Women in precarious top positions: examining the glass cliff-phenomenon in The Netherlands

(1)

University of Groningen

Faculty of Economics and Business

Master of Science – International Business and Management

Master Thesis

Women in precarious top positions: examining the glass

cliff-phenomenon in The Netherlands

Bertine Riezebos

H.Riezebos@student.rug.nl

s1789554

December 2012

(2)

2

Women in precarious top positions: examining

the glass cliff-phenomenon in The Netherlands

This study builds on previous researches on the glass cliff-phenomenon, primarily on the work by Ryan and Haslam (2005/2008/2009) and Adams, Gupta and Leeth (2009). The glass cliff-phenomenon holds that women more so than men are appointed to precarious leadership positions, associated with risk of failure and criticism. This will be tested in the context of 64 of the largest companies in The Netherlands by looking at the number of women in executive and supervisory boards, the positions females hold in these boards, and the industries the companies are in.

The present study will show that the glass cliff does not hold in The Netherlands over the years 2006-2011, based on the available data. These years were chosen since companies were struggling with the economic crisis and thus with declining company performance and uncertain situations. Findings indicate that the number of females in the supervisory boards of companies did change significantly, but binary logistic regression analysis showed that these women were appointed in companies in good financial condition rather than in poorly performing companies. The number of females in the executive board only grew with three women, thereby contradicting the glass cliff-phenomenon. The findings correspond with the findings by Adams et al. rather than those by Ryan and Haslam.

(3)

3

1. Introduction

This study draws on research in the field of women in top positions. Specifically, this study looks at a recently defined phenomenon, called the ‘glass cliff’. The glass cliff is defined as “the

phenomenon whereby women are more likely than men to be appointed to leadership positions associated with increased risk of failure and criticism because these positions are more likely to involve management of organizational units that are in crisis” (Haslam and Ryan, 2008, p. 530). The phenomenon was established as a sequel to the ‘glass ceiling’, developed by Bass and Aviolo (1994), which is described as a barrier that prevents women’s advancement into the executive ranks of organisations. Literature has suggested that the glass ceiling might be broken, or at least cracked, lately (e.g. Meyerson and Fletcher, 2000). Nonetheless, other researchers notice a new development in the field of women in top positions. To be more precise, they claim that when women eventually do get the opportunity to advance to top positions, they usually find themselves in positions that involve risk and uncertainty (Judge, 2003; Haslam and Ryan, 2008), or they find themselves leading an organisational unit that is in crisis (Ryan and Haslam, 2005).

However, the studies that have been conducted to research the glass cliff-phenomenon have quite some drawbacks, such as limited sample size, bias due to experimental designs, and drawbacks associated with using a matched samples design. This study will investigate the phenomenon again, thereby looking at the specific positions women can hold. As stated before, the glass cliff-phenomenon posits that women will be appointed to top positions sooner when the company is facing risk, uncertainty and crisis, as well as declining performance. For this reason, the period 2006-2011 is taken, since this involves the global economic crisis and therefore definitely shows declining company performance for most companies in the sample. The total sample consists of 64 of the largest companies of The Netherlands. This study will research several elements. Firstly, it will be tested whether more women indeed got appointed to the executive and supervisory boards since company performance declined. Secondly, analyses will show if women will most likely fill the CEO and CFO positions, rather than the other positions in the executive board, and which committees they mostly join in the supervisory board. Lastly, the present study will look at differences in top positions across industries considering that, if the glass cliff-phenomenon proves to be true, women should be appointed to the industries that are struggling most.

1.1 Present study

(6)

6 was Belgium with a disappointingly 8 per cent of top executives being female. However, this was nearly 20 years ago. What is the situation at the moment in The Netherlands? According to a study conducted by the ‘Sociaal en Cultureel Planbureau’ (SCP; Social and Cultural Planning Office) together with the ‘Centraal Bureau voor de Statistiek’ (CBS; Statistics Netherlands), the most striking

development has been that, within the largest companies in The Netherlands, the percentage of women in top executive positions has increased tremendously during the past few years, despite or maybe (in line with the present study) due to the current economic crisis, as can be seen in table 1. The biggest increase could be identified in the board of directors of the largest companies, although the notion has to be made that the percentage of women at the top was nearly zero a few years ago, thus, it is not too odd to find a big increase there. The study, called ‘Emancipatiemonitor 2010’, showed that in 2007

none of the 25 largest companies in The Netherlands had any women in its executive boards, whereas this has grown to 5,6 per cent women in 2010. Of course, this still is a poor result and hardly worth applauding. In the top 100 companies in The Netherlands, the result is similar, with 5,3 per cent of senior executives being female. All these numbers show that women are still very much underrepresented in top positions of the largest companies in the country. What is the reason for this? And will there be certain situations in which women get appointed to top positions sooner than in different situations? This study will look at these questions, specifically focusing on a claim developed by Ryan and Haslam (2005): women get appointed to top positions sooner when the firm or organisational unit is facing uncertainty and risk, as well as declining performance. This situation has been coined the glass cliff-phenomenon.

Several studies claim that the glass cliff-phenomenon is very much existent in today’s business community. However, most of these studies have focused on experiments or laboratory settings (e.g.

Ryan and Haslam, 2003; Bruckmüller and Branscombe, 2010). The ones that actually did conduct research on real-life companies (e.g. Ryan and Haslam, 2005) focused on other countries than The Netherlands or only focused on a limited sample. Suggestions for future research include looking at several leadership positions in order to find out whether or not more females have reached these positions (Ryan and Haslam, 2009; Adams, Gupta and Leeth, 2009). This study attempts to add to existing literature by focusing on the glass cliff-phenomenon with regard to these positions, both in the executive board and in the supervisory board of companies. Moreover, Adams et al.’s research claims

(7)

7 Table 1: Proportion of women in executive and supervisory boards. Source: www.cpb.nl

After analysing the glass ceiling-phenomenon and the glass cliff-phenomenon, this study will try to unravel whether the glass cliff-phenomenon exists in The Netherlands. My aim is to examine if, and if so how, the glass cliff-phenomenon is evident in this country.

2. Theoretical background

2.1 How did the glass cliff-phenomenon come about?

(8)

8 that time (Zelechowski and Bilimoria, 2004). According to Meyerson and Fletcher (2000), however, the new millennium would provide the possibility to celebrate the progress that had been made by women in that time. These authors claim that women by then were holding seats on corporate boards, thereby running major companies. This could have a lot to do with the fact that women’s aspirations to grow to top positions were increasing during the last decades of the 20th century (Gerstein, Lichtman and Barokas, 1988). The present study uses the definition of Dreher (2003, pp. 544-545) to explain top positions as those in the “ranks of corporate officers (i.e. CEOs, COOs, operating company presidents, executive vice presidents, CFOs, and corporate vice presidents in charge of key corporate positional areas like research and development)”. Despite the progress made during the last decades of the 1900s,

Meyerson and Fletcher (2000) still needed to acknowledge the fact that women continued to be very underrepresented in the higher levels of the workforce. Not more than 10% of senior managers in the

Fortune 500 companies were female at the time. So, as the authors proclaimed, the glass ceiling was

still very much existent during the year 2000. As a matter of fact the authors showed that, in 2010, still only 3% of Fortune 500 CEOs were female, although the percentage of women holding top positions (board positions) in those companies had grown to around 18% (www.economist.com; see figure 1).

Figure 1: Percentage of women in business in the US. Source: www.economist.com.

2.1.1 The glass ceiling

To describe the phenomenon that women were not able to advance to senior positions, the term ‘glass ceiling’ has been coined many years ago. When looking at female career paths, this is a term that cannot be ignored. It describes the phenomenon whereby women experience barriers that prevent them from advancing to senior management positions, while men do not experience such barriers. It can be seen as a transparent barrier that disables women to climb further on the corporate ladder.

(9)

9 management level, that will prevent women from rising to those positions, such as executive positions in the board of directors.

By using already existing literature, Dreher (2003) presented three problems arising from this glass ceiling-phenomenon. Firstly, when female managers in lower levels of the companies perceive a gender obstacle in their chances of reaching senior management positions, their motivation and desire to reach this level can evaporate. Women could feel that hard work and perseverance will not pay off in the long run, which will have a negative effect on productivity in all levels of the company (Vroom, 1964). Secondly, a rather big concern is that the glass ceiling creates a uniform character in the board rooms of large companies, by only welcoming men in executive positions. According to Elsass and Graves (1997) and Janis (1982), both in Dreher (2003), this uniformity will lead to too much homogeneity. The third and final reason Dreher highlights is that, in times of tight labour market conditions, gender-based barriers can further reduce the supply of needed talent and resources. According to the resource-dependence theory, it is to a firm’s advantage to eliminate barriers and, in some cases, to focus on moving women into key management positions because they may match better with certain customer segments (Jacobs, 1992). Also, by diversifying a board in terms of gender or age, a firm signals drastic change to its stake- and shareholders, which might be required in times of crises (Haslam and Ryan, 2008).

Even though the problems associated with the glass ceiling-phenomenon are big, they are still surmountable, as also Tung (2004) suggests. Dreher’s research (2003) concludes that the number of women in lower-level managerial positions is positively related to the number of women in senior level positions.

(10)

10 do not all companies introduce more women to their boards of directors? For this, we need to find out more about the barriers that were preventing women to advance on the corporate ladder.

Even though Adler and Izraeli (1994) early on suggested that companies should be seeking ways to promote women to top positions in order to maximise the effectiveness of their human capital resources, Oakley (2000) found that many corporations still did not actually hire more women for these positions. Oakley conducted research on the factors that prevent women from growing to top positions. She claims that, in all parts of the world, it is very rare to find a female senior executive, and especially a Chief Executive Officer (CEO). She identified two categories of barriers that prevented women from advancing to higher levels in the organisation. The first category, barriers created by

corporate practices, includes barriers stemming from objective with regard to recruitment, retention

and promotion, whereas the second category, behavioural and cultural causes, revolves around issues of stereotyping, tokenism, power, preferred leadership styles and the psychodynamics of male/female relations. Since the latter category is not very relevant for the present study, as the glass cliff-phenomenon is more based on corporate practices, I leave out the explanation of this category.

With regard to the first category, Oakley identified that very few women in upper management have actual line experience, which is a prerequisite for the CEO position in most instances (Oakley, 2000). Rather, these women have experience in staff support areas such as human resources or public relations, which are often seen as less-important positions. Also, corporations do not focus on creating diversity initiatives or policies that weaken the barriers for advancing to senior positions.

2.1.2 Women’s growing aspirations

Luckily, for women with ambition at least, the glass ceiling seems to have cracked since the early 2000s (e.g. Dreher, 2003; Altman and Shortland, 2008). Altman, Simpson, Baruch and Burke (2005)

have found that women belonging to “Generation X” (people born between 1960 and 1980) are climbing up the metaphorical ladder in the organisation, and that they are more ambitious than the baby boomers have ever been. Also, research has shown that women were increasingly starting to express similar aspirations to men when it comes to advancing to senior level positions (Catalyst, 2001). For example, in The Netherlands, management as an occupation has grown tremendously since the ’70s and this has given more opportunities to women to advance to managerial levels, although they remain underrepresented and are employed more in lower- and middle-management levels (Wirth, 2001). Wirth also claims, basing her conclusions on the International Labour Organization’s (ILO) SEGREGAT database, that women in The Netherlands took advantage of these growing amount of opportunities by increasing their participation in senior levels from 10 per cent in 1979 to 18 per cent in 1990, which is a remarkable increase; especially when comparing it to the percentages in other countries. A recent study, published in 2012, of the ‘Sociaal en Cultureel Planbureau’ (Social and

(11)

11 positions) supplements the ILO by showing that the number of women in top positions has certainly increased during the last couple of years, but that this increase still does not happen rapidly, especially in the business community, which one might expect since the government passed a bill stating that companies should have 30% women in their boards by 2016 (Staatsblad, 2011). Even though it is a remarkable breakthrough in business for women that they are increasingly taking on the responsibility of top positions, it appears that the glass ceiling has not been entirely broken yet, but has rather been cracked.

2.2 The glass cliff developed by several authors

Several authors identified a new development, a new barrier behind the glass ceiling. Since the early 2000s, various studies have been conducted to examine this development, which has been named glass cliff-phenomenon.

