standard errors are clustered. The same equation is used to estimate the coefficient of Shortfall for the cumulative abnormal returns as a dependent variable.

Sensitivity analysis

An additional method to assess whether firms that have more female directors in the board at the event date are rewarded in terms abnormal returns is to compute a sensitivity analysis of AR and CAR to the Shortfall number (value). Thus, the same event study method is used for different sub-samples of firms.

First, the sample firms are ranked based on the shortfall number as follows. Firms that do not need women on their boards and have a Shortfall number equal to 0 respectively, constitute one sample. Similarly, sub-samples of firms with a Shortfall number equal to 1, 2 and 3 are created. Second, after eliminating the firms that, according to their board composition, do not need to add women in the boardroom to comply with the quota, the sample firms are divided into High and Low Shortfall according to the Shortfall fraction. A firm is

(31)

30

considered to have High Shortfall if the Shortfall fraction is above the median value of the variable and Low Shortfall if it is below the median.

3.2.4. The effect of internationalization

To inspect whether foreign ownership has an impact on shareholders’ value in the day of the event an OLS regression analysis is used. The dependent variable is the represented by the firm level abnormal return in the event day and cumulative abnormal returns the (-1;+1) event window. The independent variable in the regression analysis is the foreign ownership, which represents the degree of internationalization of the sample firms. It is a continuous variable measured by the Foreign Ownership as a percentage of Total Ownership (FOTO), following Mishra (2013) and Hassel et al. (2003). The percentage is calculated by summing up the percentage of shares owned by foreign shareholders situated outside the United States.

(32)

31

distribution of cash dividends in the future (Shih, 2013). Thus, the more intangible assets a firm owns, the higher its value will be. Leverage controls for the effect of capital structure on shareholders’ value and it is equal to Liabilities divided by total assets (Gande et al. 2009). Leverage can increase the firm value by creating a tax shield (Baker and Martin, 2011). Yet, leverage can also decrease the shareholders’ value by preventing a firm to invest in profitable projects (Myers, 1977). Finally, Market to Book ratio is measured by firm market capitalization divided by the book value of its assets. This variable is seen as a proxy for over or undervaluation of a company’s securities (Baker and Wurgler, 2002) as well as for the firm growth opportunities and profitability (Fama and French, 1995).

In addition, a Fama French 12 industries vector is employed in the regression to control for the industry effects. The standard errors at the industry level are clustered. All the firm level financial data as well as the abnormal returns and cumulative abnormal returns are winsorized at 0.01 level in order to remove the impact of outliers in the regression analysis (Pinkowitz et al., 2015).

Using the described variables, the following equation8 is utilized to estimate the impact of foreign ownership on shareholders’ value:

AR i = β0 + β1 Foreign Ownership i, + β2Firm size i + β3 Profitability i + β4Asset Tangibility i +

β5 Leverage i + β2Market-to-Book value i + Industry effect i + u i (2)

Foreign ownership in countries with mandatory gender quotas

In this part of the study, aside from the construction of the Foreign Ownership variable the same methodology as described in the previous part is followed. The independent variable

8_{The same equation is used to estimate the coefficient of Foreign Ownership for the cumulative abnormal}

(33)

32

is now equal to the percentage of equity held by investors located in countries that have already adopted a mandatory quota for female representation on board of directors (see Table 2, Appendix A). The OLS regression used to test hypothesis 3.a is the following:

AR i = β0 + β1 Foreign Ownership in countries with gender quota i, + β2Firm size i +

β3 Profitability i + β4Asset Tangibility i + β5 Leverage i + β2Market-to-Book value i +

Industry effect i + u i (3)

3.2.5. Market reaction in other democratic states

In this part of the study the same methodology for calculating abnormal returns as in the first part of this chapter is used. The daily stock returns (Rit), ex post returns (NRit),

abnormal returns (ARit) and cumulative abnormal returns (CARit) for individual companies and

states are calculated using the formula (1), (2), (3) and (5). Subsequently, the aggregated abnormal return (AARt) value is calculated using the formula (4). The aggregation of abnormal

returns and the cumulative abnormal returns is first determined for each state and then, for the full sample and the sub-samples.

(34)

33

4. Results

Based on the data and methodology described above, the findings of this paper are presented in the following paragraphs.

