There has been much research and conjecture concerning the barriers women face in trying to climb the corporate ladder, with evidence suggesting that they typically confront a ‘glass ceiling’ while men are more likely to benefit from a ‘glass escalator’. But what happens when women do achieve leadership roles? And what sorts of positions are they given? This paper argues that while women are now achieving more high profile positions, they are more likely than men to find themselves on a ‘glass cliff’, such that their positions are risky or precarious. This hypothesis was investigated in an archival study examining the performance of FTSE 100 companies before and after the appointment of a male or female board member. The study revealed that during a period of overall stock-market decline those companies who appointed women to their boards were more likely to have experienced consistently bad performance in the preceding five months than those who appointed men. These results expose an additional, largely invisible, hurdle that women need to overcome in the workplace. Implications for the evaluation of women leaders are discussed and directions for future research are outlined.
A recent example of the scrutiny to which women leaders are exposed is provided by a lead article in The Times (Judge, 2003): Women on the Board: Help or Hindrance? Noting that more women are securing positions on company boards, Judge goes on to suggest that, although this development is seen as positive within the business and general community, in fact women leaders are having a negative impact on company performance. Thus she argues:
‘So much for smashing the glass ceiling and using their unique skills to enhance the performance of Britain's biggest companies. The triumphant march of women into the country's boardrooms has instead wreaked havoc on companies' performance and share prices.’
To support this argument the article presents data suggesting that companies with women on their boards tend to perform more poorly than those whose boards are wholly male. Using 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 of directors, Judge (2003) reports that of the top ten companies in the index (i.e. those with the highest percentage of women on their boards), six have underperformed relative to the FTSE 100 throughout 2003. In contrast, Judge reports that the five companies on the bottom of the index – companies that are wholly male – have all outperformed the FTSE 100 in 2003. From this analysis Judge concludes that ‘corporate Britain may be better off without women on the board’ (p. 21).
Problems with the analysis
However, on their own, these figures are far from conclusive and a number of serious methodological problems can be identified in Judge's (2003) analysis. First, the article reports no statistical analysis, stating simply that six of the top ten companies underperformed. Furthermore, closer examination of the original Cranfield Index (Singh and Vinnicombe, 2003) reveals that, curiously, Judge (2003) failed to report the performance of the two companies at the very bottom of the index (i.e. those with the lowest percentage of women on board). Importantly, both of these companies underperformed relative to the FTSE 100 in 2003. Therefore, a more complete picture indicates that six of the ten top companies with women directors (i.e. 60%) underperformed relative to the FTSE 100, while two of the bottom five companies without women directors (40%) under-performed – a difference that is far from statistically significant (χ2(1)=0.40 p=0.53).
In addition, the measures of women in leadership and of company performance used by Judge (2003) are broad and loosely defined. Women in leadership is operationalized simply as the percentage of women on the board of a given company – a measure that takes into account neither (a) changes in the number of women on the board of a given company, nor (b) their date of appointment or length of service. Similarly, the measure of company performance, the average share price over the year relative to the FTSE 100, is crude and does not take account of fluctuations in performance over time. Furthermore, even if the direction of the causal link between women in leadership and company performance is of the form suggested by Judge (2003), the crudeness of variable operationalisation makes it impossible to test this relationship properly. In particular, an appropriate test of the relationship would need to take account of a person's date of appointment. For example, if a poor board member has been in situ since 1997 then it is unlikely that the company would only experience a drop in performance in 2003.
An alternative explanation
Taken together – journalistic licence notwithstanding – the evidence garnered by Judge (2003) hardly justifies the conclusion that women are ‘wreaking havoc’ on the financial success of companies. And yet, having said that, her report raises important issues that demand closer empirical attention. However, a more sophisticated analysis is required to examine the purported link between having women in leadership positions and a company's financial performance. Moreover, if such an association can be identified, the merit of multiple explanations of the relationship needs to be considered.
In particular, if the relationship identified by Judge (2003) holds, one obvious alternative explanation of the association would simply involve reversing its causal sequencing (Haslam and McGarty, 2003). Thus, rather than the appointment of women leaders precipitating a drop in company performance, it is equally plausible that a company's poor performance could be a trigger for the appointment of women to the board. If this is the case, women may be being preferentially placed in leadership roles that are associated with an increased risk of negative consequences. As a result, to the extent that they are achieving leadership roles, these may be more precarious than those occupied by men. Extending the metaphors of the ‘glass ceiling’ and the ‘glass elevator’, we propose referring to this predicament as the ‘glass cliff’.
There is considerable anecdotal evidence for this phenomenon. For example, recently, within the UK, W.H. Smith has received considerable coverage in the business news both for its ‘tumbling share prices’, profit falls and proposed job cuts, and for its appointment of a woman, Kate Swann, as its CEO (BBC, 2003a, 2003b, 2004). In her new role, Swann's first, unenviable, task was to turn the company around and restore the retailer's fortunes.
Moreover, within the management literature there is evidence which suggests that women are appointed to management positions under circumstances that differ from those of male managers. For example, women managers tend to occupy particular types of management positions, being more likely to hold support roles in personnel, training, or marketing, rather than performing critical operating or commercial functions (Vinnicombe, 2000). Furthermore, there is a higher proportion of women managers in service sectors (e.g. retailing and banking) than in more industrial sectors (e.g. manufacturing, mining, and information technology; Davidson and Cooper, 1992; Goodman et al., 2003; Singh and Vinnicombe, 2003).
Lastly, there is evidence from the finance literature that has established a link between company performance and managerial turnover. In particular, Kaplan (1995) has shown that when stock performance declines companies are more likely to make changes to their boards of directors than when their performance is stable or improving. There is also evidence of a positive correlation between a company's performance and the number of subsequent directorships held by directors of that company (Fama and Jensen, 1983; Ferris, Jagannathan and Pritchard, 2003). Directors of a company that performs well typically succeed in being appointed to other boards, while those of companies that perform poorly tend to be ‘scarred for life’. Both these types of findings serve to underline the point that company performance is not simply the endpoint of appointment decisions, but should also be seen as the basis and impetus for future decisions of this form.
It is therefore important not only to focus on women leaders themselves, but also to take into account the circumstances surrounding their appointment. However, contemporary analyses of leaders and leadership have tended to neglect these situational variables, placing an emphasis instead largely on personality and individual differences (Cappelli and Sherer, 1991; Meindl, 1993).
The present study
In order to get to the bottom of the phenomenon (potentially) identified by Judge (2003) what is needed is a more nuanced analysis of women in leadership and company performance; one that can take into account situational factors such as the time of appointment and fluctuations in company performance, and can thus shed some light on the causal relationship between the appointment of women leaders and a company's financial performance. To this end, an archival study was conducted which spoke to the same data set that was of interest to Judge (2003). This investigated the share price performance of FTSE 100 companies on the London Stock Exchange both immediately before and after the appointment of a male or female board member. In this way, two alternative accounts can be tested: the one proposed by Judge (2003)– that women leaders produce relatively poor financial outcomes for their companies; and an alternative – that woman are appointed in conditions of relatively poor company performance.
Using the 2003 Cranfield Index (Singh and Vinnicombe, 2003) as a guide, the websites of all FTSE 100 companies were searched in order to identify those companies that had appointed a woman to their board of directors during 2003, and to establish the month in which these appointments were made. Relevant facts were established by accessing annual reports, press releases and director biographies. In total, 19 female board appointments were made in 2003; 17 companies were found to have appointed one woman in 2003, while one company was found to have appointed two women (at different times) in 2003.
In order that relevant gender comparisons could be made, a search was conducted to identify 19 FTSE 100 companies that had appointed a man to their board of directors in 2003. As far as possible, companies were matched for the time of appointment and for business sector (e.g. banking, retail, information technology).
Two measures of company performance were computed. The first was a broad measure of annual company performance, similar to that used by Judge (2003). Using the online London Stock Exchange Share Monitoring Service (LSE, n.d.), the performance of each of the companies was calculated as the percentage movement over the 12 months preceding 17 December 2003. A negative value represents a loss in share price over that time, whereas a positive value represents a gain in share price.
In order to investigate fluctuations in company performance a second indictor of company performance was calculated. Using the online monthly trading summaries provided by the London Stock Exchange (n.d.) an average monthly share price was calculated as the total value of shares traded divided by the number of shares traded. The average monthly share price was calculated for the six months before and after the appointment of a board member.