2.2.1 Judge (2003)

The research by Judge (2003) is probably the research with the least credibility. Her study has only been published in The Times, rather than in a noteworthy business journal, and her study has quite some drawbacks, which will be discussed later in this paper. The study by Judge focused on the Financial Times Stock Exchange Index (FTSE 100) and researched the relationship between company performance and the number of women in the boards of directors. For this, she uses an index compiled by the Cranfield School of Management (Singh and Vinnicombe, 2003), which ranks the FTSE 100 companies in relation to the percentage of women on their boards. By looking at share price performance her analysis shows that companies that decline to embrace women in the board for political correctness and legitimacy purposes actually perform better than those companies that promote sexual equality. The way Judge analysed this is by looking at the performance of the top five companies (i.e., those with the highest percentage of women in their boards) and the bottom five companies (i.e., those with the lowest percentage of women in their boards) from the FTSE 100.

(12)

12 Even though the research by Judge is not necessarily very scientific, it was the research that lay the basis for the development of the glass cliff-phenomenon.

2.2.2 Ryan and Haslam (2004/2005)

Ryan and Haslam (2004) responded to the research of Judge (2003) through an article on the BBC website (www.bbc.co.uk) in which they give a little away of their future research. They believed that if the relationship existed between company performance and women in boards of directors, an alternative explanation could also be valid, one that involved reversing the causal sequencing (Haslam and McGarty, 2003). They believed it would be equally plausible that a company’s poor performance

could be a trigger for the appointment of women to the board. If this were to be the case, women might be preferentially placed in positions involving increased risk of negative consequences. These positions may, as a result, be more precarious than those that male colleagues hold. Their article on the BBC website was, therefore, the first to coin the term ‘glass cliff’, which they then describe as women’s promotion into “risky, difficult jobs where the chances of failure are higher”. Ryan and Haslam claim that Judge’s article conveyed a pessimistic message about the contributions of women in the contemporary work environment. Their upcoming research would show that such pessimism may have been premature, that women leaders would be appointed when company performance was low, rather than their appointment leading to poor company performance.

In the study they published in the British Journal of Management, Ryan and Haslam (2005)

conducted an extensive archival study of the performance of the FTSE 100 companies both before and after the appointment of a male or female board member. Their analysis found that there was a much more complex story than Judge had suggested in the Times. They state that a glass cliff is a dangerous place to be. Companies that are consistently underperforming compared to before, tend to attract attention to themselves and their boards. In such circumstances, the company’s leaders are likely to be blamed rather than other factors that may have an effect on company performance. Women who take on those positions may be exposed to criticism more so than men in the same position, according to

Ryan and Haslam. They are also more likely to be held accountable for negative outcomes that were already in motion before they took on their new responsibilities.

(13)

13 with 19 female board appointments in 17 companies. Their control group consisted of another 19 companies, all who had appointed a male board member in 2003. As far as possible, the samples had been matched in terms of when the appointment was done and in which business sector the company finds itself in order to control for errors. They believed that by using a matched samples design, they could identify the exact difference between how company performance developed before a man was appointed and how it developed before a woman was appointed.

Two measures for company performance were used: (1) annual company performance based on the London Stock Exchange Share Monitoring Service and (2) average monthly share price. The first measure was calculated as the percentage movement of share price over the 12 months preceding December 2003, whereas the second measure was the average monthly share price, calculated for the six months before and after the appointment of the board member.

Consistent with Judge’s conclusion, correlational analysis of Ryan and Haslam showed that there was a significant negative correlation between the percentage of women in board positions and performance as measured by change in share price. The annual performance of the company, however, did not differ depending on whether a male or female had been appointed. Nonetheless, companies that appointed men showed a significant increase in share price in the 12 months until December 2003, while companies that appointed women did not show such a significant increase, although this does not shed light on the causal relationship.

In order to find fluctuations of share price before and after the appointment of men and women, the average monthly share price was chosen as an indicator to establish the correct causal relationship. In total, data were available for 15 companies that appointed a female, and 16 companies that appointed a male for five months preceding the appointment and three months after the appointment. Data analysis showed that, on average, companies that appointed male board members showed a relatively stable performance over time, whereas those that appointed women showed a different picture: when a women was appointed in the first half of the year (when the stock market was down), company performance showed a significant increase over time. As Ryan and Haslam (2005, p. 85)

state: “Between five and two months prior to the appointment of a woman, these companies experienced very low share price. Thereafter, however, company performance increased significantly.” Those companies that appointed a female in the second half of the year (when the stock market was up), had a rather stable company performance over time.

Ryan and Haslam’s research concludes that, contrary to Judge’s conclusion, the appointment of

(14)

14 member. This leads the authors to conclude that women are more likely to be placed in positions of leadership when there are times of general financial downturn and downturn in company performance. This creates a situation in which women are likely to be placed on a ‘glass cliff’, “in the sense that their leadership appointments are made in problematic organizational circumstances and hence are more precarious.” (Ryan and Haslam, 2003, p. 87).

2.2.3. Ryan and Haslam (2007) – reasons why the glass cliff exists

Ryan and Haslam (2007) wanted to extend their research from establishing that the phenomenon exists to answering questions of why it is that women are more likely than men to be placed in riskier leadership positions. The authors base these reasons on implicit theories of leadership in which women are seen as better suited than men to manage crisis situations. These theories reflect beliefs “(1) that women are best equipped to deal with the socio-emotional challenges that (potential) crises present, (2) that men are not suited to these challenges, (3) that men are best equipped to deal with the task demands of success [rather than failure], or (4) that women are “not up to” such tasks (Eagly and Johnson, 1990; Schein, 1973)” (Ryan and Haslam, 2007, p. 557). Nonetheless, the authors also identify social psychological and social structural factors that might have an influence on the existence of the glass cliff-phenomenon. They come up with four different reasons:

1. Hostile and benevolent sexism: one reason could be that women are appointed to risky and

precarious positions because sexist men actively strive to disadvantage women and have a desire to see them fail (hostile sexism). On the other hand, women can be assigned to such positions by men who believe they do these women a ‘favour’ by presenting them “challenging” positions, that are actually problematic (benevolent sexism).

2. Group dynamics and ingroup favouritism: glass cliffs might also be an expression of

intergroup discrimination, such that male decision makers focus on ingroup favouritism by reserving the more attractive (non-precarious) positions for fellow ingroup members (Balls, 1992; Monkturner, 1992; Gallagher, 1994; Powell and Butterfield, 2002). Outgroup members are left to occupy those positions that the ingroup does not want.

3. Perceived quality of leadership positions: research by Ashby, Ryan and Haslam (2007)

shows that subtle measures of desirability of positions indicate that people are less likely to recommend risky leadership positions than non-risky ones to friends. Ryan and Haslam (2007)

believe that this may result in less men that find themselves on glass cliffs than women, because “their genderbased ingroup directs them away from such positions and toward more secure career opportunities” (p. 559).

4. Signalling change: historically, managerial roles were linked to male characteristics.

(15)

15 feelings of others. This leads to a new stereotype: “think crisis-think female”. The results of the research by the authors showed that when the context of leadership was predominantly male, the glass cliff-phenomenon did indeed emerge. For companies with a successful history of male leadership, participants overall chose a male candidate, whereas they tended to choose a female candidate when the company with a successful history of male leadership was in a serious crisis. When the history of leadership had been predominantly female, no such relationship occurred. Although it is nice to believe that women are appointed to precarious leadership positions because their capabilities match the tasks they can expect when facing these positions (“think crisis – think female” developed by Ryan, Haslam, Hersby and Bongiorno, 2007), it may well be that they find themselves in these positions because of the mere reason that they are not men. Failure may indicate that the traditional think manager-think male approach is not working and that change is required. Alternatively, the board of directors may feel that the appointment of a woman will attract favourable attention, showing to internal and external audiences that the organisation is embracing change. Bruckmüller and Branscombesupport this argument by claiming that the glass cliff-phenomenon exists because of the status quo-bias. The authors believe that as long as an organisation performs well, it sees no need to change the status quo. Therefore, if there has been a long tradition of men in management positions, this would not need changing. However, as soon as the organisation experiences a crisis or uncertain times, the organisation might see that there should be something changing and if it perceives the male manager to be responsible for the crisis, it might select a women to take on the position more quickly.

2.2.4 Haslam and Ryan (2008)

Haslam and Ryan (2008) felt the need for supplementing their archival research with controlled experimental research. They claim that, while high in external validity, their 2005 research did not answer all questions they had. There are three reasons why they felt that their research needed additional experimental research. First, they wanted to establish whether the glass cliff-phenomenon is robust, one that can be generalised to other contexts. Second, experimental research can clarify the causal relationship between gender and company performance (Haslam and McGarty, 2004). Third, experimental research can investigate the psychological processes related to the phenomenon. The 2008 research was conducted in order to gather evidence that the glass cliff indeed exists by way of a controlled experimental setting. The second aim was to provide an initial understanding of the psychological factors that might contribute to this.

(16)

16 contribute to the appointment of women to glass cliff positions. In their studies, “a risky leadership position was defined as a role on an executive board of directors in a company with consistently declining performance (and hence an increased risk of failure), as opposed to a role in a company with continuing success (a position that was more safe)” (Haslam and Ryan, 2008, p. 533). In the first study, 95 graduate students were the participants, who all had to analyse one of two job advertisements. The first was an advertisement for a company in a female-dominated industry, while the second was one for a company in a male-dominated industry. Participants received information on the financial strength of the company, after which they received information and photographs of three potential candidates for the job opening, two (one man and one woman) of which were clearly suitable for the job. A third candidate (male) was less suitable, but was included to enhance credibility. By indicating their level of agreement with six different statements, participants could evaluate the candidates. Participants, overall, ranked the female candidate higher than the male candidate when company performance was declining rather than improving.

Participants in studies 2 and 3 were high-school students and businessmen and –women, respectively. The approach to these two studies was similar to study 1, as job vacancies were analysed considering men and women for the specific positions. Both studies yielded the same outcomes as study 1: female candidates were more likely to be selected when the festival/company’s performance was declining, whereas men were chosen for the position when company performance was increasing. For Haslam and Ryan (2008) these studies proved, once again, that their claim is true: a glass cliff definitely exists for women.

2.2.5 Adams, Gupta and Leeth (2009)

What Adams, Gupta and Leeth (2009, p. 2) wanted to do with their paper was “to provide an additional test of the Ryan and Haslam hypothesis that female executives are over-represented in precarious leadership positions.” Their paper thus is a direct response to the studies by Ryan and Haslam (2005, 2008), although it uses companies from the US rather than the UK. Moreover, although the paper by Ryan and Haslam uses ‘leadership positions’ in general to investigate the glass cliff-phenomenon, Adams et al. take into account the CEO position specifically. They conclude that looking at the CEO position helps to control for issues such as differences in motivation and ability, because people who reach the top of an organisation are more likely to be homogeneous in terms of “motivation, ability, and political savvy” (p. 2). Also, as the CEO is the most powerful and visible member, the CEO position seems a reasonable setting to establish if female executives are over-represented in firms that find themselves in precarious situations, according to the authors.

(17)

17 61 female CEOs. Using stock return data from the Center for Research in Security Prices (CRSP), the authors were able to use 48 firms in the sample of which all required data was available. Data shows that the number of female CEOs slightly increases over the time frame, and that this pace is slightly higher from 2000 onwards. The industries in which women are mostly appointed as female CEOs are business services, apparel and accessory stores, miscellaneous retail, and chemicals and allied products. For each firm, Adams et al. identify the data on when each female became CEO and obtain daily returns for each firm for a period including 250 trading days before the appointment and 100 trading days following the appointment. The reason the authors present for using daily stock returns as an indicator of firm financial health is that in reasonably efficient securities markets the price of shares incorporates a lot of performance-related information.

As a control, the authors use comparison samples of firms that appointed male CEOs over the years 1992-2004. Findings indicate that females (males) are appointed as CEO in industries that have a larger proportion of female (male) workers. What is more, the results of the study by Adams et al.

indicate that women are appointed as CEO in firms that are doing better than firms that appoint male CEOs, a result that is very different from the conclusion by Ryan and Haslam (2005). Also, company performance after the appointment does not differ significantly between firms that appointed female CEOs and those that appointed male CEOs. Thus, the authors can conclude that their research does not find evidence for the claim that women are appointed more often to leadership positions when the firm is in precarious financial health. In order to control for several influences that could have an impact, the authors conduct some additional tests. First, they look at firms that are relatively more risky. After conducting the test, they can conclude that pre-appointment returns are significantly larger for female CEOs relative to male CEOs for low-risk firms, whereas this difference is not statistically significant in high-risk firms. The second test they conduct is needed to control for periods when the stock market is down and when it is up. Results show that, regardless of market conditions, females appear to be appointed to the CEO position following better company performance relative to the appointment of male CEOs. Lastly, the authors control for the ‘recentness’ of the glass cliff-phenomenon. Here, their data show that there “have been fewer instances of females being appointed at poorly performing firms in recent years compared to appointments made in the earlier years of the sample period.” (Adams et al., 2009, p. 9).

The authors conclude that their conclusions, together with recent findings that suggest that women leaders are more willing to take risks than men, suggest that women may self-select into precarious leadership positions. Also, they claim that female CEOs do not begin in a disadvantaged situation and they should benefit from or be held accountable for company performance while they are holding the CEO position. The conclusions by Adams et al. show a very different picture than the research by Ryan and Haslam (2005, 2008), which is why the authors suggest that additional research

(18)

18

2.2.6 Ryan and Haslam (2009)

This paper by Ryan and Haslam (2009) is a response again to the paper written by Adams, Gupta en Leeth (2009). The latter paper questioned the universality of the glass cliff-phenomenon after finding no differences in company performance before or after the appointment of male and female CEOs. In their paper, Ryan and Haslam respond by stating that they agree that glass cliffs are neither universal nor inevitable. However, they believe that people should be careful interpreting the findings of Adams et al. as well. As a reason they present the argument that the “nature and significance of women’s precarious leadership positions becomes more apparent when one goes beyond archival financial data and compares the broad circumstances of male and female leaders.” (Ryan and Haslam, 2009, p. 13). Thus, what they mean is that one cannot just look at the financial data, but one should also consider the psychological processes involved in selecting female executives over male executives.

In their article, Ryan and Haslam (2009) present a limitation to their own archival study from 2005. They acknowledge that their initial investigation revealed evidence for the glass cliff-phenomenon, but also demonstrated risks that were inherent in the interpretation of their correlational, archival data. Also, they do not speak of the impact that a third, unmeasured, variable could have on the relationship between company performance and the appointment of females in top management positions. Although they claim that their focus was never on financial performance, but rather on precariousness and risk, they do take stock data as a measure of this risk and thus (in)directly look at financial performance as well.

Ryan and Haslam (2009, pp. 14-15) state that a lack of correlation (as in Adams et al.’s data) cannot establish lack of causation (due to the possible impact of suppressor variables). For the authors, the findings by Adams et al. give rise to two interesting questions: first, why do female CEOs not confront financial glass cliffs? And second, are female CEO positions still precarious? One needs to constantly ask him-/herself these questions, thereby cautiously interpreting the results of Adams et al. Suggestions for future research include: encompass psychological data, be experimental as well as archival, look at women in a range of executive roles, and explore issues of process associated with differences in leadership experiences.

3. Critique on earlier research

(19)

19 the studies mentioned earlier. The directions for future research will form the basis for the present study.

3.1 Judge (2003)

Starting with Judge’s (2003) article, there are three main drawbacks. Firstly, her article does not present any statistical analysis. By not presenting any statistical data, it cannot be concluded that

Judge’s results are statistically significant as no solid proof has been provided. Further, Judge did not actually discuss the performance of the two companies on the bottom of the list, whose performance was less than that of the FTSE 100 companies. Also, Judge made use of too broad and loosely defined measures of women in leadership and company performance. This means that elements are included which should not have been included at all. What is more, Judge did not take into account the date of appointment of the female leader, which should really be considered, since only this data would make it possible to identify the correct causal relationship of whether declining company performance leads to more women in top positions or vice versa.

3.2 Ryan and Haslam (2005)

Continuing with Ryan and Haslam’s (2005) research, there are a few rather big limitations as well. Firstly, although the authors claim to look at FTSE 100 companies, their main focus is solely on 19 companies that appointed women in top positions. In order to have a control sample, Ryan and Haslam

(20)

20 Nonetheless, I believe the study by Ryan and Haslam (2005) to have some positive points as compared to Judge’s 2003 study. First of all, the authors look at performance both before and after the appointment of the women, which Judge does not take into account. Moreover, Ryan and Haslam also take into account the reverse explanation: declining company performance might lead to the appointment of more women in top positions, rather than the reverse being true as suggested by Judge.

3.3 Haslam and Ryan (2008)

Attention should be paid to the limitations of Haslam and Ryan’s (2008) research as well. What I believe to a poor decision in their study, is that they base their analysis on a set of experiments in which they test hypothetical situations. In these situations men and women are alternatively hypothetically placed into top positions when companies perform poorly and when they perform well. In their studies, high-school and college students and business people had to determine whether a man or a women would be better suitable to lead the company in successful times and in crisis situations. Although such studies present evidence that there might be an inclination to believe that women should lead companies in crisis while men should lead successful companies, they do not provide evidence that women indeed do get selected more often to such companies. For this, archival studies are necessary, such as the one Ryan and Haslam performed in 2005.

3.4 Adams, Gupta and Leeth (2009)

Adams et al. recognised the limitations of the research by Ryan and Haslam (2005). Therefore, they took these limitations into account and presented a more complete research themselves. For this reason, I cannot find any major flaws in their research. In fact, the research by Adams et al. presents a guideline for the present research in the sense that I look at the various positions executives can take in top management teams. This direction for future research as suggested by Adams et al. will be extended by including positions in supervisory board committees as well. Also, industry type will be considered to examine if females get appointed to industries that are struggling harder than others. This will be studied by taking a large sample of companies in The Netherlands, which Adams et al. also did with their USA sample.

4. Hypotheses

(21)

21 closer look should be given to the various leadership positions that women might hold if the glass cliff-phenomenon is true. Not only will I pay attention to positions in the executive board, I will also extend the research by Adams et al. by including positions in the supervisory boards, looking at both the chairman and the people in supervisory board committees. What is more, I want to extent this research beyond the reasoning that lower performance leads to more women in top positions, by also looking at various industries.

First studying the company level, I will look at both the executive board and the supervisory board. Since The Netherlands knows a two-tier system, meaning that public limited and private limited companies are required to have both an executive board as well as a supervisory board, this study needs to look at both types of boards. Although Adams et al. only took into consideration the CEO position, thereby only looking at the executive board, I believe that not only executive positions will recruit more women, but that supervisory positions will also be filled with women more often during a time of crisis. Supervisory boards are appointed to keep companies on track, to check the activities of the executive board and to oversee overall performance. In The Netherlands, public limited and private limited companies are obligated by law to appoint a supervisory board. The main position of this supervisory board is to oversee the policy pursued by management, which is complemented by advising management. Moreover, supervisory board members are in charge of appointing or letting go of members of the executive board, improving the balance sheet and giving their authorisation for certain decisions. Following the glass cliff-phenomenon, the first hypothesis becomes:

H1: Companies with lower performance will recruit more women to their executive and supervisory boards than companies with higher performance.

(22)

22 came about due to changes in the external environment of firms, their presence has become so important, that few companies would want to eliminate that role again (Zorn, 2004). Now that we are in a time of global economic crisis, many companies are relying on the person responsible as CFO. He or she will have to help the company get through this crisis that it faces, by either keeping performance on the same level or trying to limit the damage as much as possible. Because the CEO and CFO positions face most of the risk and uncertainty that is caused by external influences, I claim – assuming that the glass cliff-phenomenon holds – that women will fill these positions, rather than filling positions that face less risk and that will cause significantly less severe consequences when they make the wrong decisions. Therefore, my second hypothesis posits:

H2: During declining company performance, women will most likely be appointed to CEO and CFO positions, rather than all other executive positions.

It is not only important to look at the various positions that executive board members can hold, it is also important to look at the positions that supervisory board members can hold. As said before, the second company level element that I will research is whether committees within the firm are also relevant for the glass cliff-phenomenon. For this, I will consider the committees that are most common in Dutch companies: the audit committee, the remuneration committee, the selection and appointment committee, and the corporate governance committee. As a study by Catalyst (2006)shows, women are crucially underrepresented as chairs of supervisory boards and the most important supervisory board committees, which can exclude them from key leadership opportunities, including decision-making and agenda-setting. However, if the glass cliff-phenomenon is true, women now do get appointed as chairs. Obviously, chairmen will be ultimately responsible for the decisions and actions of these boards and the committees, and their decisions will have most (severe) consequences. Therefore, my third hypothesis posits:

H3: During declining performance, more women can be found as chairmen of supervisory boards than in more successful times.

(23)

23 example, the selection and remuneration committee or the corporate governance committee. As a consequence, my fourth hypothesis states:

H4: During declining performance, women will be appointed to the audit committee rather than all other committees.

In times of an economic crisis, when companies face declining performance, some companies will be more severely hit than others. The ING Economisch Bureau (Economic Bureau) presented their expectation for the economic growth per sector in The Netherlands for the years 2012 and 2013. According to this bureau, most of the sectors in the country will experience economic decline rather than growth in 2012 and some also in 2013, as can be seen in table 2.

Table 2: Volume mutations in percentages per industry. Source: www.ing.com

The table above shows the volume mutations per industry. These volume mutations indicate the growth or decline in production per industry. When production declines in an industry, this means that demand for the specific product/service declines as well, which is a bad signal for the companies in the particular industry. Those industries with the highest decline in production, experience the highest decrease in demand and are hit hardest. The sector that will be hit hardest in 2012 is ‘Construction’, with a decline in production of 0,02%. After ‘Construction’, ‘Industry’ holds the second place and ‘Transportation’ third place. Of course, it differs per year, as can also be seen in the figure, but to keep things constant, I will look at these three sectors and expect those to be hit most severely by the global economic crisis. This means that, even though these industries are mostly male-oriented, more female managers should be appointed to top positions within these companies if the glass cliff-phenomenon holds. Therefore, my final hypothesis states:

(24)

24

5. Methodology

With the intention of analysing the relationship between a dependent variable and several independent variables, multiple regression analysis seems appropriate. According to Hair Jr, Black, Babin and Anderson (2009), multiple regression analysis is a statistical technique that “can be used to analyze the relationship between a single dependent (criterion) and several independent (predictor) variables” (p. 157). The objective of this technique is to take the values that are known – those of the independent variables – and use these to predict the dependent value. Normally, multiple regression analysis can only be used when all the variables are metric. However, it is possible to use nonmetric data as well by transforming ordinal or nominal data into dummy variables, which is exactly what I have done.

Multiple regression analysis will only be done if outcomes of my preliminary analysis indicate that this will be appropriate. Since it can fairly easily be seen whether or not the number of women in specific boards or positions increases, the hypotheses can quickly be rejected if data shows that the number of women did not increase. Only in the instance that a hypothesis seems to hold – when data indicate that the number of women did indeed increase – will I conduct multiple regression analysis.

6. Data description

(25)

25 supervisory boards, the positions managers hold within these boards, and the industries the companies are in.

For this, I make use of a different method than the one chosen by Ryan and Haslam (2005). While they opted for a matched samples design, I will not select this method. By choosing a matched samples design, Ryan and Haslam limited their sample in a way that might have distorted the results. Even though they claim that women were selected in companies that were not doing so well and men in companies that were doing better, they did not base this on their entire database, but only on the 38 companies selected. It could very well be that the other companies in which male executives were selected had lower performance as well, just like the companies that selected female candidates. By excluding these companies from the analysis, their study does not present the complete picture. I understand that they tried to match the samples as well as possible, but they will never find a perfect match and that is why I believe it to be better to include the full dataset rather than a subsample.

6.1 Subject of study

This study attempts to identify whether the number of female (non-)executive directors within a selection of 64 large companies of The Netherlands increases when company performance declines. The sample was drawn from a top 100, which in turn was taken in order to provide a bigger picture than when the top 25 would have been taken as a target sample. Also, within the top 100 there is a larger variety of industries and companies represented, which will enhance the generalisability of the present study. Ultimately, I ended up with 64 companies for which data is available for all six years (2006-2011). The Netherlands being a country that employs a two-tier system in public and private limited companies, makes it worthwhile to look at both the executive board and the supervisory board. What is more, according to a study by the European Parliament (2008), female activity rates in The Netherlands are very high compared to other EU member states (70% at that time), which makes the country an ideal choice for examining female activity rates in the upper echelons of organisations, and The Netherlands is in the group of three countries with the lowest unemployment rate of women.

6.2 Variables

(26)

26

6.2.1 Dependent variable

Number of women in top positions. This variable refers to the amount of women in boards of

directors. With women in top positions I mean, following Dreher (2003, pp. 544-545), those in the “ranks of corporate officers (i.e. CEOs, COOs, operating company presidents, executive vice presidents, CFOs, and corporate vice presidents in charge of key corporate positional areas like research and development)”. Thus, they are basically those that are called ‘executives’.

The dependent variable will be subdivided into several elements, being: - The number of women in executive boards

- The number of women in supervisory boards - The number of women in the position of CEO/CFO

- The number of women in committees of supervisory boards. The committees that will be considered are the ‘main committees’, being the audit committee, the appointment committee and the corporate governance committee.

- The number of women that becomes chair of supervisory board committees

6.2.2 Independent variables

Company performance. Even though company performance can be measured through various

non-financial measures (such as reputation or customer satisfaction) as well as various non-financial measures (such as share price), I opt for one of the most common financial measures, being return on equity (ROE). The reason I choose return on equity to indicate company performance is because it is easily comparable across divisions and companies (Eccles, 1991), as well as being able to tell shareholders how effectively their money is being employed. Since share price – which has been used as indicator of company performance by both Ryan and Haslam (2005) and Adams et al. (2009)– is only based on

the perceptions of the public, I believe this to be a more subjective indicator and therefore opt for ROE instead. For the analyses, I will make use of both the performance level and the delta performance. This is important to include, because even though a company can have high performance compared to other companies, it could be that this company has experienced a major drop in company performance during the past year. On the other hand, it could also be that the performance level of a specific company is very low compared to other companies, but that it did experience a significant increase in company performance over the past year. Which company will then be more likely to add a women to its boards? The glass cliff-phenomenon does not posit anything on this subject, so it is worthwhile to be investigated in the present study.

Type of industry. As a study by the ING Economic Bureau (2012) showed, several industries are

(27)

27 industries that are mostly problematic are ‘Construction’, ‘Industry’, and ‘Transportation’. Since the glass cliff-phenomenon holds that women will get appointed to, among others, top positions of organisational units that are in crisis, they should get appointed to top positions in the three aforementioned industries more often than in all other industries. Since it is very difficult to measure industry through the use of SIC codes, a dummy variable will be used with a value of 1 when the company is in one of the three industries and a value of 0 otherwise. A dummy variable is able to translate a nonmetric variable into a metric variable by assigning either a 0 or a 1 to the specific firm, depending on whether it possesses a particular characteristic (Hair Jr, Black, Babin and Anderson, 2009), which in this case is based on whether the firm is in one of the three aforementioned industries. Industry type is based on SIC codes present in the database by Dr. Van Veen.

6.2.3 Control variables

Firm size. Firm size can be measured by several different elements, such as its assets or liabilities,

its operating revenue, number of employees, and value added. For the purpose of this study, number of employees in the firm is chosen as the determinant of firm size. According to Kumar, Rajan and Zingales (1999, p. 13), value added is preferred over output/revenue, because “the complexity of the

organization has to do with the value of its contribution, not with the value of the output sold.” A report by the European Commission (1994), called ‘Enterprises in Europe’, states that the value added

per employee is reasonably stable across size-classes, which indirectly means that firm size in terms of the number of employees will be very similar to firm size in terms of value added. Nonetheless, according to the report, coordination costs that are present in organisational theories of the firm are mostly in terms of number of employees and not their productivity. Therefore, a measure based on the number of employees is suggested by the European Commission. Also, this measure has been used for a very long time already and is widely accepted (e.g. Pashigian, 1968).

Board size. Board size in this study refers to the number of men and women in both the executive

board, as well as the supervisory board. It is thus the total amount of directors in the company. This is based on the directors listed within each company in the database compiled by dr. Van Veen.

It is important to also define ‘declining company performance’, since I have assumed throughout this analysis that declining company performance will lead to more women in the boards of companies. Therefore, a thorough understanding of this concept is required.

Declining company performance: this factor will be determined by looking at returns on equity for

(28)

28 Thus, the crisis is seen as one of the influences on declining company performance. Although there is no common definition, I will make use of the explanation by Jickling (2008, p. 1): “disruptions in financial markets rise to the level of a crisis when the flow of credit to households and businesses is constrained and the real economy of goods and services is adversely affected”.

6.3 Data gathering

For this study I will make use of a database compiled by, amongst others, Dr. Kees van Veen who has been compiling data on the top 100 companies in The Netherlands. This study will make use of a stratified data sample of his database, as it focuses solely on The Netherlands and only on a selection of companies from that country. The database covers information on both executive boards and supervisory boards, and since this study will look at the number of non-executive and executive women in companies, I will make use of all of this information. The information that has been added personally, as this was not available yet, are the various positions that these (non-)executives have held and hold in their respective companies, regarding both CEO/CFO positions, as well as committee positions. Where this information could not be found in either annual reports or on the internet, this is seen as a missing variable and will therefore be omitted from further analysis. Companies that deliberately choose not to have supervisory board committees were included in the analysis. In the event that new directors had to be added to the database, the annual reports of the companies have been consulted to check whether the director is male or female. In this, gender can be determined by the use of gender specific language or by a photo inserted in the annual report. To the extent that no clear indication was present in the annual reports as to whether the specific director is male or female, internet websites have been analysed and companies in the database have been cross-referenced in order to check whether the director is already in the database at a different company. The same steps have been taken where it is not clear which position the director holds or in which committee (s)he is present.

(29)

29 Except for data on the amount of directors and which positions they hold, more information on the companies is needed. Via Orbis, identification numbers can be found for companies, so these were looked up with the list of 64 companies at hand. Once all identification numbers are known, they can easily be entered in Orbis altogether, which is what I did. After this, I selected all the data I still needed to gather, which was: SIC code where this was not available in the dataset yet, year of incorporation, total number of employees, and company performance for which I used Return on Equity. Orbis easily looks up all of this data and is able to export this to Excel. Where data was not available for one of these elements, I looked it up in annual reports and on company websites. This way, I was able to gather all the data needed.

In order to be able to use ‘industry’ as a variable in the analysis, I created a dummy variable. Using the SIC codes to determine the industry that firms were in, I created two groups. The first group consists of companies in three industries under analysis, being ‘Construction’, ‘Industry’, and ‘Transportation’. The companies in these three industries received a value of 1. The second group consists of all other companies and these companies received a value of 0.

7. Results

7.1 Statistical pre-tests

Since the database of dr. Van Veen has been used very often during the past few years for various studies, I can assume that it is a reliable database. Moreover, the data that already existed within this database and the data that has been added by me personally comes straight from Annual Reports of all companies in the sample and from Orbis. Since it can be assumed that data published in these reports and in Orbis is correct, the reliability of the sample is very high and therefore, the reliability and consistency of this research is high as well.

(30)

30 authors, VIF values of 4 or higher may already create reasons for concern, where others think that values of 10 or higher are problematic (Eckey, Kosfeld and Türck, 2005). According to O’Brien

(2007), these ‘rules of thumb’ should not be assumed too easily; researchers should rather let the

threshold depend on the situation. However, for this research the threshold does not really matter, since the highest VIF value is 4,001 for ROE2010 and therefore is rather low. Multicollinearity can therefore be ruled out in this study.

7.2 Hypotheses

The dataset used for this analysis consists of 64 companies, all present in The Netherlands. Some are listed on the AEX index, some on the AMX index, some on the AScX index. This means that some companies are considerably larger than others, which might have an influence on the number of women in top positions in those companies, but this is something to which I get back later.

Looking at the data from the database, it can quickly be determined how many women were in the executive and supervisory boards of the 64 companies during the years 2006-2011. By putting these numbers into a graph, it can easily be seen whether or not there is indeed a trend of hiring more women into top positions when company performance declines. Looking at graph 1 below and tables 10 and 11 in the appendix, it can be seen that there is a slight trend of increasing the amount of women in executive and supervisory boards, but a ‘boost’ because of declining company performance cannot easily be determined. The graph does show that the amount of men in both boards decreases since 2009, which can indirectly mean that men are less often chosen for these boards during times of crisis.

Graph 1: Number of men and women in executive and supervisory boards in 2006-2011

(31)

31 can be seen in this graph, the number of women in committees did slightly increase, although it started to decrease (a little) again in 2011.

Graph 2: Number of women in supervisory board committees in 2006-2011

7.2.1 Executive and supervisory boards

During this study I have determined top positions to be those in both the executive boards and the supervisory boards of companies. In order to test whether there is a difference between the number of women that get appointed to these boards, I analyse them separately. Starting with the amount of women in executive boards, I look at how many women were in the executive board of one of the 64 companies in the dataset. As can be seen in graph 1 above, not many women were in executive positions during the years 2006-2011. In fact, a quick count shows that the number of women in executive boards increases from 7 to 12 from 2006 to 2011, with a mean of 0.11 and 0.19 respectively (see table 10 in the appendix). Therefore, no test is needed and it can be quickly determined that the first part of hypothesis 1, which stated “Companies with lower performance will recruit more women to their executive and supervisory boards than companies with higher performance” can be rejected.

In graph 1, we can see that the amount of women in supervisory positions grows remarkably faster than the amount of women in executive positions, which could mean that this hypothesis might be supported. In fact, the total amount of women in supervisory positions grew from 26 in 2006 to 56 in 2011, with a peak in 2010 when the amount of women in supervisory positions was 57. Since this is a rather large increase over the six years under study, statistical tests are required to examine whether or not this is a significant increase in the number of women in supervisory boards.

7.2.1.1 Logistic regression

(32)

32 dependent variable is continuous. In the present study, however, this is not the case. The dependent variable, being ‘gender selected for the position in the supervisory board’, can only be either ‘male’ or ‘female’, thereby being dichotomous. For this reason, I opt for a binary logistic regression analysis. The outcomes of the logistic regression analysis will be presented in table 3, in three different models. In model A, only the control variables have been included. Model B includes both the control variables and the independent variables, while model C includes the interaction variable next to the aforementioned variables. The table shows the magnitude of the effects that variables have on the dependent variable ‘gender selected for the position in the supervisory board’, by making use of standardised scores, which are named exp(B)-values (Lammers, Pelzer, Hendrickx & Eisinga, 2007). When the exp(B)-value is smaller than 1, there is a negative effect. Values larger than 1 indicate a positive effect between the variables. Next to this, the table also shows whether the values are significant, thus whether a CV or IV significantly affects the DV. The total variance of the different models can be shown as well in order to determine the total explained variance (Sieben and Linssen, 2009). The hypothesis on the supervisory board stated: “Companies with lower performance will

recruit more women to their supervisory board than companies with higher performance”. This will

be tested using model C, as all variables are included in this model.

Model A only includes the control variables ‘Board size’ and ‘Company size’. The first variable is measured by calculating the average board size per company over the years 2006-2011, while the latter is based on the number of employees per company in 2011. Both variables are non-significantly related to the dependent variable. In fact, ‘Company size’ does not show any relationship because it has a value of 1,000 and thus does not indicate a positive or negative effect. On the other hand, ‘Board size’ is positively related to ‘gender selected for the position in the supervisory board’, although this effect is only significant in models B and C, indicating that a larger board size does have some influence on the number of women selected for the supervisory board. Model A has a total explained variance of 0,025. This indicates that the control variables in this model only explain 2,5 percent of the total variance in ‘gender selected for the position in the supervisory board’.

(33)

33 when it is lower. This surely contradicts the glass cliff-phenomenon, which expects that women are selected for positions more often when performance is low and decreases over time. The total explained variance for the model, grows from 2,5 percent in model A to 7,6 percent in model B.