4.1. Descriptive statistics and correlation

The descriptive statistics are presented in Table 1. The board size varies between 4 and 23 members with a mean value of 10 directors. Although the female representation accounts for a maximum value of 8 women in the board the mean value is only 2 women in the board of directors. Yet, there are still companies that do not have women in the boardroom. The median value of the shortfall number equals 1.22 and the shortfall fraction is 0.16. This means that the median company in the sample in terms of Shortfall has to add one female member in the boardroom. Additional description of the shortfall number is presented in Table 1, Appendix B. The table shows that almost 12% of the companies in the sample have no women on board and they need to appoint 3 female directors by the end of 2021. Remarkably, 37% of the sample firms already comply with the requirements of the quota, 20% need to increase the female representation on board by one and 31% by two members.

(35)

34 Table 1: Summary statistics

Variables Observations Mean Standard

deviation

Minimum Maximum

AR (0) 70 -0.010 0.024 -0.107 0.065

CAR (-1;+1) 70 -0.012 0.036 -0.181 0.049

Board size 70 10.2 4.38 4 23

Number of Female directors 70 2.1 1.97 0 8

Proportion of Female directors 70 0.17 0.12 0 0.5

Shortfall (number) 70 1.229 1.024 0 3 Shortfall (fraction) 70 0.164 0.149 0 0.5 Firms size 70 14.49 2.14 9.61 19.72 Profitability 70 -0.02 0.37 -2.64 0.33 Asset Tangibility 70 0.186 0.182 0 0.495 Leverage 70 0.556 0.34 0.1 2.238 Market-to-Book 70 0.007 0.014 0.0003 0.0623

Foreign Ownership (% total) 70 19.57 11.99 1.4 58.97

Foreign Ownership (% in countries with gender quotas)

70 3.31 2.67 0 17.25

Correlation matrixes for the regressions are presented in Table 2 and Table 3 in Appendix B. The first matrix refers to the correlation between the variables that determine the impact of shortfall on the abnormal returns. Both the shortfall number and shortfall fraction are negatively correlated with the abnormal returns and cumulative abnormal returns and also negatively correlated with the board and firm size. This means that the implementation of the quota led to negative returns and it had a higher negative influence on smaller firms that have a small board of directors. Table 3 shows that the independent and control variables in the regression (2) and (3) do not have a strong linear correlation with the dependent variables (AR

(0) and CAR (-1;1)). Profitability and asset tangibility of the firms are negatively correlated with

(36)

35

4.2. Market reaction

Table 2 reports the average abnormal returns and average cumulative abnormal returns of the 143 Russell 3000 firms headquartered in California. Row (1) presents the return values on the next trading day after the law was signed (October 1, 2018), which constitutes the event date. The values of AR(0) , CAR(-1;+1) and CAR(-2;+2) are negative. These results are

economically and statistically significant at 1% level. Firms that were affected by the quota experienced, on average a decline of 1,2% in their return on the event day. Further, these firms had a decline of 1.9% in their cumulative abnormal returns in the (-1;+1) event window meaning that the stock price return declined further after the announcement of the law being signed. The average cumulative abnormal return in the (-2;+2) event window is equal to -1,1%, showing a stabilization in the return evolution. The results stay statistically significant after testing for cross-sectional independence of abnormal returns and cumulative abnormal returns using Patell (1976) t-test9. The results are robust to alternative estimation methods of abnormal returns and cumulative abnormal returns, a smaller estimation period as well as to the use of Russell 3000 Index return as a proxy for market return (see Table 4, Appendix 2).

The findings confirm the first hypothesis of this paper, showing that investors of the affected firms perceived the quota legislation as being value destroying. Hence, the findings are consistent with the theory that boards are structured to maximize the value of shareholders (Ahern and Dittmar, 2012). These results are in line with the findings of Hwang et al. (2018) and Ahern and Dittmar (2012) who report negative returns for firms affected by the gender quota law in the case of California and Norway, respectively.