To test for the existence of a relationship between women in leadership and company performance, correlational analysis was performed to assess the strength of the relationship between the percentage of women on the board of a company and its annual performance in 2003. Consistent with Judge's (2003) claims, results revealed that for FTSE 100 companies there was a marginally significant negative correlation between the percentage of women in leadership positions and performance as measured by change in share price, r(97)=−0.14, p=0.09 (one-tailed). Thus, the higher the percentage of women on a company's board, the poorer the company's performance.
A t-test was also conducted to see if the annual performance of a company differed depending on whether a male or a female board member had been appointed. Analysis revealed that there was no significant difference in performance in 2003 for those companies that appointed a women (M=7.54%) compared to those companies that appointed a man (M=10.08%), t(36)<1, ns, although the trend was in the direction suggested by Judge (2003). Further, while neither companies which had appointed a male nor those which had appointed a female performed significantly different from the FTSE 100 (which showed an increase of 10.71% in the 12 months up to December 17, 2003, both ts<1), those companies that appointed men showed a significant increase in share price over that time, t(18)=2.81, p<0.02, while those that appointed a women did not show a significant increase, t(18)=1.96, p=0.07. However, as argued in the Introduction, such analysis does not shed light on the direction of causation, and such broad measures do not take account of the time of appointment or fluctuations in company performance over time.
Fluctuations in monthly performance
In order to investigate fluctuations in the performance of companies immediately before and after the appointment of an individual to their boards of directors, changes in average monthly share prices were calculated. The relative monthly performance was defined as the percentage change in share price from the previous month, with a positive value indicating an increase in average monthly share price and a negative value indicating a decrease in share price. Due to the availability of data, the relative monthly performance was calculated five months prior to the appointment to their boards of directors and three months after the appointment. In total, data were available for 15 companies with female appointments and 16 companies with male appointments.
In order to investigate company performance in the five months prior to the appointment of a board member and three months after the appointment a 2 (gender of appointee: male, female) x 2 (time of appointment: first half of the year, second half of the year) x 8 (relative monthly performance: 5 months prior to 3 months post) mixed-model analysis of variance (ANOVA) was conducted with repeated measures on the last variable. The eight levels of the repeated measures variable allowed for tests of linear, quadratic, cubic, fourth-, fifth-, sixth- and seventh-order trends in relative monthly performance.
The results revealed a main effect for time of appointment, F(1, 27)=11.50, p<0.01. These reflected overall seasonal variations in the stock market such that performance in the months immediately before and after appointments made in the first half of the year (M=0.49%) was lower than those made in the second half of the year (M=2.63%). However, this was qualified by a significant two-way interaction between relative monthly performance and time of appointment, F(7, 189)=4.20, p<0.001. Contrasts revealed a significant linear interaction between relative monthly performance and time of appointment, Flin(7, 27)=18.73, p<0.001. For appointments in the first half of the year, relative monthly performance tended to be most negative three to four months prior to appointment, increasing to reveal more positive performance in the months immediately before and after appointment. In other words, for people appointed in the first half of the year, the appointment of new board members was generally associated with improved performance. In contrast, for those appointments made in the second half of the year, relative monthly performance was stable and generally positive.
However, this two-way interaction was qualified by a marginally significant linear three-way interaction between gender, time of appointment, and relative monthly performance, Flin(7, 27)= 3.17, p<0.09, and a significant seventh-order three-way interaction, F7ord(7, 27)=7.21, p<0.02. In order to decompose these effects, separate analyses were conducted to examine the effect of time of appointment and fluctuations in monthly performance as a function of gender of the appointee (see Figure 1).
Analyses revealed that these three-way interactions arose from variation in the performance of companies that appointed women to their board. As can be seen from Figures 1a and 1b, on average, those companies that appointed a male board member showed a relatively stable performance over time, both for appointments in the first half and the second half of the year, with none of the polynomial contrasts being significant (all ps>0.05).
However, for those companies that appointed a woman to their board, company performance did vary significantly over time as a function of time of appointment, Flin(1, 13)=21.01, p<0.01, Fcub(1, 13)=4.78, p<0.05, F7ord(1, 13)=3.91, p<0.05. As can be seen in Figure 1c, when a woman was appointed in the first half of the year (i.e., when the stock market was down), company performance showed a clear and significant linear increase over time, Flin(1, 7)=14.53, p<0.01. 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.
In contrast, for those companies that appointed a woman to their board in the second half of the year (i.e. when the stock market was up), the pattern was more complex (see Figure 1d). Here there was a significant linear decrease in performance, Flin(1, 6)=7.51, p<0.04, such that performance before the appointment of a woman was generally positive while after the appointment it was relatively stable. However, in addition, this analysis revealed a significant cubic trend, Fcub(1, 6)=5.70, p=0.05. This arose from the fact that performance prior to appointment was not consistently high but fluctuated between being high and being stable.
The glass cliff
This archival examination of the performance of FTSE 100 companies questions the rather simplistic assumption that women leaders are responsible for poor company performance. Instead a more complex story is revealed, one that points to the need to take account of situational factors when examining organizational and leadership outcomes (e.g. see Miendl, 1993; Haslam, 2001). First, it is important to note that, contrary to Judge's (2003) intimation, the appointment of a woman to the board of directors was not associated with a subsequent drop in company performance. Indeed, in a time of a general financial downturn in the stock market, companies that appointed a woman actually experienced a marked increase in share price after the appointment (Figure 1c), whereas those appointments made in less unsettling times were followed by a period of share price stability (Figure 1d).
However, potentially more interesting findings uncovered by the study related to fluctuations in company performance leading up to the appointment of men and women to boards of directors. In particular, it is apparent that for companies which appointed men to their boards of directors in 2003, company performance was relatively stable, both before and after the after the appointment. Furthermore, this was true regardless of the time of year that the appointment was made, and hence regardless of the state of the stock market (see Figures 1a and 1b). For companies that appointed women to their boards of directors in 2003 a more interesting pattern of results emerged. In a time of a general financial downturn in the stock market, companies that appointed a woman had experienced consistently poor performance in the months preceding the appointment (Figure 1c). In contrast, when the stock market was more stable, companies that appointed a woman had experienced positive (but fluctuating) performance (Figure 1d).
Importantly, then, this archival study has helped to unearth an interesting phenomenon. That is, it appears that women are particularly likely to be placed in positions of leadership in circumstances of general financial downturn and downturn in company performance. In this way, such women can be seen to be placed on top of a ‘glass cliff’, in the sense that their leadership appointments are made in problematic organizational circumstances and hence are more precarious.
Implications for the evaluation of women leaders
Positions on glass cliffs can be seen as being exceedingly dangerous for the women who hold them. Companies that have experienced consistently bad performance are bound to attract attention to themselves and to those on their boards of directors (as the Swann's experience at W.H. Smith shows). Moreover, consistent with the traditional, ‘romantic’ model of leadership – and as Judge's, 2003, attributions show all too vividly – in such circumstances explanations of organizational outcomes (e.g. negative share performance), are likely to be couched in terms of the personalities and individual abilities of the leaders involved (e.g. their leadership style or financial competence) rather than on the situational and contextual variables surrounding the company (Cappelli and Sherer, 1991; Meindl, 1993; Meindl, Ehrlich and Dukerich, 1985; see also Haslam, 2001; Haslam et al., 2001). In this way, compared to men, women who assume leadership offices may be differentially exposed to criticism and in greater danger of being apportioned blame for negative outcomes that were set in train well before they assumed their new roles. This is particularly problematic in light of evidence that directors who leave the boards of companies which have performed poorly are likely to suffer from a ‘tarnished reputation’ (Ferris et al., 2003) and are less likely be offered future directorships (Brickley, Linck and Coles, 1999; Gilson, 1990).
However, it must be noted that, in reality, a company's financial performance, especially one that is floated on the share market, is determined just as much (if not more) by shareholder perceptions as it is by the actual behaviour of its board members (cf. Lord and Maher, 1991). Thus, the financial direction that a company takes after the appointment of any board members, especially over a short time-span, is likely to be more indicative of the confidence that the shareholders have in the appointment than of the actual ability of the appointed board member. Obviously, companies are aware of the importance of shareholder confidence, and therefore the appointment of a woman leader when things are going badly could be seen as a corporate strategy designed to signal to the shareholders that radical change is on the way (Furtado and Rozeff, 1987).
Directions for future research
Although this archival examination of the performance of FTSE 100 companies points to the existence of a glass cliff, it clearly represents only a preliminary investigation of this phenomenon. Further research is therefore required in order to provide more thoroughgoing examination of the psychological processes underlying these findings. One obvious question that needs to be addressed is the nature of the corporate (and general) motivations underlying the appointment of women to precarious positions. Here a variety of motivations could be posited, ranging from confidence in the particular abilities of women leaders (e.g. Eagly, Johannesen-Schmidt and van Engen, 2003) to overt sexism. It is also important to investigate women leaders' sensitivity to, and perceptions of, glass cliff positions (Schmitt, Ellemers and Branscombe, 2003). Do they see the precariousness as a form of discrimination, or simply as an opportunity to achieve?
As suggested above, another important avenue for research is to examine the way in which women who take on these glass cliff positions are subsequently evaluated by their colleagues, staff and by the public at large. Because women typically constitute a minority on the board of any company (Singh and Vinnicombe, 2003) are they more visible and thus more open to criticism (Miller, Taylor and Buck, 1991)? Are women more likely to be blamed for negative organizational outcomes? Are they equally likely to be praised for any positive outcomes?
Lastly, it is also necessary to establish whether the glass cliff phenomenon extends outside the boardroom and into other leadership arenas. Theoretically, it is also important to see whether the phenomenon is associated with low-status group membership in general and hence is a manifestation of a general social psychological process rather than one specific to gender (e.g. as discussed by Haslam, Postmes and Ellemers, 2003; Ryan and David, 2003; Ryan, David, and Reynolds, 2004). Along these lines, there is some evidence to suggest that, at least in Japan, poor company performance is associated with the appointment of ‘outsiders’ to boards of directors (Kaplan and Minton, 1994).
In order to explore these possibilities, several experimental and archival investigations are underway and all have revealed patterns commensurate with the above findings (Ryan and Haslam, in prep-a, in prep-b). In particular, there is evidence to suggest that within politics, women who rise to public office often do so under difficult circumstances. Looking, for example, at Margaret Thatcher's political career, a series of glass cliff situations can be identified. Thatcher's first brush with politics was to run as a Conservative candidate (twice) in a strong, safe Labour seat, losing both attempts. She was made Education Minister in the early 1970s when student radicalism was at its peak, facing student riots and strong criticism. Lastly, in 1979 she became Prime Minister at a time when Britain was facing rampant unemployment and economic recession. In Australia too, there have only ever been two women state premiers (Joan Kirner in Victoria and Carmen Lawrence in Western Australia). Both were appointed mid-term and after their party had been exposed to humiliating scandals. As a result, both faced the prospect of unwinnable elections which they duly lost.
It is already well established that women face greater challenges than men in their attempts to climb to the top of the corporate ladder. Moreover, it is apparent that even if they arrive there, women are likely to receive greater scrutiny and criticism than men, and to secure less positive evaluations, even when performing exactly the same leadership roles (Eagly et al., 1992). It now seems apparent that in addition to these obstacles, the leadership positions that women occupy are likely to be less promising than those of their male counterparts. So, in addition to confronting a glass ceiling and not having access to a glass elevator, they are also likely to be placed on a glass cliff. Furthermore, as the content of Judge's (2003) article indicates, if, upon finding themselves in a leadership position, they fail (as they are more likely to than men because their positions are more precarious), they may be singled out for blame and humiliation, at the same time that the unpropitious conditions of their appointment are overlooked.
The debate surrounding this process is likely to become more ferocious as women assume more leadership roles in the future (Singh and Vinnicombe, 2003). One important contribution of this paper is to help restore some balance to this debate. It does this by helping to render visible factors that are customarily ignored in the analysis of organizational leadership (Haslam, 2001; Miendl, 1993). Ironically too, it is apparent that, if overlooked, these factors can easily promote the very inequality that women's advancement is intended to redress.
Thanks to Tom Postmes for statistical guidance. This research was funded by a grant from the Leverhulme Trust (Grant F.00144.V).
Michelle Ryan is Post-Doctoral Researcher in the School of Psychology at the University of Exeter, and currently holds a five-year Academic Fellowship funded by the Research Council of the United Kingdom. Her main area of interest is examining gender and gender differences from a social psychological perspective, particularly differences within an organizational setting. Together with Alex Haslam, Michelle was recently awarded funding from the European Social Fund to further research the Glass Cliff.
Alex Haslam is Professor of Psychology at the University of Exeter. Work with colleagues at Exeter and elsewhere is informed by a social identity perspective and deals with core organizational topics such as leadership, motivation, communication, stress and power. His most recent book is Psychology in Organizations: The Social Identity Approach (2nd Ed. Sage, 2004). He is currently Chief Editor of the European Journal of Social Psychology.