• No results found

EFFECT OF THE APPOINTMENT OF A FEMALE CEO ON THE MARKET VALUE OF A FIRM

N/A
N/A
Protected

Academic year: 2021

Share "EFFECT OF THE APPOINTMENT OF A FEMALE CEO ON THE MARKET VALUE OF A FIRM"

Copied!
38
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

EFFECT OF THE APPOINTMENT OF A

FEMALE CEO ON THE MARKET VALUE OF A

FIRM

Sophie Groen

S1964208

Master Thesis, FEB, University of Groningen

Supervisor: Peter P.M. Smid

(2)

1. Introduction

The number of women in top management functions is a topical issue. Some European countries, such as Norway, Italy, France, Spain and most recently Germany, already

introduced gender quotas for corporate boardrooms. We hear more and more about women in top management but it is still a men’s world. The percentage of females in top management functions of firms of the Standard & Poor’s 1500 was 1.6 in 1992, it increased to 5.8 percent in 2000 and even rose to 8.7 percent in 2011, but growth decreased over the last couple of years (Dezsö, Gaddis Ross, and Uribe, 2016). For the position of Chief Executive Officer the number is even lower: in February 2016 only 20 of S&P’s 500 companies had a female leader (Catalyst, 2016). A journalist from the Huffington Post attentively noticed: ‘Female CEOs are kind of like shark attacks – extremely rare, but so well-covered by the media we think they’re pretty common’ (Peck, 2015). In times when gender equality is highly valued in the Western world, women still experience a so-called glass ceiling.

It is well known that certain characteristics are in general more expressed in women than in men and vice versa. These characteristics may help determine how a CEO runs a company. The question here is if the expectation of future firm performance with female leadership and the fact that female CEOs are still scarce lead to a specific reaction from the market. How do investors appraise a new female CEO? There is no consensus in literature yet about this subject.

Research question

Does the announcement that a female is appointed as CEO in the USA lead to significant other stock returns than the appointment of a male?

(3)

The relevant literature will be reviewed in chapter 2. Chapter 3 will elaborate the dataset. In chapter 4 I will explain the methodology, which includes the models used to measure a possible abnormal stock return, the statistical tests used to interpret the outcomes and the regression model. Then in chapter 5 the results are presented and in chapter 6 I will give the conclusion of this study.

2. Literature review

The traditional division of the tasks between men and women is still tangible. Currently 55 percent of university graduates in Europe is female, but their employment rate is 21 percent lower and the average wage gap is 15 percent (McKinsey, 2007). Women experience the burden of combining work and in most cases the responsibility of the household: the so-called double burden syndrome. And the perception of many females is that they have to make sacrifices in their personal live in order to have success in their careers. McKinsey (2007) used a sample of managers around the world, in which 33 percent of the women was single and 54 percent was childless, compared to respectively 18 and 29 percent for men. This is empirically supported by U.S. data: the more successful women are, the fewer children they have while this relation is reversed for men (Hewlett, 2002).

Where a minimum quota for females was mentioned before in the introduction, Dezsö et al. (2016) argue that there is an implicit maximum quota for females in practice. They find empirical evidence that firms make an effort to have a small number of women in the top but do not make an effort to have more. So in case there is already a position in top management filled by a woman the chance that another women will be appointed reduces. Furthermore, Hutzschenreuter, Kleindienst, and Greger (2015) find in an empirical study that outgoing CEOs who have the informal power to help select their successor are more likely to select a similar person. They assume that a person with the same demographic characteristics, e.g. gender, will continue their work and vision the best. Therefore the chance that the new CEO will be a female is lower when CEOs are primarily men.

(4)

characteristics typical for women may have a positive influence on the performance of the firm. Robert E. Quinn developed a model with eight different types of leaders; facilitator, mentor, innovator, broker, producer, monitor, coordinator and director (Quinn, 1990). In order to be an effective leader one should move easily from one type to another dependent on the situation. Kim and Shim (2003) find that the mentor- and the broker role are more expressed in women. The mentor role includes the qualities of effective listening, providing support and encouraging personal development, but the leader might be seen as soft and might have less authority. The broker role contains negotiation skills and the ability of building a network, but a negative point may be opportunism. Affinity with these roles creates an interactive work environment. These findings are empirically supported by Post (2005) and he states that female leadership leads to more cohesion, more cooperation and more participation. Moreover, there are fewer conflicts when more women are present in a work environment (Webber, 1987).

A general known characteristic of women is that they are more risk-averse than men. This is supported by Beckmann and Menkhoff (2008). For an empirical study they analysed fund managers in several countries and find indeed that female fund managers have a higher level of risk aversion and show less competitive behaviour. Another characteristic that is more expressed in women is ethical sensitivity. The empirical findings of Bear, Rahman, and Post (2010) support that if the number of female board members increases the corporate

responsibility strength ratings also tend to increase. But the presence of only one female member is not enough to increase the Corporate Social Responsibility of the firm. Bear et al. (2010) give as an explanation for this that it is harder for minorities to give their opinions and to be heard.

(5)

A final interesting difference between male and female directors is that the female directors have a higher attendance (Adams and Ferreira, 2009). Furthermore, they tend to motivate their male colleagues as the attendance of male directors rise when the board is more gender diverse.

But do these ‘female’ characteristics add value and leads diversity in the top management to better corporate results? The conclusions of prior research on this topic are divergent. Rose (2007) finds no empirical evidence of a connection between the performance of the firm and gender diversity in the boardroom, using Tobin’s Q as a measure for firm performance. The Tobin’s Q ratio measures the market value of the firm in relation to the total asset value of the firm. Gregory-Smith, Main, and O’Reilly (2014) also find no empirical proof that female representation in boards leads to better firm performance and state that female appointments are desirable from a moral point of view rather than economically relevant. Adams and Ferreira (2009) even warn about binding gender quotas as they find reduced firm value as a result of gender diversity of directors. Haslam, Ryan, Kulich, Trojanowski, and Atkins (2010) distinguish between accountancy-based measures of performance (Return on Assets and Return on Equity) and stock-based measure of performance (Tobin’s Q). They do not find empirical evidence that there is a significant relationship between accountancy-based measures of performance and the presence of female board members, but do find a negative relation between female representation in the board and a stock-based measure of

performance. Firms with at least one female board member were valued at 121 percent of the book value compared to 166 percent for firms with no women on the board. As a possible explanation they give that investors have stereotypic beliefs that women are less suitable as a leader. But besides that they mention the ‘glass cliff’ theory as a possible explanation. This is a theory that states that women are more likely to become a leader during periods of

organizational crisis or poor performance, for which the same authors found empirical evidence in an earlier study (Ryan and Haslam, 2005). If investors are aware of this phenomenon they might value the company lower.