Row (2), (3) and (4) report the abnormal returns and cumulative abnormal returns of the sample firms on the three other event dates. The results help to establish whether there was

9_{In addition, the sing and rank tests were performed. Their values also suggest statistical significance of AR} (0)

(37)

36

an anticipation effect at the introduction of the quota and its passage to the state institutions. The value of AR(0), CAR(-1;1) and CAR(-2;2) are not statistically significant in any of the three

event dates, meaning that they were not statistically different from 0. Thus, there is no empirical evidence for the anticipation of negative effects of the gender quota on firm value before the announcement of the law.

Table 2: Abnormal returns (AR) and cumulative abnormal returns (CAR) per announcement date

Event date #Firms AR(0) CAR (-1;+1) CAR(-2;+2)

(1) Passed governor (01/10/2018) 143 -0.01202*** -0.01906*** -0.01153*** t-Statistics (-6.15128) (-4.50429) (-2.52738) Standardized t-Statistics (-6.69092) (-3.07157) (-3.07157) Patell Z (-6.44364) (-2.47022) (-2.47022) (2) Passed House (29/08/2018) 143 -0.00019 -0.00343 0.00259 t-Statistics (-0.10853) (-0.77327) (0.54044) (3) Passed Senate (31/05/2018) 145 0.00868 0.00685 0.00563 t-Statistics (1.19782) (1.00846) (0.82674) Introduced (03/01/2018) 143 -0.00132 -0.00212 -0.00768 (4) t-Statistics (-0.61345) (-0.51507) (-1.75252)

Note: This table reports average abnormal returns and average cumulative abnormal returns for firms

headquartered in California that are constituents of Russell 3000 index, as of September 2018. AR (0), CAR (1;+1) and CAR (-2;+2) are calculated based on Capital Asset Pricing Model on the four event dates as follows: law introduction, passage to the Senate, passage to the House and the enactment of the law. The event date of the enactment of the law is the next trading day after the announcement, which is October 1st_{. The estimation period}

starts 252 days before the event and ends two days before the event. A firm is included in the portfolio if it has at least 100 non-missing observations in the estimation window and a return observation on the event date. The value of t-Statistics that test the hypothesis that the AR and CAR does not differ from zero are reported in parentheses. For the last event date (October 1st_{, 2018) the standardized t-Statistics and Patell Z (1976) are reported. ***, **}

(38)

37

To have a clear understanding of how the portfolio returns of Russell 3000 firms headquartered in California evolved, a graphical representation of average cumulative abnormal returns in a (-5;+5) event window is displayed in Figure 1. The graph shows a sharp decline of the portfolio cumulative abnormal return in the event day which is followed by a decline of similar magnitude on day one after the event. After a slight positive adjustment on day two, the abnormal returns continue to decrease on day three, four and five. This indicates that investors were reacting to the news for few days after the announcement of the law. Figure 1: The evolution of the cumulative abnormal returns in the (-5;+5) event window

Note: This graph shows the movement of the cumulative abnormal returns from 5 days before, until 5 days after

the event date (October 1, 2018). The sample is composed of California headquartered firms that are constituents of Russel 3000 Index as of September 2018. The solid line represents the mean cumulative abnormal returns of the sample firms, while the dotted line represents the mean +/-1.96 (standard error) of those values. The output was obtained from the event study application of WRDS.

4.3. The effect of Shortfall

(39)

38

sample of firms in Column (1) are again negative and statistically significant at 1% level. The average abnormal returns are notably decreasing with the Shortfall number (see Column (3), (4) and (5)). Firms that need to add one woman in the boardroom have a -0.9% abnormal return, which is statistically significant at 10% level while firms that need to add two or three female directors experience a higher decline in their returns, of -1.2% and -1.8% significant at 5% and 1% level, respectively. It is notable that the abnormal returns decrease with the increase of the number of women that firms need to add on their boards. A firm that needs to appoint two women in the boardroom has, on average, 0.3% lower abnormal returns than firms that need only one additional female director. If a firm needs to appoint three women in the boardroom then its abnormal return is, on average, 0.9% and 0.6% lower than the abnormal returns of firms that need to add one and two women on board, respectively.

