The relationship between short term firm performance and women quotas for the board : evidence from German firms

(1)

1

Amsterdam Business School

BACHELOR’S THESIS

The relationship between short term firm performance and women quotas

for the board: evidence from German firms

Name : Maria Bodnarescu

Student ID number : 10864709

Bachelor’s Programme : Economics and Business Administration Specialization : Finance and Organization

(2)

2

Abstract

This study investigates the effect of the women quota introduced in 2016 in Germany. Two types of women quota policies were introduced: binding target applicable for small firms and mandatory quota applicable for big firms. In this paper big firms refers to publicly listed companies with more than 2000 employees. The government set a women quota of 30% for the proportions of women on the board of directors for those companies. On the other hand, small firms are defined as publicly listed companies with less than 2000 employees. In this case, the board of directors has to set a binding target for the proportions of women on their board. Data from Germany and Austria were used to investigate whether the introduction of women quota had a positive impact on firm performance. Using a difference in difference approach, this paper compares return on assets and profit margins of Austrian and German companies from 2014–2017, where only German firms were hit by the reform. This study’s main finding is that there is no effect of women quota on firm performance. Neither the difference in ROA nor profit margin were significantly different than 0 for both policies. The results suggested that characteristic differences between men and women are not enough to find a positive short-term impact on performance. This study is the first to investigate a binding target’s effect (for proportion of women on board) set by the company.

(3)

3

1 Introduction

In recent years, there has been increasing public debate regarding women’s under-representation in the corporate world. MSCI investigated 4,200 companies globally and found that women held only 17.3% of board seats in 2017 (Eastman, 2017). Further, 22.6% of companies did not have a single woman on the board; only 31.5% of firms had at least three female board members. In the IT sector, the proportion of women on board was even lower—28.5% of tech companies had no woman on board; only 18% had three or more female board representatives (Eastman, 2017).

However, these numbers were even lower in previous years. In 2015, 27.5% of companies had no female board member; only 20.1% firms globally had at least three female board members (Eastman, 2017). One of the reasons behind the positive trend is that many countries implemented a quota for women in the board of directors, dictating that companies must have a minimum proportion of women on the board.

Many studies have been conducted previously regarding characteristic differences between genders in other fields, such as psychology; however, there is not much data for corporate finance. This study will focus on how heterogeneity affects the board’s behaviour, as well as this change’s impact on performance. In particular, this study will investigate the effect of two different women quotas implemented in Germany.

For board of directors, a statuary women quota of 30% was implemented in Germany on 1.1.2016. This regulation is applicable to listed companies with over 2000 employees. For firms with less than 2000 employees, the board of directors is required to set a binding target for female quota in their company ‘Frauenquote – sie betrifft mehr Unternehmen, als man denkt!’, n.d.). This study is performed to compare whether either of the policies has an effect on firm performance. This paper will investigate the research question: To what extent does gender diversification in the board increase firm performance?

Dale-Olsen et al. (2013) argued that women quota has no significant effect in the short-run in Norway and that either women who replaced men were less capable or the board’s influence in the short-run was too small to be detectable.

One motivation for this study is to see whether this also applies to other countries such as Germany. One main reason why the results of Germany and Norway could differ is that Norway is one of the wealthiest countries worldwide in terms of GDP per capital. Additionally, the economy’s size is greater in Germany than in Norway. Thus the competition environment, firms operate in, is likely to differ.

(5)

5

The paper has the following structure: first, previous literature on effect of statutory women quota is discussed. As Germany is the first country to implement self-imposed women quota, there is no previous literature on that specific policy. Extensive literature on the effect of women on the board is instead discussed. One reason why there might not be enough women on board is the bias in some companies. With self-imposed quota, firms could be motivated to overcome old patterns and engage exactly those many women on the board so that diversification equilibrium is reached. Thus each firm could exploit diversification’s positive effect, without being forced to engage more women than it would be optimal.

In the third section, description of economic methodology follows. Regression model is described in particular, as well as sample selection and construction. In order to establish whether there is a casual effect of women quota on firm performance, a control group was used. Austrian companies have been used in this study as a control group for German firms, because both countries are institutionally similar. On top of that both countries are

geographically close and part of the European Union, Austria and Germany share the same official language.

Correspondingly, in the next session, the result of this analysis is presented. A separate regression for both policies is performed. Although based on current literature, it is likely that there is no effect on performance for the mandatory women quota, there is generally a positive effect of women on board performance that could cause the second policy to have a positive effect. In the last section, conclusion and evaluation of this paper will follow.

2 Literature Review

2.1 Effect of statutory women quota on firm performance

One of the problems with government regulation for companies is that ‘mixed results are partially the result of governance measures that have a very modest level of reliability and construct validity’ (Larcker et al., 2013). For women quota, the literature also differs depending on the country and performance measure investigated.

Most important to note is that introduction of a new law like women quota could cause companies to rush to meet the requirements. Companies might end up with women getting hired, even though they do not meet professional expectations of positions they fill, simply because of the timing.

(6)

6

According to Gregory-smith et al. (2014), the moral value of including women in the board is already sufficient to implement such policy. They also found that although the board’s behavior is different when women enter board of directors, there is no evidence that there is an effect on a company’s bottom line performance (Gregory‐Smith et al., 2014).

On the other hand, Adams and Ferreira concluded that imposed gender diversity decreases firm performance in the US (2009). In this case, the reason for the decrease is that a gender diverse board has fewer takeover defenses. Thus, for well managed companies,

mandating women quota can negatively impact firm performance (Adams & Ferreira, 2009). Adams and Ferreira (2009) also concluded that women quota has positive effects on attendance records of board of directors. Men were less likely to join monitoring committees and boards with only male directives allocating less effort in monitoring (Adams & Ferreira, 2009). Thus there might be a positive long-term impact of a diverse board, not accounted for in the study.

In other fields like politics, there were already more long-term studies performed on the effect of women quota. Like, Beaman et al. investigated women quota’s role for

leadership positions in Indian village councils (2009). They found that exposure to female leaders weakens stereotype, improving perception of female effectiveness, such that gender roles are evaluated and women are more likely to attain such a position. According to Beaman et al., similar pattern could arise for private firms also (2009).

Based on data in Norway, Ahern and Dittmar (2012) argue that a statutory women quota has a negative impact on company’s performance. This is because any mandated severe constraint on selection of board of directors can lead to a large economic decline in company value. Another problem is that women who replaced men were significantly younger and less experienced, thus less able to perform such responsibility (Ahern & Dittmar, 2012).

Depending on country and performance measure investigated, it can be concluded that effect of women quota is mixed. Thus, overall based on current literature, it is reasonable to assume that there is likely to be no effect of statutory women quota on firm performance, such that ROA and profit margin are likely to be the same before and after 2016, for listed companies with more than 2000 employees.

Hypothesis 1: There is no relationship between the statutory women quota and firm

(7)

7 2.2 Effect of women on firm performance

This paper’s other important issue is whether a binding target set by board of directors

(instead of mandated quota by government) has a positive effect on firm performance. Jalbert et al. (2013) collected data over a period of ten years (1997–2006) from 6305 companies. They found that female CEOs perform better than male CEOs in terms of return (ROI and ROAA), market valuation, institutional ownership and sales growth (Jalbert et al., 2013).

Another reason it would be beneficial to include females on board is that the labour market might not be in equilibrium due to reputational bias. A board member, proven to be successful in other board positions will more likely be offered the job as a board

representative (Ferris et al., 2003).

Earlier, most board positions were held by male representatives. The reputational bias makes it more likely for a board representative to be male. This stickiness might be a

contributor why the market cannot operate sufficiently on its own to provide the board’s efficient diversification.

For countries with substantial gender imbalances and large sanctions for non-compliance, women quota could help break the glass ceiling (Comi et al., 2017). Thus, it allows high skilled and productive female workers to reach a position in the board of directors (Comi et al., 2017).

Also, according to Chang et al., CEO’s ability differences influence firm performance (2010). It is likely that not only CEO’s, but also board members’ abilities play a role in firm performance. Combining study results of Comi et al. and Chang et al., it can be concluded that positioning high skilled productive female on the board could improve firm performance.

Women behave differently than men and thus are likely to make different investment decisions than men. Like, companies with male and female executives are likely to issue debt or do acquisitions (Huang & Kisgen, 2012). Huang and Kisgen (2012) also argued that a characteristic difference between men and women is that women are less overconfident: women provided broader earnings forecasting and were more likely to exercise an option earlier. Women are also on average more risk adverse and hence are more likely to make decisions, more favorable for investors in the long-term (Huang & Kisgen, 2012).

Moreover, according to Ararat et al., increasing demographic diversity enhances firm performance by increasing monitoring (2012); thus, reducing the wedge’s negative effect. As a more diverse board has a positive effect on firm’s performance, increasing proportion of female representatives on board should increase performance.

(8)

8

Another study found a positive relationship between a gender diverse board and firm performance based on the data of 127 US companies (Erhardt et al., 2013).

As Germany is the first country to implement self-imposed women quota, there is no previous literature on that specific policy. Bias can prevent companies from hiring women, leading to a small portion of women on board. One important aspect of self-mandated quota is that for a diverse board rather than government the company chooses the number of female representatives, so that with self-imposed quota, firms can be motivated to overcome old patterns, engaging exactly that many women on board required to reach diversification equilibrium.

Thus, each firm might be able to exploit diversification’s positive effect, without being forced to engage more women than would be optimal; such that it could be the case that a binding target (set by company) for proportion of female board members has positive effect on performance.

Hypothesis 2: There is a positive relationship between the binding target for women quota

and firm performance.

3 Data methodology

3.1 Regression models

For big companies, a statutory women quota of 30% was implemented in 2016; while for small companies, the supervisory board had to set a binding target for the company. In order to test the effect of the two policies on performance, four panel data regressions are

estimated: two for each dependent variable, one for small and one for big companies. The two different measures, using a proxy for firm performance, are ROA and profit margin. The study uses two different proxies in order to evaluate a company’s performance from different angles. As already stated, a separate regression is performed for each measure. A difference in differences regression is used in order to evaluate whether there is an effect on performance in case of each policy (stationary and self-imposed women quota). The regression model used to test the effect is:

ROAi= 𝛽0+ 𝛽1*post+ 𝛽2* German+ 𝛽3 *postxGerman + 𝛽4 * LnSize

+ 𝛽5* manufacturing + 𝛽6 * Finance + 𝛽7 * LnSales+ 𝛽8 * LnAge+ 𝜀i

Profit margini= 𝛽0+ 𝛽1*post+ 𝛽2* German+ 𝛽3 postxGerman+ 𝛽4 * LnSize

(9)

9

Table 1 definition of variables used

Variable Description

Dependent

Profit margin net income/total sales based on info given in respective balance sheets

ROA return on asset: net income /total assets calculated same way

Independent

German dummy variable: 0 if Austrian company, 1 if German company where policy was introduced

post dummy variable: 1 if after 2016 (policy implemented) , 0 if before 2016

postxGerman German* post (interaction coefficient between two variables on which DiD is measured)

Control

Manufacturing dummy variable: 1 if company in the manufacturing industry, 0 if otherwise (as long as a company produces anything , it is accounted for in the manufacturing dummy)

Finance dummy variable: 1 if company involved in finance industry, 0 if otherwise LnSize natural logarithm of number of employees in a company

LnAge natural logarithm of company age (years since it has been founded) LnSales natural logarithm of total sales

3.2 Control variables

To account for effect of industry type on firm performance, companies are divided into three categories: manufacturing, financial and others. In order to prevent perfect multicollinearity and thus biasness, all other industries, which fit neither category (mostly in service industry), are not accounted for in industry dummies. Main industries in the sample not included as industry dummy are: IT, telecommunication, media, real estate and retail.

The main reason to divide firms into three different industries is that the amount of assets held by a company is industry based. In capital-intensive industries like manufacturing, companies tend to hold more intangible assets. Based on this, the ROA is likely to be lower compared to other industries such as, retail where companies hold mostly current assets.

These variables are also included in profit margin analyses. Although the industry’s type is unlikely to have an equally strong impact on profit margin as on ROA, it is still likely to have an impact, such that it is important to include 2 industry dummies in both regressions.

In case of size, age and sales, a natural logarithm has been used, as the relationship between the respective variable and performance is non-linear. A log-linear approximation is likely to better represent the variable’s true effect on firm performance. For example, each additional year of age, has a higher impact on performance in early years of a company’s

(10)

10

existence compared to when a firm exists longer. As the same logic applies to the other two variables, it makes sense to look at percentage change (ln), instead of absolute values.

3.3 Sample construction

The data in this study is collected mostly manually from yahoo finance. Some of the data is gathered from the balance sheets as well as websites from the respective company to fill in information like company age as well as any gaps.

In Germany there are in total about 100 big publicly listed companies of which 75 have the data easy accessible and fall under the restriction. This study investigates 75 German publicly listed companies in each of the regression. Although there are about 3,500 small companies in Germany, the time horizon did not allow collecting all of those data points, as the data was collected manually. For this reason, 75 small publicly listed companies have been chosen randomly.

In the small companies’ sample, 25 small firms from Austria have been included in the regression as a control group. In the big companies’ sample, 27 small firms from Austria have been used as a control group. The reason for this specific size in Austria is that there are 27 big and 25 small Austrian companies for which the data is easy available and the

companies fall under the restriction.

The data used in this study is from 2014-2017. The regression is based on a short time frame such that the performance of the correlation between German firms and Austrian firms is constant except for the effect of the women quota introduction. Additionally, it is not possible at the moment to investigate a longer time horizon as the policy was only implemented in 2016.

In this study, several restrictions have been implemented for the data selection. First of all, for a company to enter the random draw, it must have conducted business in all years from 2014-2017. Secondly, firms which are just subsidiaries to other companies have not been considered in the selection. This is because it is likely that the parent company is going to have a substantial impact on firm decisions and performance

The reason to include in addition to German also Austrian companies is that: Austrian firms have been used a control group to establish whether there is a casual effect between women quota and firm performance. In both countries the legal, political and economic environment is very similar. Furthermore, during the last couple of years there has been a lot of economic integration between Austria and Germany and both countries are part of the European Union. Additionally, in both countries as well the same language is spoken.

(11)

11

On the contrary, although Austria and Germany are structurally similar they are not exactly the same: for example Germany is a much bigger country than Austria. Based on these small changes, there could be some deviation in performance between Austrian and German companies.

3.4 Type of tests performed

Using the data collected, a time series panel data regression was performed using Stata. However, for that the right test specification is needed. First, to check for multicollinearity and thus potential errors in the model, a covariance analysis between dependent variables is performed. Secondly, the model type (fixed vs random effects) is determined using the Haussmann test. Lastly, the regression for both policies with both performance measures is tabulated and interpreted using correct model specifications.

4 Results

4.1 Descriptive Statistics

Table 2 big companies basic descriptive statistics

Variable Mean Std. Dev. Min Max

Profit margin 0.033 0.244 -3.006 0.736 ROA 0.034 0.0832 -0.717 0.679 German 0.722 0.448 0 1 Post 0.500 0.501 0 1 postxGerman 0.361 0.481 0 1 Manufacturing 0.546 0.499 0 1 Finance 0.155 0.362 0 1 LnSize 9.351 1.224 7.612 13.373 LnAge 4.145 0.995 1.945 7.597 LnSales 21.787 1.537 18.059 26.164

(12)

12

Table 3 small company basic descriptive statistics

Variable Mean Std. Dev. Min Max

Profit margin 0.004 0.481 -5.31 0.730 ROA 0.020 0.168 -1.356 0.796 German 0.737 0.441 0 1 Post 0.500 0.501 0 1 postxGerman 0.368 0.483 0 1 Manufacturing 0.432 0.496 0 1 Finance 0.137 0.344 0 1 LnSize 5.740 1.476 0.423 7.559 LnAge 3.77 0.908 2.484 7.509 LnSales 18.18 1.528 12.447 20.874

Descriptive statics of small and big firms are tabulated in tables 2 and 3. Although, variables are mostly similar, several important differences can be noted. First, the profit margin for small companies is lower than for big companies. The reason for this is that there are more firms with negative net income in the small firm sample.

Generally younger firms tend to more often have negative net income, than well-established older firms. Tables 2 and 3 show that small firms’ age is lower than big firms’ age. A lower profit margin for small firms is also consistent with the economic theory. As expected by logical reasoning, profit margin variance for small firms is twice as high as for big firms.

Another key difference is that LnSales is greater for big companies than for small companies. This is not a big surprise either, because big firms are larger than small firms such that they generate on average more absolute revenue.

In contracts, there are several similarities between big and small firms. The most important being that in both groups, the industry composition is similar: about half of the companies are based in the manufacturing industry and 15% in the finance sector.

(13)

13

Table 4 large companies correlation

LnSize LnSales LnAge German Manufacturing Post finance

LnSize 1.0000 LnSales 0.6529 1.0000 LnAge 0.0701 0.0748 1.0000 German 0.1582 0.1897 -0.1985 1.0000 Manufacturing 0.0883 0.1674 0.1987 0.0348 1.0000 Post 0.0000 0.0310 0.0000 0.0000 0.0000 1.0000 finance -0.1739 0.0635 0.0478 0.0111 -0.4694 0.0000 1.0000

Table 5 small companies correlation

LnSize LnSales LnAge German Manufacturing Post finance

LnSize 1.0000 LnSales 0.4498 1.0000 LnAge 0.0196 -0.1128 1.0000 German 0.1666 0.1553 -0.2148 1.0000 Manufacturing 0.2697 0.0114 0.1803 -0.0584 1.0000 Post 0.0000 0.0267 0.0000 0.0000 0.0000 1.0000 finance -0.1087 -0.0222 -0.0899 -0.2489 -0.3469 0.0000 1.0000

Another aspect, which could cause multicollinearity of the results and thus biasness, is: dependent variables being correlated with each other. Table 5 shows the correlations between different dependent variables for small companies. Except for Ln Sales and LnSize that have a correlation of 0.45 and thus are highly correlated, none of the other variables have a strong correlation with another, such that there should be no issue performing regression in the intended way.

The same applies when we investigate correlations in large companies (table 4). Except for LnSales and LnSize, there is no strong positive relationship. The reason for the strong relationship between both variables is likely to be: the bigger a company is the more sales it generates on average.

4.2 Regression output

As discussed in the methodology part, four different regressions were performed, one for each of the following case:

 ROA small firms

(14)

14

 ROA big firms

 Profit margin big firms

However to perform the right regression, the correct test specification is needed. Based on the Haussmann test, the random effect model better fits the analysis. Although in one of the four cases, the null hypothesis was rejected, random effect is used for every regression to

maximize comparability in this study. The Haussmann test’s results can be found in the appendix.

Table 6: big company performance

ROA Profit margin

LnSize -0.0061188 (0.008625) -0.0202501 (0.0165131) LnSales 0.0095708 (0.0066282) 0.017268 (0.0132661) LnAge 0.0068345 (0.0083081) -0.0225698 (0.0153579) German 0.0043173 (0.0188084) -0.0230744 (0.0420435) post 0.0066198 (0.0089563) 0.0344315 (0.043411) Finance -0.0256172 (0.0263655) 0.0865406* (0.0491591) Manufacturing -0.0069084 (0.0189713) 0.0307075 (0.035286) postxGerman -0.005355 (0.0105328) -0.0630461 (0.0510934) Constant -0.1424076 (0.1107734) -0.2556123 (0.2164395) N (observation) 388 388 N (group) 97 97 R2 0.0042 0.0556 RHO 0.70875401 0.12238924

Note: *, **, and *** represent 10%, 5%, and 1% significance levels respectively. The numbers represent the respective coefficient. The numbers in brackets represent the standard error of the variable.

The results of big companies’ regression performance can be found in table 6. Most variables used in table 6 are in both regressions statistically not significant. The finance

(15)

15

variable in the profit margin regression is the only variable with a statistical significance in table 6, but only at 10% significance level.

Furthermore, when inspecting the two performance measures output, it can be noticed that the results are statistically similar. For example, R2 is very low for both regressions, meaning that only a small portion of the variance in dependent variables can explain the variance in the respective performance measure. This in turn means that most of the firm performance is explained by variables not included in this study.

Based on first hypothesis, there should be no relationship between statutory women quota and firm performance, such that postxGerman should not be significantly different than 0. Based on results in table 6, this is indeed the case. Such that it can be concluded that in the short-run, a mandated women quota has no impact on firm performance in Germany.

Table 7 small company performance

ROA Profit margin

LnSize -0.0078184 (0.0107373) -0.0049906 (0.0334753) LnSales 0.0271878*** (0.0094057) 0.1012738*** (0.0271156) LnAge -0.0023751 (0.0156556) -0.0774285 (0.0495447) German -0.0460766 (0.0361334) -0.1437414 (0.1100942) post -.0004118** (0226351) 0.0699672 (0.0476219) Finance 0.0580572 (0.0440942) 0.0821127 (0.1396508) Manufacturing -0.0216448 (0.0310749) -0.0283798 (0.098351) postxGerman 0.0091483 (0.0264152) -0.0994217* (0.0556597) Constant -0.3882546** (0.1735729) -1.407831*** (0.5123501) N (observation) 380 380 N (group) 95 95 R2 0.0906 0.1144 RHO 0.52608493 0.73911386

Note: *, **, and *** represent 10%, 5%, and 1% significance levels respectively. The numbers represent the respective coefficient. The numbers in brackets represent the standard error of the variable.

(16)

16

Table 7 shows the same regression as table 6, except that this time small companies were used to perform the regression. For this group, the board of directors set a binding target for the proportion of women on board. Based on theory, a binding target (set by the company) for the proportion of female board members should have a positive effect on performance. Such that hypothesis 2 stated that there is a positive relationship between the binding target for women quota and firm performance.

As it can be seen in table 7, the variable postxGerman is 10% statistically significant in the profit margin regression. However, based on the sample size and ROA, this is not significant enough to consider postxGerman significant. The hypothesis was rejected using a z-test (b (postxGerman) =0 vs b (postxGerman) =1) at 5% significance. Such that it can be concluded that there is no effect of a binding target on the performance of women on board.

There are several reasons why the result differs from the hypothesis. Based on the theory, each firm should be able to exploit diversification’s positive effect, without being forced to engage more women than it would be optimal. Such that each firm would apply the right amount of diversification to reach equilibrium. However, in practice, this is not always the case. Self-imposed women quota being a new concept, the board of directors might not have enough information to diversify to the right extent.

Another possible reason why there is no effect of this type of women quota on firm performance is that there is simply no effect of such equilibrium being reached. Lastly, it could be the case that the time frame investigated was too short to find an effect.

5 Conclusion and Evaluation

5.1 Conclusion

This study investigated the short-term effect on performance of two different women quota policies: binding target and mandatory quota. One reason why there might be different results in the long run is that, there is more time for boards to optimize. The reason to analyze short-run effects is that the policy was implemented just recently, such that there is no long short-run study feasible at this moment. Also, the time frame investigated in this study was 2014–2017.

In this paper, two different measures were used as a proxy for firm performance, namely ROA and profit margin. Overall, in case of both policies inspecting both measures, no evidence on a significant positive effect was found. Thus, it can be concluded that in Germany both types of women quota have no effect on firm performance in the short-run.

(17)

17

More specifically, for the mandatory women quota, this is in line with the hypothesis. Based on past literature, the expectation was not to find any effect of mandatory women quota on performance.

Additionally, for the self-imposed woman quota there was also no effect found as well. As Germany is the first country to implement such type of women quota, there is no comparable study for this specifically. Correspondingly, other papers on the effect of women on company performance were reviewed instead. In most of the studies, a positive effect of women on firm performance was found. Based on logical reasoning, a positive effect of women on performance was expected: if firms were allowed to optimize, such that the efficient percentage of women (diversity) on board is achieved, this would enhance firm performance as well. However, in this study the opposite was found. Even in case of self-imposed women quota, there is no positive effect on performance.

One key point, which should not be overlooked, is that there might be a social

positive aspect, not investigated in this study. If the assumption that women are discriminated against is true, imposing a women quota might help women achieve more equal treatment. Moreover, this law could have a positive effect of women being willing to invest more in their career as the glass ceiling is broken.

5.2 Evaluation: Limitations of the study

This study’s most important limitation was that not all German and Austrian companies that were or would have been affected were used. This is because data had to be collected manually and thus there had to be a random draw used instead. Collecting all data would have required about three to four months of pure data collection that was not feasible in the time frame. Another issue, which arose due to time constraint, is that only the most recent number of employees was used. Thus, the change over the past four years in firm size has been disregarded, such that number of employees in a firm is an approximation.

Another limitation was that only ROA and profit margin were analyzed as a proxy for firm performance. They are just two of many accounting measures, out of many more that could have been investigated to get a broader picture. Only accounting measures were used and no market measures, which might yield a different performance profile.

Furthermore, due to sample size firms were divided into three industry categories instead of more. In capital-intensive industries like steel or utilities, companies tend to hold more intangible assets, such that their ROA is likely to be lower compared to other industries

(18)

18

like retail where companies hold mostly current asset. Industry diversification applied in this paper may not be deep enough to account for those differences.

However, in this paper, a problem with diversifying it further based on sample size is that there would not be enough companies in each industry category to analyze whether the effect is significant. Another problem with the industry category was that it was difficult to categorize firms as today many have various side businesses, besides their main business (some companies could have fallen into 2 categories).

An additional restriction in this study was that Austrian companies are not a perfect control group for German firms. Although both countries are structurally similar enough that the performance of German firms can be approximated by Austrian firms, they are not perfect substitutes. The biggest difference between both countries is that Germany is a much bigger country than Austria. In the future, a study could be conducted to investigate whether this size difference leads to differences in performance. For this the performance of Austria companies could be compared to firms in Bavaria instead of Germany. Bavaria is not only more comparable to Austria in terms of size, but also in terms of culture and mentality (based amongst other things based on their close geographic position).

Lastly, time horizon is another important limitation to consider. The time frame investigated in this paper is four years only. The reason for this is that the policy has only been implemented two years ago, such that there is simply not more data available. There is only a small time horizon analyzed. However, in the long run, there might be a different trend when investigating the data. It would be interesting to investigate this effect in a few years. The only problem in a few years would be to find a test group as the Austrian government implemented women quota for the board of directors for firms as of this year.

(19)

19

6 References

Adams, R.B., & Ferreira, D. (2009). Women in the boardroom and their impact on governance and performance. Journal of Financial Economics, 94(2), 291–309.

Ahern, K.R., & Dittmar, A.K. (2012). The changing of the boards: the impact on firm valuation of mandated female board representation. Quarterly Journal of Economics, 127(1), 137–97.

Ararat, M., Aksu, M., Cetin, T. A. (2015). How Board Diversity Affects Firm Performance in Emerging Markets: Evidence on Channels in Controlled Firms. Corporate

Governance: An International Review, 23(2), 83-103

Beaman, L., Chattopadhyay,R., Duflo,E., Pande,R., & Topalova.P. (2009). “Powerful Women: Does Exposure Reduce Bias?” Quarterly Journal of Economics 124: 1497–540.

Chang, Y. Y., Dasgupta, S., & Hilary, G. (2010). CEO Ability, Pay, and Firm Performance. Management Science, 56(10), 1633-1652.

Comi, S., Grasseni, M., Origo, F.,& Pagani, L.(2017). Where Women Make the Difference. The Effects of Corporate Board Gender Quotas on Firms' Performance across Europe. University of Milan Bicocca Department of Economics, Management and Statistics, 37, 1-39.

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.

Eastman, M.T. (2017). WOMEN ON BOARDS. n.p. MSCI, 1-23

Erhardt, N. L., Werbal, J.D.,& Shrader, C.B.(2003). “Board of Director Diversity and Firm Financial Performance.” Corporate Governance: An International Review,11(2), 102– 108

(20)

20

Ferris, P. S., Jagannathan, M. and Pritchard, A. C. (2003). Too Busy to Mind the Business? Monitoring by Directors with Multiple Board Appointments. Journal of Finance,

58(3), 1087-1111

Frauenquote – sie betrifft mehr Unternehmen, als man denkt! [Brochure]. (n.d.)

Frankfurt am Main: Audit Committee Institute e.V..

Für welche Unternehmen die Frauenquote künftig gilt (2015, march 3). Morgenpost Retrieved from https://www.morgenpost.de/politik/article138143111/Fuer-welche-Unternehmen-die-Frauenquote-kuenftig-gilt.html

Gregory‐Smith, I., Main, B.G. M., & O' Reilly, C. A. (2014). Appointments, Pay and Performance in UK Boardrooms by Gender. Economic Journal,124(574), 109-128.

Huang, J., & Kisgen, D.J.(2012). Gender and Corporate Finance: Are Male Executives Overconfident Relative to Female Executives? Journal of Financial Economics, 108, 822-839.

Jalbert, T., Jalbert,M.,& Kimberly F.(2013). The Relationship between CEO Gender, Financial Performance and Financial Management. Journal of Business and Economics

Research, 11(1), 25-33.

Larcker,D. F., Richardson, S. A. ,&Tuna, I.(2007). Corporate Governance, Accounting Outcomes, and Organizational Performance. Accounting Review, 82(4), 963-1008.

Maas,S.(2017, july 1).70 Prozent der Unternehmen wollen nicht mehr Frauen. Deutschlandfunk. Retrieved from http://www.deutschlandfunk.de/

Staley, O.(2016, May 3). You know those quotas for female board members in

Europe? They’re working. Quartz. Retrieved from

(21)

21

7 Appendix

7.1 Results of the Haussmann test

Table A1 ROA small

Note: use re based on result

Table A2 profit margin small

(22)

22

Table A3 ROA big

Note: use fe based on result

Table A4 Profit margin big