(6)

Ferreira (2009) argue that boards with a higher level of gender diversity pay more attention to monitoring, because women will sooner accede monitoring committees. Furthermore,

McKinsey (2007) finds that a selected group of European firms with the greatest level of gender diversity in higher management positions scores much better than the average of their industry. The selected group had an Earnings Before Interest and Taxes of 11.1 percent, a Return On Equity of 11.4 percent and a stock price growth from 2005 to 2007 of 64 percent, where the industry averages were respectively 5.8, 10.3 and 47 percent. Arena, Crillo, Mussolino, Pulcinelli, Saggese, and Sarto (2015) find empirical evidence for a significant positive relation between gender diversity and firm performance, but only if there is a so-called ‘critical mass’ of women. The presence of only one woman has no effect. Dezsö and Gaddis Ross (2012) do also find empirical evidence that the presence of women in the top management of S&P 1500 firms leads to improved firm performance, but only if innovation is an important element of the strategy of the company.

What if we take it one step further and look at the gender of the CEO, the person with the final responsibility. Khan and Paulo Vieito (2013) find empirical evidence that US firms with a female CEO over the period 1992-2004 outperform firms led by man. They also find,

consistent with the theory, a lower level of firm risk for female CEOs. Martin, Nishikawa, and Williams (2009) argue that this knowledge is used in practice as they find in an empirical research that firms with a relatively high level of risk are more likely to appoint a woman as CEO, with the intention to reduce the risk. The findings of Peni (2012) also empirically support the hypothesis that a female CEO leads to a better firm performance. Zhang and Qu (2015) approached the issue from another perspective and argue that a possible gender change in a CEO turnover leads to a weaker firm performance due to an extra disruption in the

succession process. They find in their empirical study in China, where female CEOs are much more common than in the USA, that a male-to-female succession has a negative effect on firm performance and that it increases the odds of an early departure of the new CEO.

(7)

chairperson, president or any other C-suite position, compared to a male successor. On the contrary, Lee and James (2003) argue that the market reacts more negatively to the

appointment of a new female CEO than to the appointment of a male CEO. Moreover, they find in their empirical research that the effect on the share price is more negative when a female is appointed to the position of CEO than to another position in the top management. Both the empirical studies of Martin et al. (2009) and Ola and Proffitt (2015) do not detect a significant difference between the appointment of a male and a female CEO.

But other factors than gender, such as the source of the new CEO, the age or the industry, may also influence the reaction of the investors and should be taken into account. Lauterbach, Vu and Weisberg (1999) find empirical evidence that the postsuccession performance of external successors in the top management is better than the performance of internal

successors. And CEOs from outside the company receive on average a higher salary than new CEOs that were already connected to the firm for a longer time (Peyrache and Palomino, 2013). With a better performance of external managers one would expect that the financial market would react more positively on the appointment of outsiders for positions in the top management. This is in line with the empirical findings of Charitou, Patis and Vlittis (2010). They find significantly higher abnormal stock returns around the announcement day of a new external CEO for American companies. These findings are confirmed in an empirical study for poorly-performing companies on the French market by Dherment-Ferere and Renneboog (2000). They conclude that returns increase with more than 2 percent after the announcement of an external CEO and decrease by almost 1 percent after the announcement of an internal CEO. An explanation for the latter might be that the financial market holds the insider partly responsible for the recent poor performance. For well-performing companies they find no significant decline after the internal CEO appointment. This is in line with the empirical findings of Datta and Guthrie (1994) who state that firms with lower profits and growth are more likely to hire an outside CEO. A remarkable finding in this area is from the study of Lee and James (2003). As mentioned earlier they found empirical evidence that the financial market reacts more negatively to the announcement of a new female CEO. And in the case of a female CEO appointment they find that the market reacts even more negatively when the female CEO is recruited from the outside.

(8)

that age comes with experience and that is off course an advantage for a CEO. Therefore it is remarkable that over 2013 and 2014 the average return of the shares of the 10 companies with a CEO younger than 40 was 39.7 percent in contrast to 36 percent for the total S&P 1500 (Krantz, 2015). A well-known assumption is that the older you get, the more risk-averse you become. According to the empirical study of Rolison, Hanoch, Wood and Liu (2014) this statement is in general true. They find that when it comes to financial situations men tend to take fewer risks as they become older, but they could not find a significant decrease in risk taking for women. Serfling (2014) specifically studied the riskiness of corporate policies with regard to the age of the CEO and finds indeed empirical evidence that a negative relation between the stock return volatility and the age of the CEO. This can be explained by the fact that CEOs above the age of 59 maintain lower leverage ratios, make more diversifying

acquisitions, enter in more diversified operations and invest less in research and development. The effect on the riskiness of the company is even bigger when both the CEO and the second man in control are older.

And perhaps does the industry in which the firm operates matter. There are still big

differences between industries concerning the percentage of women in the labor force. In the USA it varies from 9.3 percent in the construction industry to 74.6 percent for education and health services (Bureau of Labor Statistics, 2015). Cook and Glass (2011) consider the

industry as a relevant factor as they find in their empirical study a bigger increase in the share price after the announcement that a woman is appointed to a senior leadership position in industries where more than 50% of the labor force is female.

All in all there is no clear consensus about this topic in the literature. Considering the different views and the fact that female CEOs are still relative uncommon in the USA, my hypotheses are:

Ÿ The financial market reacts more negatively, that is lower stock returns, to the appointment of a female CEO than to the appointment of a male counterpart.

Ÿ The stock returns are higher when the new female CEO is promoted from within the firm. Ÿ The stock returns are higher when the new female CEO is older.

(9)

3. Data

The event is specified as the announcement that a female is or will be appointed CEO in a publicly listed U.S. firm during the years 2002-2015. To obtain the data I used Orbis. Orbis is a database with information on millions of companies worldwide published by Bureau van Dijk. From the retrieved events, there were 5 female-to-female CEO successions. They are removed from the dataset and analysed separately. So the main dataset only contains male-to-female successions.

The exact announcement date and other information about the incoming CEO are retrieved from several press agency’s; Business Wire, PR Newswire, Bloomberg, Reuters and RTT News. The historical stock prices are obtained from Yahoo.finance.

The dataset contains 114 observations.

Ÿ In 84 cases the CEO was promoted from within the firm and in the other 30 the CEO was recruited from the outside.

Ÿ Of the appointed women 56 were 50 years or younger at the announcement date, 55 were older than 50 and of the remaining 3 the age is unknown. 50 years is chosen as the cut-off point, as the average age of a newly appointed CEO is 50 years (Todaro, 2003).

Ÿ In 44 of the cases the CEO turnover occurs in an industry where more than 46.8% of the labor force is female and in 70 cases in an industry where less than 46.8% of the labor force is female. This number is chosen because the average percentage of female workers in an industry is 46.8% (Bureau of Labor Statistics, 2015). The firms are allocated to an industry according to the North American Industry Classification System.

Figure 1: Distribution over the subsamples

Source of the CEO Age of the CEO Industry

(10)

To be able to compare the reaction of the market to the announcement of a new female CEO with the reaction to the announcement of a new male CEO I constructed a matched sample. This sample again only contains male-to-male successions. The events are matched by date, so each event is matched with the announcement of a new male CEO within 2 months.

This dataset also contains 114 observations.

Ÿ The CEO was an insider in 94 cases and was recruited from the outside in 20 cases. Ÿ 46 of the appointed men were 50 years or younger at the time of the announcement and the other 68 were older than 50 years.

Ÿ In 34 cases the CEO turnover occurred in an industry where more than 46.8% of the labor force is female and in the other 80 cases it occurred in an industry where less than 46.8% of the labor force is female.

So both the female and the male sample are divided in 6 subsamples, which further will be referred to as internal, external, ≤50 years, >50 years, female industry and male industry. And each pair will further be referred to as opposite samples, e.g. the internal sample is the

opposite sample of the external sample.

4. Methodology

Firstly, I will describe the event window and the estimation window. The second section is about the abnormal return. I calculate the abnormal return in two ways, which I will elaborate in section 3 and 4. In section 5,6,7 and 8 I will explain statistical tests, which I use to

determine the significance of the abnormal returns. In section 9 I will elaborate statistical tests which I use to see if there is a significant difference between the female and male sample and between the opposite samples. Section 10 contains the explanation of a cross-sectional regression model.

4.1 Event window and estimation window

(11)

will also use an event window of 4 days (τ= -4,-3,-2,-1), which further will be referred to as the pre-event window. The estimation window and the event window should not overlap and therefore I use an estimation window of 180 trading days from τ=-190 to τ= -11. By

including days after the event date I also capture a possible abnormal return if the

announcement was made after the closing of the market. On the other hand the returns on the days prior to the event could also be affected due to information leakage before the official announcement or due to the fact that the information might be available for the market slightly before the news agencies had the chance to report the news.

Figure 2: Time line

Estimation window Event window

τ0 τ1 τ2 τ=0 τ3

In the formulas the estimation window is (τ0,τ1) and the event window is (τ2,τ3).

4.2 Abnormal return

The return is calculated as follows:

R!! =P!!− P!,!!! P!,!!!

(1) Where Piτ is the closing price adjusted for dividends and stock splits of stock i on day τ and

Pi,τ-1 is the closing prices adjusted for dividends and stock splits of stock I on the trading day

prior to day τ.

The abnormal return is calculated by subtracting the expected return from the actual return.

AR!!  = R!!− E R!!

(2) Where ARiτ is the abnormal return on stock i in period τ, Riτ is the actual return on stock i in

period τ and E(Riτ) is the expected return on stock i in period τ without conditioning on the

(12)

I determine the expected return by using two different models: the constant mean return model and the market model. According to Brown and Warner (1985) both methods are equally able to detect abnormal performances. And using more complicated multifactor models does in general not lead to better results in event studies (MacKinlay, 1997).

4.3 Constant mean return model

In this model the expected return on the event date is the average return during the estimation window.

AR!!= R!!− R !!,!!

(3) Where ARiτ is the abnormal return on stock i in period τ, Riτ is the actual return on stock i in

period τ and R !!,!! is the average return in the estimation window.

For an event window of multiple days the average cumulative abnormal return is calculated as follows:

CAR!!,!! = AR!

!!

!!!!

(4) Where CAR !!,!! ) is the average cumulative abnormal return in the event window and AR! is

the average abnormal return on a certain date in the event window.

4.4 Market model

The market model assumes a linear relation between the return of a market index and the return of the individual security. An advantage of the market model compared to the constant mean return model might be that it reduces the variance of the abnormal return as it removes the part of the return related to the variation in the market return (MacKinlay, 1997). I use the S&P’s 500 as the market index. This is a stock market index based on 500 companies that are listed on either the NYSE or the NASDAQ.

AR!!= R!!− α!− β!R!!

(13)

Where ARiτ is the abnormal return on stock i in period τ, Riτ is the actual return on stock i in

period τ, Rmτ is the return on the market in period τ, αi is the intersection of the regression line

and βi is the slope coefficient of the regression line.

4.5 Parametric test from MacKinlay

θ = CAR τ!, τ! var CAR τ!, τ! !!

(6) Where θ is the test statistic, CAR τ!, τ! is the average cumulative abnormal return in the

event window and var CAR τ!, τ! is the variance of the average cumulative abnormal returns in the event window (MacKinlay, 1997).

The variance of the average abnormal returns is obtained by formula 7. var AR! = 1

N! σ!! !

!!!

(7) Where var AR! is the variance of the average abnormal returns on date τ, σ2ε is the variance

of the error terms/ abnormal returns of an individual event and N is the number of observations.

For a multiple-day event window formula 6 requires the variance of the cumulative abnormal returns in the event window as showed in formula 8.

var CAR τ!, τ! = var AR! !!

!!!!

(8) Where var CAR τ!, τ! is the variance of the average cumulative abnormal returns in the

(14)

test statistic exceeds the critical value or is lower than the negative equivalent of the critical value.

4.6 T-test

Unlike MacKinlay (1997), Brown and Warner (1985) use a t-test as a parametric test. This could by a valid alternative if one expects that the variance increases around the event.

t =

CAR

!!,!!

σ

!"#!!,!!

N

(9) Where t is the test statistic,

CAR

!!,!! is the average cumulative abnormal return in the event window,

σ

!"#!!,!! is the standard deviation of the cumulative abnormal returns in the event window and N is the number of observations. To determine the critical value I use the critical values of the student’s t distribution. The degrees of freedom used are

df = N − 1

(10) where df is the degrees of freedom and N is the number of observations.

The null hypothesis is that there are no abnormal returns in the event window and is rejected if the test statistic is higher than the critical value or lower than the negative equivalent of the critical value.

4.7 Corrado rank test

Normality of the distribution is assumed for both the t-test and the parametric test described by MacKinlay (1997). However in the case of daily returns the tails of the distribution are generally fatter than the tails of a normal distribution (Fama, 1976). Therefore I also do a Corrado rank test; a non-parametric test that does not assume a normal distribution.

According to MacKinlay (1997) it is best to use both the parametric and the non-parametric test to check for robustness.

(15)

highest abnormal return. Then the average distance from the middle rank on the event date is divided by the standard deviation of the average daily excess rankings to obtain a test statistic.

𝐶T = 1 N !!!! K!,!− K! 1 T !!!!! ! N1 !!!! K!,!− K! ! + N1 ! K!,!− K! !!! !! !!!! ! (11)

Where CT is the test statistic, N is the total number of observations, Ki0 is the rank of the

abnormal return on the event date, Ki is the median rank and T is the total number of days in

the estimation and event window (Corrado, 1989). Just like with the parametric test from MacKinlay I use the Z-score to determine the critical value, which is 2.575 for a significance level of 1%, 1.96 for a significance level of 5% and 1.645 for a significance level of 10%. The null hypothesis is that there are no abnormal returns in the event window and is rejected if the test statistic exceeds the critical value or is lower than the negative equivalent of the critical value.

4.8 Jarque-Bera test

To determine whether the distribution of the cumulative abnormal returns is normal I use the Jarque-Bera test. This test checks if the skewness and kurtosis match those of a normal distribution (Jarque and Bera, 1980). Skewness is used as a measure for the asymmetry of a distribution and kurtosis is a measure for the shape of the tails of a distribution.

𝐽𝐵 =𝑁 6 𝑆!+

1

4 𝐾 − 3 !

(16)

4.9 F-Test and independent T-tests

To check whether the average cumulative abnormal returns of the female and the male sample significantly differ and if the average cumulative abnormal returns of the opposite samples significantly differ I use an independent t-test. There are two tests: one in case the two

samples have equal variances and one in case the variances are different. This can be checked with an F-test. F =σ!"#! !!,!! ! σ!"# ! !!,!! !   (13) Where F is the F-statistic, σ!"#

! !!,!!

! is the variance of cumulative abnormal returns in the

event window of the first sample and σ!"#! ! !!,!! is the variance of the second sample. The

null hypothesis is that the variances of the two samples are equal and can be rejected if the test statistic exceeds the critical value of the F-distribution or is lower than the negative equivalent of this critical value. The degrees of freedom used are

df! = N!− 1 and df! = N!− 2,

(14) where df1 and df2 are the degrees of freedom and N1 and N2 are the number of observations in

respectively sample 1 and sample 2.

The formula for the independent t-test if the variances of the two samples are equal is as follows:

t = CAR! !!,!! − CAR! !!,!!

σ!! 1

N! +N1!

(15) Where CAR! !!,!!  and CAR! !!,!! are the average cumulative abnormal returns in the event window of respectively sample 1 and sample 2, N1 and N2 are the number of observations in

(17)

σ!! = N!− 1 σ!"#! ! !!,!! + N !− 1 σ!"#! ! !!,!! N!+ N!− 2 (16) Where σ!"#! ! !!,!! and σ !"#! !!,!!

! are the variances of the cumulative abnormal returns in the

event window of respectively sample 1 and 2 and N1 and N2 are the number of observations of

respectively sample 1 and 2. For the critical value I use the Student’s T-distribution with a degrees of freedom of

df = N!+ N!− 2,

(17) where df is the degrees of freedom, N1 is the number of observations in sample 1 and N2 is the

number of observations in sample 2.

If the variances of the two samples differ, the t-statistic can be computed as follows:

t = CAR! !!,!! − CAR! !!,!! σ!"# ! !!,!! ! N! + σ!"# ! !!,!! ! N! (18) Where CAR! !!,!!  and CAR! !!,!!  are the average cumulative abnormal returns in the event window of respectively sample 1 and sample 2, σ!"#

! !!,!!

! and σ

!"#! !!,!!

! are the variances

of the cumulative abnormal returns in the event window of respectively sample 1 and 2 and N1 and N2 are the number of observations in respectively sample 1 and 2. To determine the

critical value I use the student’s T-distribution. The degrees of freedom used are:

(18)

Where df is the degrees of freedom, σ!"#! ! !!,!! and σ

!"#! !!,!!

! are the variances of the

cumulative abnormal returns in the event window of respectively sample 1 and 2 and N1 and

N2 are the number of observations in respectively sample 1 and 2. For both tests the null

hypothesis is that the average cumulative abnormal returns do not differ and can be rejected if the t-statistic exceeds the critical value or is lower than the negative equivalent of the critical value.

4.10 Cross-sectional models

In this study I do not only want to see if there are abnormal returns at or around the event date but also if these abnormal returns are related to specific characteristics of the event. This can be achieved by doing a cross-sectional regression using Ordinary Least Squares in EViews.

AR! = δ! + δ!x!"+ ⋯ + δ!x!"+ ε!

(20) Where ARi is the abnormal return of stock i, δm, m=0, …, M are the regression coefficients,

xmi,m=1, …, M are the characteristics of stock i and ε!is the zero mean disturbance term.

For both the source (internal/external) of the new CEO and whether or not females are above average represented in the industry I use dummy variables.

Table 1: Dummy variables

Source Industry Internal External >46.8% females in labor force ≤46.8% females in labor force Dummy 1 0 1 0

OLS gets the most valid results assuming that the disturbance terms are cross-sectionally uncorrelated and homoskedastic. Homoskedasticity means that the errors have constant and finite variance. In this case the disturbance terms could as well be heteroskedastic. If the homoskedasticity assumption is not satisfied the OLS will still give unbiased estimators but the standard errors can be inappropriate and therefore we could come to the wrong

(19)

5. Results

In this chapter the results are presented. Section 1 gives an overview of the average abnormal returns in the total samples. In section 2,3,4 and 5 the results are given for respectively the one-day event window, the three-day event window, the five-day event window and the pre-event window. Section 6 contains the results of the ordinary least squares regression. And the results of a female-to-female succession are presented in section 7.

5.1 Average abnormal returns

Table 2 shows the average abnormal returns for the days in the event windows and figure 2 show these returns graphically. In general the shares of the firms with a new male CEO have higher average abnormal returns than the shares of the firms with a new female CEO.

Table 2: Average abnormal returns

Figure 3: Average abnormal returns t= CMRM Females (N=114) MM Females (N=114) CMRM Males (N=114) MM Males (N=114) -4 -0.30% -0.10% 0.01% 0.23% -3 -0.34% -0.19% 0.11% 0.05% -2 -0.42% -0.36% -0.17% -0.10% -1 -0.11% -0.08% 0.01% -0.20% 0 -0.24% -0.17% -0.14% -0.22% 1 -0.35% -0.27% 0.58% 0.53% 2 0.09% 0.08% 0.34% 0.16%

(20)

5.2 Results one-day event window

Table 3: Average abnormal returns for the one-day event window and distribution and significance of the average abnormal returns

There are no significant average abnormal returns for the female samples in the one-day event window. For the male sample the average abnormal returns are significant for both the

external subsamples and the male oriented industry with the market model, according to the parametric test from MacKinlay.

N AAR Normal distribution N AAR Normal distribution Females CMRM Males CMRM Total 114 -0.24% no 114 -0.14% no Internal 84 -0.49% no 94 0.11% no External 30 0.46% no 20 -1.33%! no ≤50 years 56 -0.95% no 46 -0.34% no >50 years 55 0.54% no 68 -0.01% no Fem ind. 44 -0.08% no 34 0.38% no Male ind. 70 -0.34% no 80 -0.37% no Females MM Males MM Total 114 -0.17% no 114 -0.22% no Internal 84 -0.42% no 94 0.00% no External 30 0.51% no 20 -1.23%!! no ≤50 years 56 -0.85% no 46 -0.16% no >50 years 55 0.48% no 68 -0.26% no Fem ind. 44 0.12% no 34 0.22% no Male ind. 70 -0.36% no 80 -0.40%! no

CMRM=Constant Mean Return Model, MM=Market Model

*,**,***#significant at the 10%, 5% and 1% significance level respectively, t-test#

!,!!,!!! significant at the 10%, 5% and 1% significance level respectively,

parametric test from MacKinlay#

,✚✚,✚✚✚ significant at the 10%, 5% and 1% significance level respectively, Corrado

Rank Test #

(21)

Table 4: Average abnormal returns for the one-day event window and significance of the difference between the average abnormal returns of the female and male sample and the opposite samples

On the event date the average abnormal returns in the female sample do not significantly differ from the average abnormal returns in the male sample. However, in the female sample the average abnormal returns in the two age groups significantly differ according to both models and for the male sample this is the case for the external and internal sample.

N AAR N AAR Females CMRM Males CMRM Total 114 -0.24% 114 -0.14% Internal 84 -0.49% 94 0.11%✚✚ External 30 0.46% 20 -1.33%✚✚ ≤50 years 56 -0.95%✚ 46 -0.34% >50 years 55 0.54%✚ 68 -0.01% Fem ind. 44 -0.08% 34 0.38% Male ind. 70 -0.34% 80 -0.37% Females MM Males MM Total 114 -0.17% 114 -0.22% Internal 84 -0.42% 94 0.00%✚✚ External 30 0.51% 20 -1.23%✚✚ ≤50 years 56 -0.85%✚ 46 -0.16% >50 years 55 0.48%✚ 68 -0.26% Fem ind. 44 0.12% 34 0.22% Male ind. 70 -0.36% 80 -0.40%

CMRM=Constant Mean Return Model, MM=Market Model

*,**,***#significantly differs from the other gender sample at the 10%, 5% and 1% significance level respectively#

,✚✚,✚✚✚ significantly differs from the opposite sample at the 10%, 5% and 1%

(22)

5.3 Results three-day event window

Table 5: Average cumulative abnormal returns for the three-day event window and distribution and significance of the average cumulative abnormal returns

In this event window the average cumulative abnormal returns are not significant for both the female and the male samples.

N CAAR Normal distribution N CAAR Normal distribution Females CMRM Males CMRM Total 114 -0.70% no 114 0.45% no Internal 84 -1.37% no 94 0.36% no External 30 1.19% no 20 0.87% no ≤50 years 56 -0.94% no 46 0.06% no >50 years 55 -0.03% no 68 0.71% yes Fem ind. 44 -0.70% no 34 0.87% no Male ind. 70 -0.70% no 80 0.27% no Females MM Males MM Total 114 -0.53% no 114 0.11% no Internal 84 -1.11% no 94 0.19% no External 30 1.08% no 20 -0.25% no ≤50 years 56 -0.89% no 46 -0.42% no >50 years 55 0.09% no 68 0.48% yes Fem ind. 44 -0.40% no 34 0.34% no Male ind. 70 -0.62% no 80 0.02% no

CMRM=Constant Mean Return Model, MM=Market Model

*,**,***#significant at the 10%, 5% and 1% significance level respectively, t-test#

!,!!,!!! significant at the 10%, 5% and 1% significance level respectively,

parametric test from MacKinlay#

,✚✚,✚✚✚ significant at the 10%, 5% and 1% significance level respectively, Corrado

Rank Test #

(23)

Table 6: Average cumulative abnormal returns for the three-day event window and significance of the difference between the average cumulative abnormal returns of the female and male sample and the opposite samples

The average cumulative abnormal returns are lower for all the female subsamples than the male subsamples except for the external sample, but for none the difference is significant. There is a significant difference between the female internal sample and the female external sample according to the constant mean return model.

N CAAR N CAAR Females CMRM Males CMRM Total 114 -0.70% 114 0.45% Internal 84 -1.37%✚ 94 0.36% External 30 1.19%✚ 20 0.87% ≤50 years 56 -0.94% 46 0.06% >50 years 55 -0.03% 68 0.71% Fem ind. 44 -0.70% 34 0.87% Male ind. 70 -0.70% 80 0.27% Females MM Males MM Total 114 -0.53% 114 0.11% Internal 84 -1.11% 94 0.19% External 30 1.08% 20 -0.25% ≤50 years 56 -0.89% 46 -0.42% >50 years 55 0.09% 68 0.48% Fem ind. 44 -0.40% 34 0.34% Male ind. 70 -0.62% 80 0.02%

CMRM=Constant Mean Return Model, MM=Market Model

*,**,***#significantly differs from the other gender sample at the 10%, 5% and 1% significance level respectively#

,✚✚,✚✚✚ significantly differs from the opposite sample at the 10%, 5% and 1%

(24)

5.4 Results five-day event window

Table 7: Average cumulative abnormal returns for the five-day event window and distribution and significance of the average cumulative abnormal returns

The only average cumulative abnormal return that is significant at the 10% significance level in this event window is the female internal subsample with the constant mean return model, with an average cumulative abnormal return of -2.07%. For the one-day and three-day event window the average (cumulative) abnormal return was not significant. This difference could be explained by a relative big abnormal return of -0.62% on t=-2 (see Appendix B).

N CAAR Normal distribution N CAAR Normal distribution Females CMRM Males CMRM Total 114 -1.03% no 114 0.61% no Internal 84 -2.07%* no 94 0.45% no External 30 1.89% no 20 1.37% yes ≤50 years 56 -2.20% no 46 0.57% no >50 years 55 0.20% no 68 0.65% no Fem ind. 44 -0.52% no 34 0.84% no Male ind. 70 -1.35% no 80 0.52% no Females MM Males MM Total 114 -0.82% no 114 0.18% no Internal 84 -1.65% no 94 0.21% no External 30 1.51% no 20 0.01% no ≤50 years 56 -1.88% no 46 -0.21% no >50 years 55 0.28% no 68 0.44% no Fem ind. 44 -0.12% no 34 -0.28% no Male ind. 70 -1.25% no 80 0.37% no

CMRM=Constant Mean Return Model, MM=Market Model

*,**,***#significant at the 10%, 5% and 1% significance level respectively, t-test#

!,!!,!!! significant at the 10%, 5% and 1% significance level respectively,

parametric test from MacKinlay#

,✚✚,✚✚✚ significant at the 10%, 5% and 1% significance level respectively, Corrado

Rank Test #

(25)

Table 8: Average cumulative abnormal returns for the five-day event window and significance of the difference between the average cumulative abnormal returns of the female and male sample and the opposite samples

The average cumulative abnormal returns are higher in the male samples than the female samples except for again the external subsamples and the female oriented industry with the market model. But only for the internal subsample with the constant mean return model they differ significantly. Furthermore there is, just as for the three-day event window, a

significance difference in average cumulative abnormal returns between the female internal and external subsample, this time for both models.

N CAAR N CAAR Females CMRM Males CMRM Total 114 -1.03% 114 0.61% Internal 84 -2.07%**✚✚ 94 0.45%** External 30 1.89%✚✚ 20 1.37% ≤50 years 56 -2.20% 46 0.57% >50 years 55 0.20% 68 0.65% Fem ind. 44 -0.52% 34 0.84% Male ind. 70 -1.35% 80 0.52% Females MM Males MM Total 114 -0.82% 114 0.18% Internal 84 -1.65%✚ 94 0.21% External 30 1.51%✚ 20 0.01% ≤50 years 56 -1.88% 46 -0.21% >50 years 55 0.28% 68 0.44% Fem ind. 44 -0.12% 34 -0.28% Male ind. 70 -1.25% 80 0.37%

CMRM=Constant Mean Return Model, MM=Market Model

*,**,***#significantly differs from the other gender sample at the 10%, 5% and 1% significance level respectively#

,✚✚,✚✚✚ significantly differs from the opposite sample at the 10%, 5% and 1%

(26)

5.5 Results pre -event window

Table 9: Average cumulative abnormal returns for the pre-event window and distribution and significance of the average cumulative abnormal returns

The negative average cumulative abnormal returns are significant according to the t-test for the total sample with the constant mean return model and for the female CEOs below 50 years and the female oriented industry with both models. The t-test has value in the case of the total sample and the female oriented subsample for the constant mean return model, because the values of the Jarque-Bera test show that a normal distribution can’t be rejected.The average cumulative abnormal return for the female oriented subsample with the constant mean return model is also significant according to the Corrado rank test. The significance in this event window can be explained for the female industry sample by the average abnormal returns of -1.53% and -1.32% on t=-4, for respectively the constant mean return model and the market model (see Appendix B). These are the lowest and third lowest average abnormal returns on days in one of the event windows.

N CAAR Normal distribution N CAAR Normal distribution Females CMRM Males CMRM Total 114 -1.17%* yes 114 -0.04% no Internal 84 -1.16% yes 94 -0.34% no External 30 -1.20% no 20 1.34% yes ≤50 years 56 -1.74%** no 46 -1.22% no >50 years 55 -0.33% yes 68 0.76% no

Fem ind. 44 -2.70%***✚ yes 34 0.18% yes

Male ind. 70 -0.21% no 80 -0.14% no Females MM Males MM Total 114 -0.74% no 114 -0,02% no Internal 84 -0.75% no 94 -0.40% no External 30 -0.69% no 20 1.77% no ≤50 years 56 -1.41%* no 46 -0.94% no >50 years 55 0.15% no 68 0.61% no Fem ind. 44 -1.97%** no 34 -0.07% no Male ind. 70 0.04% no 80 0.00% no

CMRM=Constant Mean Return Model, MM=Market Model

*,**,***#significant at the 10%, 5% and 1% significance level respectively, t-test#

!,!!,!!! significant at the 10%, 5% and 1% significance level respectively,

parametric test from MacKinlay#

,✚✚,✚✚✚ significant at the 10%, 5% and 1% significance level respectively, Corrado

Rank Test #

(27)

Table 10: Average cumulative abnormal returns for the pre-event window and

significance of the difference between the average cumulative abnormal returns of the female and male sample and the opposite samples

In the male sample the average cumulative abnormal returns are higher in every subsample except the male oriented industry for the market model. Only for the female industry sample the average cumulative abnormal returns of the female and male sample differ significantly, according to the constant mean return model. Here again the low abnormal return of -1.53% on t =-4 and -0.99% on t=-3 for the female industry sample may play a role (see Appendix B).For the female sample the average cumulative abnormal returns of the female industry sample and the male industry sample differ significantly. And for the male sample this is the case for the two age samples and the internal and external samples.

N CAAR N CAAR Females CMRM Males CMRM Total 114 -1.17% 114 -0.04% Internal 84 -1.16% 94 -0.34% External 30 -1.20% 20 1.34% ≤50 years 56 -1.74% 46 -1.22%✚✚ >50 years 55 -0.33% 68 0.76%✚✚ Fem ind. 44 -2.70%**✚✚ 34 0.18%** Male ind. 70 -0.21%✚✚ 80 -0.14% Females MM Males MM Total 114 -0.74% 114 -0.02% Internal 84 -0.75% 94 -0.40%✚✚ External 30 -0.69% 20 1.77%✚✚ ≤50 years 56 -1.41% 46 -0.94%✚✚ >50 years 55 0.15% 68 0.61%✚✚ Fem ind. 44 -1.97%✚ 34 -0.07% Male ind. 70 0.04%✚ 80 0.00%

CMRM=Constant Mean Return Model, MM=Market Model

*,**,***#significantly differs from the other gender sample at the 10%, 5% and 1% significance level respectively#

,✚✚,✚✚✚ significantly differs from the opposite sample at the 10%, 5% and 1%

(28)

5.6 Results from the OLS Regression

Table 11: Coefficients form the OLS regression over the total samples

The most noteworthy from these results is the positive relationship between the (cumulative) abnormal returns and the age of the incoming CEO for the female sample. The relationship is significant for the constant mean return model in the case of a three- and five- day event window at the 1% significance level and in the case of the pre-event window at the 10% significance level. For the market model the relationship is significant at the 1% significance level for the three- and five-event window and at the 5% significance level for the pre-event window. The regression shows also a significant positive relation in four cases for the male sample.

For the five-day event window the regression shows both for the constant mean return model and the market model a significant negative relation between the cumulative abnormal returns and a new female CEO promoted from within the firm. And when it comes to the industry the regression shows a significant negative relationship for the pre-event window between the cumulative abnormal returns and the female oriented industry. For the male sample there are

Source (N=114) Age (N=111) Industry (N=114)

Females Coefficient Coefficient Coefficient

CMRM (1-day) -0.003736 0.00195 0.001143 CMRM (3-day) -0.016835 0.002529*** -0.002703 CMRM (5-day) -0.030372** 0.003593*** -0.000514 CMRM (pre) 0.001815 0.001757* -0.023732** MM(1-day) -0.004398 0.001758 0.002371 MM (3-day) -0.014316 0.002444*** -0.002256 MM (5-day) -0.022095* 0.003314*** 0.003959 MM (pre) 0.001468 0.001825** -0.01924**

Source (N=114) Age (N=114) Industry (N=114)

Males Coefficient Coefficient Coefficient

CMRM (1-day) 0.015377 0.00069 0.006477 CMRM (3-day) -0.002139 0.001696 0.005157 CMRM (5-day) -0.007052 0.00119 0.002822 CMRM (pre) -0.012763 0.002142** 0.002454 MM(1-day) 0.013158 0.00058 0.005318 MM (3-day) 0.00859 0.002212** 0.001626 MM (5-day) 0.006727 0.002194** -0.007977 MM (pre) -0.01765 0.002024** -0.001205

CMRM=Constant Mean Return Model, MM=Market Model *,**,***#significant at the 10%, 5% and 1% significance level respectively#

# #

(29)

no significant relationships between the cumulative abnormal returns and the source of the CEO or the industry in which the firm operates.

5.7 Female-to-female succession

Table 12: Average abnormal returns female-to-female succession

Although there are only 5 observations, it is interesting to see here that the average abnormal returns are quite higher in the days prior to the event but lower on the event date compared to the male-to-female succession.

6. Conclusion

In general the announcement that a female is appointed CEO does not lead to significant abnormal returns in this research. When compared with newly appointed male CEOs it seems in general that the abnormal returns are lower when there is an announcement of a female CEO appointment than when there is an announcement of a male CEO appointment and thus that the market reacts negatively to this news. But I found only in a few subsamples statistical evidence for this and so I conclude that there is no difference between the stock returns after the announcement of a new female CEO and the stock returns after the announcement of a new male CEO. This is in line with the findings of Martin et al. (2009) and Ola and Proffitt (2015).

But there is some statistical evidence for different stock returns between the subsamples after the announcement that a female will become the new CEO. Firstly, there are some significant results about the source of the new female CEO. The abnormal returns are for almost every event window lower, in a few cases significant, when the new female CEO works already at

t= CMRM (N=5) MM (N=5) -4 2.91% 3.07% -3 0.86% 1.06% -2 2.16% 2.88% -1 0.34% -0.02% 0 -1.43% -0.87% 1 1.09% 0.69% 2 -0.22% -0.63%

(30)

the firm than when she is recruited from the outside. This is in line with the findings of Charitou et al. (2010) about CEO successions in general. But it is in contrast with my

hypothesis and the findings of Lee and James (2003). The lower stock returns for new internal CEOs could be explained if the company is performing poorly prior to the succession, as Dherment-Ferere and Renneboog (2000) find, but that information is not included in this study.

Secondly, the age of the incoming CEO might be relevant. The stocks in the subsample where the CEO is 50 years or younger have on average lower returns than the stocks in the

subsample where the CEO is older than 50 in every event window and the average abnormal returns differ significantly on the event date. And the regression showed in almost all cases a significant positive relationship between the abnormal returns and the age of the new CEO. So it seems that investors prefer an older CEO, possibly due to more experience or a bigger risk-aversion that comes with older age as Rolison et al. (2014) and Serfling (2014) argue.

Thirdly, I found also some evidence that it matters in which industry the firm operates when it comes to the pre-event window, which might be interesting if we assume that the information of a CEO appointment is already available to investors slightly before the official

announcement. In the pre-event window we found significantly lower stock returns when the announcement of a new female CEO occurred in an industry in which there are more women active than the American average of 46.8% than when it occurred in an industry with a female labor force below 46.8%. And the regression also showed this significant negative

relationship between the cumulative abnormal returns and the female oriented industry. This is in contrast with the hypothesis and with the findings of Cook and Glass (2011). They find a bigger increase in the share price when a female gets a senior leadership position in an

industry where more than 50% of the labour force is female.

The average abnormal returns are quite higher for the female-to-female succession sample than the male-to-female succession sample in the days prior to the announcement. This is in line with the findings of Zhang and Qu (2015) that a gender change in a CEO turnover is a disrupting factor and leads to weaker firm performance. But off course we have to bear in mind that the female-to-female succession only occurred 5 times. But it might be an

(31)

For the matching with the male CEO appointments I used the announcement date. The results could have been more precise if the other characteristics where also used for the matching. In that case the subsample sizes of the females and males would have been equal.

When it comes to the statistical tests there is a limitation for the independent t-tests I used to see if the abnormal returns were significantly different between the female and the male sample and between the opposite samples. These t-tests assume a normal distribution and this was not the case in most of my samples. But, in almost all cases the variances were unequal and then there is not a good alternative non-parametric test.

My sample only contained the announcement of female CEOs during the years 2002-2015. An interesting follow-up study would be in some time when there is probably more data available to check if there has been a change in the reaction of the market over time. It could be possible that the investors have a different reaction to the appointment of a female CEO when it becomes more common, which I expect to happen in the future.

(32)

References

Adams, J.C., Mansi, S.A. (2009) CEO turnover and bondholder wealth. Journal of Banking &

Finance, 33(3), 522-533

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

Arena, C., Cirillo, A., Mussolino, D., Pulcinelli, I., Saggese, S., Sarto, F. (2015) Women on board: Evidence from a masculine industry. Corporate Governance, 15(3), 339-356

Bear, S., Rahman, N., Post, C. (2010) The Impact of Board Diversity and Gender

Composition on Corporate Social Responsibility and Firm Reputation. Journal of Business

Ethics, 97(2), 207-221

Beckmann, D., Menkhoff, L. (2008) Will Women be Women? Analyzing the Gender Difference Among Financial Experts. Kyklos, 61(3), 364-384

Brown, S.J., Warner, J.B. (1985) Using Daily Stock Returns: The Case of Event Studies.

Journal of Financial Economics, 14, 3-31

Bureau of Labor Statistics (2015) Labor Force Statistics from the Current Population Survey. Retrieved from http://www.bls.gov/cps/cpsaat18.htm

Campbell, K., Minguez-Vera, A. (2008) Gender Diversity in the Boardroom and Firm Financial Performance. Journal of Business Ethics, 83(3), 435-451

Carter, D.A., Simkins, B.J., Simpson, W.G. (2003) Corporate Governance, Board Diversity and Firm Value. Financial Review, 38(1), 33-53

(33)

Charitou, M., Patis, A., Vlittis, A. (2010) The Market Reaction to the Appointment of an Outside CEO: An Empirical Investigation. Journal of Economics and International Finance, 2(11), 272-277

Cook, A., Glass, C. (2011) Leadership change and shareholder value: How markets react to the appointments of women. Human Resource Management, 50(4), 501-519

Corrado, C. (1989) A nonparametric test for abnormal security-price performance in event studies. Journal of Financial Economics, 23(2): 385-395

Datta, D.K., Guthrie, J.P. (1994) Executive Succession: Organizational Antecedents of CEO Characteristics. Strategic Management Journal, 15, 569-577

Dezsö, C.L., Gaddis Ross, D. (2012) Does female representation in top management improve firm performance? A panel data investigation. Strategic Management Journal, 33(9), 1072-1089

Dezso, C.L., Gaddis Ross, D., Uribe, J. (2016) Is there an implicit quota on women in top management? A large-sample statistical analysis. Strategic Management Journal, 37(1), 98-115

Dherment-Ferere, I., Renneboog, L.D.R. (2000) Share Price Reactions to CEO Resignations and Large Shareholder Monitoring in Listed French Companies. CentER Discussion Paper

Fama, E.F., (1976) Foundations of finance. Basic Books, New York

Francis, B., Hasan, I., Park, J.C., Wu, Q. (2015) Gender Differences in Financial Reporting Decision Making: Evidence from Accounting Conservatism. Contemporary Accounting

Research, 32(3), 1285-1318

(34)

Haslam, S.A., Ryan, M.K., Kulich, C., Trojanowski, G., Atkins, C. (2010) Investing with Prejudice: The Relationship between Women’s Presence on Company Boards and Objective and Subjective Measures of Company Performance. British Journal of Management, 21(2), 484-497

Hewlett, S.A. (2002) Executive Women and the Myth of Having It All. Harvard Business

Review, 80, 66-73

Ho, S.S.M., Li, A.Y., Tam, K., Zhang, F. (2015) CEO Gender, Ethical Leadership, and Accounting Conservatism. Journal of Business Ethics, 127(2),351-370

Hutzschenreuter, T., Kleindienst, I., Greger, C. (2015) What Determines Demographic Similarity Between Incumbent CEOs and Their Successors? A CEO Informal Power Perspective. Managerial and Decision Economics, 36(7), 421-438

Khan, W.A., Paulo Vieito, J. (2013) Ceo gender and firm performance. Journal of Economics

and Business, 67, 55-66

Kim, H., Shim, S. (2003) Gender-Based Approach to the Understanding of Leadership Roles among Retail Managers. Human Resource Development Quarterly, 13(3), 321-342

Krantz, M. (2015) Facebook CEO leads list of U.S.’ youngest. Retrieved from

http://americasmarkets.usatoday.com/2015/01/27/zuckerberg-leads-list-of-u-s-youngest-ceos/

Lauterbach, B., Vu, J., Weisberg, J. (1999) Internal vs. External Successions and Their Effect on Firm Performance. Human Relations, 52(12), 1485-1504

Lee, P.M., James, E.H. (2003) She’-e-os: Gender Effects and Stock Price Reactions to the Announcements of Top Executive Appointments. Darden Business School Working Paper (2-11)

MacKinlay, A.C. (1997) Event Studies in Economics and Finance. Journal of Economic

(35)

Martin, A.D., Nishikawa, T., Williams, M.A. (2009) CEO Gender: Effects on Valuation and Risk. Quarterly Journal of Finance and Accounting, 48(3), 23-40

McKinsey (2007) Women Matter: Gender Diversity, a Corporate Performance Driver. Retrieved from http://www.raeng.org.uk/publications/other/women-matter-oct-2007

Ola, J.C., Proffitt, D. (2015) The Stock Market Response to CEO changes: Does Gender Matter? International Journal of Business and Management, 10(5), 1-9

Peck, E. (2015, July 13) Do You Realize Ho Few Women CEOs exist? These Executives Don’t. The Huffington Post, Retrieved from

http://www.huffingtonpost.com/2015/07/13/weber-shandwick-female-ceo_n_7771608.html Peni, E. (2012) CEO and Chairperson characteristics and firm performance. Journal of

Management & Governance, 18(1), 185-205

Peyrache, E., Palomino, F. (2013) Internal versus External CEO choice and the structure of compensation contracts. Journal of Financial and Quantitative Analysis, 48(4): 1301-1331

Post, C. (2015) When is female leadership an advantage? Coordination requirements, team cohesion, and team interaction norms. Journal of Organizational Behavior, 36(8), 1153-1175

Quinn, R.E., (1990) Becoming a master manager: a competency framework. Wiley, New York

Rolison, J.J., Hanoch, Y., Wood, S., Liu, P.J. (2014) Risk-taking differences across the adult life span: a question of age and domain. The Journals of Gerontology Series B: Psychological

Sciences and Social Sciences, 69(6), 870-880

Rose, C. (2007) Does female board representation influence firm performance? The Danish evidence. Corporate Governance, 15 (2), 404–13.

Ryan, M.K., Haslam, S.A. (2005) The Glass Cliff: Evidence that Women are

(36)

Serfling, M.A. (2014) CEO age and the riskiness of corporate policies. Journal of Corporate

Finance, 25: 251-273

Todaro, W. (2003) Want To Be a CEO? Stay Put. Retrieved from http://www.forbes.com/2003/03/31/cx_wt_0401exec.html

Webber, R.A. (1987) Changes in Perception and Behavior in Mixed Gender Teams. Human

Resource Management, 26(4), 455-467

White, H. (1980) A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity. Econometrica, 48(4), 817-838

(37)

Appendix A: Average abnormal returns of the female sample

Table A1: Average abnormal returns of the female sample

! t= CMRM (N=114) MM (N=114) -10 -0.55% -0.33% -9 -0.16% -0.15% -8 0.07% -0.07% -7 -0.82% -0.85% -6 0.35% 0.55% -5 0.60% 0.58% -4 -0.30% -0.10% -3 -0.34% -0.19% -2 -0.42% -0.36% -1 -0.11% -0.08% 0 -0.24% -0.17% 1 -0.35% -0.27% 2 0.09% 0.08% 3 -0.12% -0.10% 4 -0.59% -0.56% 5 0.77% 0.82% 6 0.51% 0.48% 7 -0.34% -0.27% 8 0.78% 0.77% 9 0.07% -0.14% 10 -0.43% -0.31%

(38)

Appendix B: Average abnormal returns of the subsamples

Table B1.1: Average abnormal returns of the female subsamples

Table B1.2: Average abnormal returns of the male subsamples

Females AAR

N t= -4 t= -3 t= -2 t= -1 t= 0 t= 1 t= 2

Constant Mean Return Model

Total 114 -0.30% -0.34% -0.42% -0.11% -0.24% -0.35% 0.09% Internal 84 0.09% -0.20% -0.62% -0.43% -0.49% -0.46% -0.08% External 30 -1.39% -0.74% 0.13% 0.79% 0.46% -0.06% 0.57% ≤50 56 -0.30% -0.13% -0.81% -0.51% -0.95% 0.52% -0.46% >50 55 -0.20% -0.39% -0.16% 0.42% 0.54% -0.99% 0.40% Fem ind. 44 -1.53% -0.99% -0.37% 0.19% -0.08% -0.80% 0.55% Male ind. 70 0.47% 0.07% -0.45% -0.29% -0.34% -0.07% -0.20% Market Model Total 114 -0.10% -0.19% -0.36% -0.08% -0.17% -0.27% 0.08% Internal 84 0.28% -0.08% -0.54% -0.41% -0.42% -0.28% 0.01% External 30 -1.16% -0.52% 0.16% 0.83% 0.51% -0.26% 0.27% ≤50 56 -0.20% -0.02% -0.73% -0.46% -0.85% 0.41% -0.25% >50 55 0.08% -0.23% -0.03% 0.34% 0.48% -0.73% 0.22% Fem ind. 44 -1.32% -0.55% -0.38% 0.28% 0.12% -0.79% 0.65% Male ind. 70 0.66% 0.03% -0.35% -0.31% -0.36% 0.05% -0.29% ! Males AAR N t= -4 t= -3 t= -2 t= -1 t= 0 t= 1 t= 2

Constant Mean Return Model

Referenties

GERELATEERDE DOCUMENTEN

Maximal and submaximal aerobic tests for wheelchair- dependent persons with spinal cord injury: A systematic review to summarize and identify useful applications for

Taking into account possible effects of type of organization, it was hypothesized that non-profit organizations would obtain significantly higher credibility scores when applying an

In our view a decision support system should also challenge the SC professional to evaluate his decisions; to reflect on his choices, the uncertainties

The reformulation as a Mealy Machine can be done in di fferent ways, in particular, the higher order functions present in the Haskell definitions may be executed over space or

(2011) European risk factors’ model to predict hospitalization of premature infants born 33–35 weeks’ gestational age with respiratory syncytial virus: validation with Italian

For all heating sections active, the heat pipe oriented vertically in an evaporator- down mode and a power input of 150 W, the overall thermal resistance was 0.014 K/W at a

ii. the links between meaning, goals /purposes, positive relational processes and other facets of psyschosocial well-being, bearing in mind some socio- demographic and

irn.tl die volA... nio Voldoonde