Surprisingly, firms that already comply with the quota requirement also experience a negative abnormal return (see Column (2)). Although it is lower than the abnormal returns of other sub-sample firms (-0.6%), it is significant at 5% level, meaning that firms that are not expected to face the constraints imposed by the limited supply of qualified female directors are still negatively affected by the new law. The negative results might be associated with investors’ expectations that the female directors in those boards might be employed by other committees that need to appoint new women. This can lead to the situation when those directors are less dedicated to that firm which will negatively affect the firm value. An alternative explanation can be that, given that the sample firms are part of Russell 3000 Index, they are firms that strive for progress and expansion and investors might expect that those firms will grow in the future. Consequently, their board will expand and their demand for female directors will increase, leading to the search and recruitment frictions induced by the gender quota.

(40)

39

Column (4)). Because the sample of firms that have to appoint three women on boards consists of eight firms only, the results might be insignificant because of the lack of power. Finally, the results for the High Shortfall sample (see Column (7)) show that the firms comprised in the sample experienced a -1.6% cumulative abnormal return, significant at 1% level.

Table 3: Sensitivity analysis of abnormal returns and cumulative abnormal returns to the Shortfall of female directors to comply with the quota

Note: This table reports the average abnormal returns and average cumulative abnormal returns for the sample of

Russell 3000 firms headquartered in California that have data on the board composition as of September 2018. AR (0) and CAR (1;+1) are calculated based on Capital Asset Pricing Model on October 1, 2018, which is the next trading day after the enactment of the quota law. The estimation period starts 252 days before the event and ends two days before the event. The sample is composed of California headquartered firms that are constituents of Russel 3000 Index. A firm is included in the portfolio if it has at least 100 non-missing observations in the estimation window and a return observation on the event date. Column (1) uses a portfolio of all the firms in the sample. Column (2) uses a sample of firms that already have the required number of female directors in their boardroom. Column (3), (4) and (5) use the Shortfall (number) to divide the sample firms intro three sub-samples. Shortfall number equals to the difference between the required number of female directors by the gender quota and the current number of female directors on the event date. Column (6) and (7) use the Shortfall fraction to divide the sample firms in High and Low Shortfall sub-samples. Shortfall (fraction) is the fraction of women directors required by the law and the current fraction of female directors on board. A firm has High Shortfall if the Shortfall fraction is above the median value and Low Shortfall if it is below the median, respectively. The value of t-Statistics is reported in parentheses. ***, ** and * denote that the results are statistically significant at 1%, 5% and 10% levels.

Although the results are not entirely consistent across the measures of the Shortfall and those of returns, which can be due to limited number of firms in sample composition, this

(1) (2) (3) (4) (5) (6) (7) Shortfall (number) Shortfall (fraction)

All firms 0 1 2 3 Low High

AR (0) -0.0100*** -0.0062** -0.0090* -0.0125** -0.0189*** -0.0154** -0.0099**

t-Statistics (-4.1835) (-2.8953) (-1.7937) (-2.1779) (-4.27890) (-2.1326) (-2.9377)

CAR (-1;1) -0.0125*** -0.0068 -0.0050 -0.0227** -0.0253 -0.0136 -0.0162***

t-Statistics (-3.5504) (-1.5609) (-1.0219) (-2.8633) (-1.5180) (-1.4256) (-3.1491)

(41)

40

analysis shows that, on average, firms that have a higher need to appoint women in their boardroom face a larger decline in their returns. Thus, the first method used in this part of the study confirms that investors react to the new law implementation considering the constraints given by the limited supply of qualified female directors. The costs associated with these constraints increase with the fraction and number of female director additions required by the law, leading to a higher negative reaction captured by the median abnormal returns and cumulative abnormal returns of sample firms.

Table 4 shows the results of the regression (1). In the absence of control variables (see Column (1) and (2), Panel A), Shortfall number has a negative effect on AR and CAR significant at 5% and 10% level. That means that for each female director that a company needs to add on its board, the abnormal return will decrease, on average, by 0.46 percentage points and the cumulative abnormal returns will decrease by 0.81 percentage points, all other factors being constant. However, when the firm level control variables (firm size and the board size) are employed in the regression (see Column (3) and (4), Panel A) the coefficient is no more statistically significant. Panel B uses the Shortfall fraction as an independent variable in the regression. The coefficient of abnormal returns and cumulative abnormal returns are not statistically significant, meaning that there is no empirical evidence of a negative effect of the Shortfall fraction on firm value.

(42)

41

Table 6: Regression analysis on the effect of Shortfall on AR (0) and CAR (-1;+1)

Panel A: The effect of Shortfall number

(1) (2) (3) (4)

Variables AR (0) CAR (-1;1) AR (0) CAR (-1;1)

Shortfall number -0.00465** -0.00895* -0.00272 -0.000144 (0.00185) (0.00432) (0.00275) (0.00363) Board size -0.000777 -0.00123 (0.000911) (0.00177) Firm size 0.00289*** 0.00881* (0.000702) (0.00383) Constant -0.00452** -0.00185 -0.0408*** -0.128** (0.00162) (0.00607) (0.0115) (0.0515) Industry effects Y Y Y Y Observations 70 70 70 70 R-squared 0.037 0.063 0.063 0.185

Panel B: The effect of Shortfall fraction

(1) (2) (3) (4)

Shortfall fraction -0.0222 -0.0604 -0.00142 0.000816 (0.0173) (0.0332) (0.0200) (0.0202) Board size -0.000615 -0.00121 (0.000734) (0.00181) Firm size 0.00343*** 0.00886** (0.000932) (0.00359) Constant -0.00658* -0.00291 -0.0535*** -0.129** (0.00286) (0.00694) (0.0121) (0.0427) Industry effects Y Y Y Y Observations 70 70 70 70 R-squared 0.018 0.061 0.056 0.185

Note: This table reports the coefficient estimates of equation (1). AR (0) and CAR (1;+1) are calculated based on

(43)

42

Despite the significant results reported by Table 3, that show a logical negative relationship between Shortfall and abnormal returns, the regression analysis does not confirm the existence of a linear relationship between the Shortfall and abnormal returns. Thus, the evidence obtained by the two methods employed to test the second hypothesis should be interpreted with caution.

4.4. The effect of internationalization

Panel A in Table 5 reports the regression coefficients of equation (2) in which foreign ownership percentage is used as independent variable. The abnormal returns are positively affected by the percentage of foreign ownership. The coefficient estimate of stand-alone Foreign Ownership variable in Column (1) indicates that every extra percent of foreign ownership in total ownership will reward the company with an increase of 0.027 percentage points in its returns. This result is economically and statistically significant at 5% level. The foreign ownership coefficient remains negative and statistically significant at 5% level when the firm level control variables are added to the model in Column (3), meaning that an extra percent of foreign ownership leads to 0.029 percentage points increase in abnormal returns. The estimate coefficients of the foreign ownership when cumulative abnormal returns are used as dependent variable in Column (2) and (4) are not statistically significant. This demonstrates that the level of internationalization measured by foreign ownership has a positive impact on the firm value only on the event date.

(44)

43

foreign ownership at the announcement date, the positive impact of foreign ownership on abnormal returns is confirmed. This leads to the acceptance of the third hypothesis.

Panel B in Table 5 reports the regression coefficients for equation (3). This model aims to determine the relationship between the percentage of Foreign Ownership in countries with gender quota and the abnormal returns and cumulative abnormal returns. The coefficient estimates, although positive, are not statistically significant. These can be caused by the lack of power given the low median value of the independent variable (3.51 %) combined with a small sample size (70 observations). Hence, the model employed does not offer empirical evidence to support the hypothesis 3.a of this paper.

Table 5: Regression analysis on the effect of foreign ownership on AR (0) and CAR (-1;+1)

Panel A: The effect of total foreign ownership

(1) (2) (3) (4)

Foreign Ownership (%) 0.000278** 0.000335 0.000299** 0.000292 (0.000112) (0.000234) (9.32e-05) (0.000181) Firm size 0.000757 0.000971 (0.00117) (0.00262) Profitability -0.0132 0.00536 (0.0125) (0.00337) Asset Tangibility -0.00584 -0.0177 (0.00969) (0.0270) Leverage 0.00349 0.0192 (0.00467) (0.0141) Market-to-Book -0.619*** -1.134* (0.119) (0.537) Constant -0.0157*** -0.0194** -0.0236 -0.0313 (0.00386) (0.00737) (0.0189) (0.0421)

Industry fixed effects Y Y Y Y

Observations 70 70 70 70

Evidence from California The effects of mandatory gender quota for corporate boards:

The effects of mandatory gender quota for corporate boards: