Article Critique- Org Behavior

560

https://doi.org/10.1177/0149206318810415

Journal of Management Vol. 46 No. 4, April 2020 560 –582 DOI: 10.1177/0149206318810415

© The Author(s) 2018 Article reuse guidelines:

sagepub.com/journals-permissions

You’re Fired! Gender Disparities in CEO Dismissal

Vishal K. Gupta Sandra C. Mortal The University of Alabama

Sabatino Silveri University of Memphis

Minxing Sun Clemson University

Daniel B. Turban University of Missouri

CEO dismissals attract considerable attention, presumably because of the visibility, publicity, and intrigue that often surrounds the decision to fire the CEO. With the goal of advancing schol- arly understanding of CEO dismissals, we examine whether CEO gender influences the likeli- hood of dismissal. We theorize and find that ceteris paribus, female CEOs are significantly more likely to be dismissed than male CEOs. Perhaps even more importantly, we find a CEO gender by firm performance interaction such that male CEOs are less likely to be dismissed when firm performance is high (compared to when it is low), whereas female CEOs have a similar level of dismissal likelihood regardless of firm performance. Notably, our results are robust to multiple analytical techniques and various econometric specifications, bringing greater credence to the validity of our findings. Implications and directions for future research are also discussed.

Keywords: CEO dismissal; gender; firm performance

Acknowledgments: Previous versions of this research were presented at the Southern Management Association Conference and discussed at the Indian Institute of Management at Shillong, where we benefited from good feed- back and constructive comments. Conversations with Alka Gupta, Junsoo Lee, and Steven Michael were useful in the development of ideas discussed here. Erik Markin and Joshua White provided editorial assistance at different stages during the writing of this article. We are very grateful for the helpful insights and suggestions from action editor Karen Schnatterly and three anonymous reviewers, all of whom played an important role in strengthening this manuscript. Of course, all omissions and errors remain our own.

Corresponding author: Vishal K. Gupta, The University of Alabama, Tuscaloosa, AL 35487, USA.

E-mail: [email protected]

810415 JOMXXX10.1177/0149206318810415Journal of ManagementGupta et al. research-article2018

Gupta et al. / Gender and CEO Dismissal 561

The increasing presence of women in chief executive roles motivates considerable interest in understanding what happens to women after they reach the CEO position (Oliver, Krause, Busenbark, & Kalm, 2018; Zhang & Qu, 2016). Some scholars suggest that women receive preferential treatment compared to men for having reached previously inaccessible leader- ship roles (“female leadership advantage” logic; Underdahl, Walker, & Woehr, 2014). Others contend that women continue to be disadvantaged even after they attain the highest position in the organizational hierarchy (Glass & Cook, 2016). Specifically, using the metaphor of the “glass cliff,” researchers argue that women in leadership positions face more perils and risks compared to their male counterparts (Ryan & Haslam, 2007). The bias against women in leadership roles is believed to be rooted in widespread stereotypical beliefs that associate the characteristics needed for success as a leader with men but not with women (“think manager– think male” effect; Eagly & Karau, 2002; Schein, 2001). With the goal of further advancing this line of inquiry, and providing a strong test of the greater precariousness of women’s leadership position vis-à-vis men (Ryan & Haslam, 2007), we investigate whether female CEOs face differential dismissal risk compared to male CEOs.

CEO dismissal refers to the forced departure of the chief executive from the firm (Haleblian & Rajagopalan, 2006; Huson, Parrino, & Starks, 2001). Three decades ago, Fredrickson, Hambrick, and Baumrin defined CEO dismissal as “a situation in which the CEO’s departure is ad-hoc (e.g., not part of mandatory retirement policy) and against his or her will” (1988: 255), a definition that continues to resonate with researchers in this area. While CEOs depart from their firms for many reasons, dismissal—sometimes also referred to as forced turnover (Farrell & Whidbee, 2002) and involuntary exit (Alexander, Fennell, & Halpern, 1993)—has long been considered the most theoretically interesting form of CEO departure (Finkelstein, Hambrick, & Cannella, 2009). Scholars posit, with considerable supporting evidence, that firm performance is the primary metric by which CEOs are assessed and, thus, an important predictor of CEO dismissal (Hilger, Mankel, & Richter, 2013). Consequently, researchers generally view CEO dismissal as “one of the main corporate governance instruments” (Fiordelisi & Ricci, 2014: 66), so that the threat of dismissal is considered a powerful tool to pressure managers to lead their firms better and pursue value-enhancing policies (Lehn & Zhao, 2006).

Unexpected departures of CEOs from public corporations receive considerable media scrutiny (Li, Lu, Makino, & Lau, 2017). The growing number of women in the C-suite, coupled with the current zeitgeist of gender equality in society (Sandberg, 2013), motivates interest in possible gender differences in CEO dismissal. Contemporary media articles con- tend that female CEOs face greater threat of dismissal compared to male CEOs because the former get blamed disproportionately more for the problems and issues facing the company (Leung, 2014; Reingold, 2016). A recent PwC investigation noted that among CEOs leaving office in large public corporations between 2003 and 2013, 38% of women were forced out compared to only 27% of men (Favaro, Karlsson, & Nielson, 2014). While these reports sug- gest gender biases in CEO dismissal, it is worth remembering that media accounts are “not held to high methodological and peer review standards that academic work is subject to” (R. B. Adams, 2016: 373) and, thus, do not advance ongoing discussions about challenges faced by female leaders in a rigorous fashion (Ryan, Haslam, Morgenroth, Rink, Stoker, & Peters, 2016). The CEO position is the highest corporate leadership role (Zhang & Qu, 2016), and CEO dismissals are often clouded in controversy and confusion, so that systematic research

562 Journal of Management / April 2020

is needed to examine whether there are gender differences in CEO dismissals. In particular, it is important to know, using rigorous statistical tools, whether CEO dismissals actually show significant gender bias, which may inadvertently perpetuate social stereotypes suggest- ing that men, but not women, have the attributes required for successful leadership.

On the basis of 641 dismissals from 2000 to 2014, we investigate whether female CEOs are more likely to be dismissed than male CEOs and whether firm performance is differen- tially related to dismissal for male and female CEOs. As such, we examine and illuminate the gendered nature of CEO dismissal, an important issue that is a matter of scholarly inter- est and also has substantial practical implications for sound corporate governance (Hilger et al., 2013). Our research extends prior work that examines whether female, compared to male, executives face greater challenges and threats (Ryan & Haslam, 2007). Rigorous scholarship on possible gender bias at the apex of the organizational hierarchy can provide valuable insights into the similarities and differences in career derailment between high- potential men and women (Krishnan, 2009; J. B. Leslie & Van Velsor, 1996). More broadly, we examine the provocative claim that the rise of women to top executive positions in large corporations suggests that gender is no longer a relevant factor in one’s career trajectory (Elsesser, 2016), presumably because gender bias is adequately addressed by corporate policies and procedures that seek to level the playing field for men and women (Barrett & Morris, 1993; Landy, 2008).

Theory and Hypotheses

Considerable evidence indicates that gender-role stereotypes, socially shared expectations about the characteristics and behaviors of men and women (Eagly & Mladinic, 1989), have limited the ascension of women to the highest level in organizations (Heilman, 2001). Gender stereotypes typically associate men with agentic characteristics, which capture achievement- oriented tendencies (e.g., aggressive), whereas women are associated with communal attri- butes, which entail concern for the welfare of others (e.g., caring; Haines, Deaux, & Lofaro, 2016). Because leadership roles are often described in masculine (agentic) terms (Schein, 2001), and women are viewed as deficient in such qualities (Heilman, 1983), gender stereo- types constrain women’s advancement to leadership positions (Koenig, Eagly, Mitchell, & Ristikari, 2011).

Despite substantial barriers, some women have ascended into top leadership ranks in the corporate world, including at very prominent firms such as GM (Mary Barra), Pepsico (Indra Nooyi), and IBM (Virginia Rometty). Unfortunately, however, there is limited understanding of whether men and women have different experiences when they occupy similar leadership roles. Ryan and Haslam (2005) introduced the idea of the glass cliff to highlight that women who break through the “glass ceiling” will find themselves in perilous situations as they will be promoted to higher risk leadership positions and, thus, face more difficulties once they are in leadership positions. The glass cliff metaphor captures the precariousness of such posi- tions due to invisible dangers of falling from the heights of corporate leadership (Ryan, Haslam, & Kulich, 2010). Evidence is mixed, however, regarding whether women are more likely to be promoted to risky top leadership positions, with some studies finding supporting results (e.g., Cook & Glass, 2014a; Ryan, Haslam, Hersby, & Bongiorno, 2011), while others do not (S. M. Adams, Gupta, & Leeth, 2009; Cook & Glass, 2014b). However, the equally

Gupta et al. / Gender and CEO Dismissal 563

important question of whether men and women are treated differently in comparable leader- ship positions has largely gone unanswered. As Oliver et al. observed, “Though research has focused on the ascent and acceptance of female CEOs, the post-promotion circumstances female CEOs face remain unclear” (2018: 113). This is a notable omission because from an equal opportunity perspective, what happens to women in leadership positions is at least as critical as the conditions under which they made it to these positions (Zhang & Qu, 2016). Thus, we examine the important research question of whether women who reach the CEO position are treated differently than men in such positions.

Two well-regarded conceptual frameworks—token theory (Kanter, 1977) and role con- gruity logic (Eagly & Karau, 2002)—cast light on the potential challenges that female lead- ers, but not male leaders, face after they make it to coveted leadership positions like the CEO role. Token theory, sometimes labeled “critical mass theory” (Bratton, 2005), suggests that numerical minorities (such as female CEOs) often experience enhanced visibility and atten- tion, exaggerated stereotypes, and heightened monitoring and scrutiny (E. B. King, Hebl, George, & Matusik, 2010). Furthermore, role congruity theory suggests that cultural stereo- types associating leadership with masculinity can undermine evaluations of women’s com- petence and ability to lead (Kark & Eagly, 2010). Taken together, token theory and role congruity theory suggest that female leaders are more likely to be the target of others’ unfa- vorable perceptions about their ability and competence. Consistent with this, investor reac- tion to announcements of female CEO appointments is significantly more unfavorable than of male CEO appointments, presumably because female CEOs are perceived as less compe- tent than male CEOs (Lee & James, 2007). Additionally, research, using an experimental design, finds that firms having an initial public offering are perceived as less likely to suc- ceed when a woman, rather than a man, is at the helm, which then lowers investments directed at female-led firms compared to male-led firms (Bigelow, Lundmark, Parks, & Wuebker, 2014).

However, the greater salience of female CEOs may also provide them with certain bene- fits (L. M. Leslie, Manchester, & Dahm, 2017). Some scholars contend that there is a “female advantage” for women who make it to the top leadership positions (Yukl & Chavez, 2002). Gender stereotypes describe women as more skilled at inclusiveness, interpersonal relations, power sharing, and nurturing others—characteristics considered essential for leadership in modern organizations—and as a consequence, women should be superior leaders (Grant, 1988; Loden, 1985). In this vein, Rosette and Tost argue that “because a feminized approach to managing others is increasingly viewed as a strength” (2010: 222), female leaders are seen as more valuable for the firm than male leaders. Furthermore, other scholars argue that because women surmount more barriers than men on their way up, women who make it to the CEO position may be “particularly gifted and/or especially good at learning and/or deal- ing with adversity” (Gupta, Mortal, & Guo, 2018: 2039). For these reasons, some scholars suggest a “female leadership advantage” exists for women who make it to the CEO position (Eagly, 2007), a sentiment that also resonates well with the popular press and the mass media (Folkman, 2012).

The literatures reviewed above offer competing predictions for men and women in CEO positions. Empirical research also paints a mixed picture. Some studies find that female CEOs have a lower likelihood of departing from the firm than male CEOs (Elsaid & Ursel, 2018), whereas other evidence suggests that female CEOs have shorter median tenure than

564 Journal of Management / April 2020

male CEOs (Glass & Cook, 2016). Importantly, however, none of these studies distinguished between dismissals and voluntary exits, thus failing to offer a direct test of whether there are gender differences in CEO dismissal. From the perspective of token theory and role congru- ity theory, female CEOs are seen as having attributes inconsistent with leadership roles, which causes female CEOs to be scrutinized more critically and evaluated more negatively than male CEOs. The growing literature on the glass cliff phenomena also recognizes that “women managers tend to receive greater scrutiny and criticism than men, and they tend to be evaluated less favorably, even when performing exactly the same leadership roles as men” (Ryan & Haslam, 2007: 550). However, from the “female advantage” perspective, female CEOs are seen as more competent than male CEOs, which might be associated with lower levels of dismissal. We theorize, however, that the heightened scrutiny of female CEOs increases the salience of social expectations about what men and women usually do or ideally should do (Rudman & Glick, 2001), which generates less favorable evaluation of women’s leadership compared to men’s. Furthermore, the perceived cultural mismatch between women and demands of leadership roles is “likely to be the most extreme at the highest levels of leadership” (Eagly & Karau, 2002: 577), such that women “often seem inappropriate or presumptuous” when they occupy senior leadership positions (Wynen, Beeck, & Ruebens, 2015: 379). In effect, we expect the lack of fit between the feminine stereotype and the CEO role causes female managers to be scrutinized more heavily and criticized more often. Consequently, all else being equal, we expect women will seem less able and competent than men to effectively lead the firm, even when they already occupy the CEO position.

Thus, based on the conceptual logic articulated above, we hypothesize that:

Hypothesis 1: The likelihood of dismissal will be higher for female CEOs than for male CEOs.

Much of the work on CEO dismissal is grounded in agency theory (Crossland & Chen, 2013), which posits that when firms do not perform well, an effective internal governance practice is to sanction the CEO (Kato & Long, 2006). For researchers working in this vein (e.g., Murphy, 1999), poor firm performance is an ipso facto case of poor CEO performance. Consistent with this idea, firm financial performance is considered the primary predictor of CEO dismissal (Hubbard, Christensen, & Graffin, 2017). As Fredrickson et al. note, the “most obvious answer” to the question of why CEOs get dismissed, and “one for which there is [considerable] empirical support” (1988: 255), is that the firm is not performing well and is not expected to perform well in the future. While much research in this area focuses on the role of the performance of the firm itself (Hilger et al., 2013), some scholars argue that the board’s expectation about future performance—based on comparisons with industry and market performance—is a critical factor in CEO dismissal (Haleblian & Rajagopalan, 2006).

Although it is unsurprising that poor firm performance is considered a key determinant of CEO dismissal (Wiersema & Zhang, 2011), some research suggests that CEOs get fired even when firm performance is good (Martin & Combs, 2011). Ertugrul and Krishnan (2011), for example, find that approximately half the dismissed CEOs in their sample were forced out when the firm was performing well. When a firm is performing poorly, implying that the CEO is not leading the firm properly, there may be a normative expectation that the board should take action and dismiss the CEO so that a new leader can take the firm in a different direction (Hilger et al., 2013). Indeed, the behavioral theory of the firm proposes, with

Gupta et al. / Gender and CEO Dismissal 565

considerable supporting evidence, that when firm performance is below expectations, based on the firm’s prior performance and/or competitors’ performance, organizational changes are likely (Argote & Greve, 2007). Thus, when firms are performing poorly, we expect that an appropriate action is to dismiss the CEO, and in making such a decision, the board demon- strates its vigilance in exercising its statutory monitoring function. However, when the firm is performing well, there is considerable ambiguity about the CEO’s leadership of the firm and no clear script for the board to follow. In such situations we expect that gender-role ste- reotypes will influence dismissal decisions.

Considerable evidence indicates that stereotypes have greater influence in more ambiguous situations (Heilman, 1997; Heilman & Haynes, 2008). Therefore, we expect gender-role ste- reotypes, and biases associated with female leaders, to have more influence when firms are performing well because of the inherent uncertainty and ambiguity in connecting CEOs’ deci- sions and actions with specific firm outcomes (Auster & Prasad, 2016). Because of their token status and role incongruity (discussed earlier), female CEOs are more likely to be perceived as having less ability and competence than male CEOs. Indeed, Ertugrul and Krishnan (2011) observed that CEOs in well-performing firms were more likely to be dismissed when they are perceived as deficient in ability or competence. By extension, we argue that because of ambi- guity and vagueness in determining the CEO’s contributions to firm performance, CEO gen- der will influence the extent to which the CEO is given credit for good firm performance. We contend that when the firm is performing well, female CEOs are less likely than male CEOs to be considered good stewards of the firm and therefore are more likely to be dismissed.

In summary, we expect that CEO gender will affect performance-dismissal sensitivity, such that dismissal will be more sensitive to performance—both absolute and relative—for male CEOs compared to female CEOs. More specifically, we expect no gender differences in dismissal for poorly performing firms, as we expect boards to feel pressured to take action regardless of CEO gender. However, we expect that as firm performance improves, the prob- ability of dismissal for male CEOs decreases but does not protect female CEOs from dis- missal. Thus, we hypothesize:

Hypothesis 2: CEO gender will moderate the performance-dismissal relationship, such that perfor- mance will be more strongly related to dismissal for male versus female CEOs. More specifi- cally, we expect that male and female CEOs are equally likely to be fired in underperforming firms, but female CEOs are more likely to be fired than male CEOs in well-performing firms.

Method

Sample

Large public firms in the United States compose the sample for this study. We start with the ExecuComp database, which contains information on top executives for large firms. We obtain board variable data from BoardEx, which has consistent coverage starting in 2000. Accounting information is from the Compustat Industrial Annual files, and stock return information is from the monthly CRSP tapes. Our final sample comprises 21,772 firm-year observations spanning 2,390 unique firms from 2000 to 2014. CEOs who have been in office less than 1 year are not included in our sample because dismissal is highly unlikely within the 1st year.

566 Journal of Management / April 2020

Variables

Our dependent variable, CEO Dismissal, equals 1 if the CEO is fired and 0 if there is no turnover.1 Identifying CEO dismissals is difficult because firms rarely explicitly state that the departing CEO is being fired (Dedman & Lin, 2002; Shen & Cannella, 2002). Therefore, we follow Parrino (1997) to identify CEO dismissal by combining content analysis of public news reports with information about CEO age and continued affiliation with the firm. Although this approach is labor intensive, it has been employed in recent studies because it is considered quite effective in capturing CEO dismissal (Jenter & Kanaan, 2015; Zhang, 2008). On the basis of a Factiva news search of press reports, we classify CEO departure as a dismissal when it is reported that the CEO is fired or forced out (which is quite uncommon) or resigns or leaves as a result of policy differences or pressure (Jenter & Kanaan, 2015).2 Furthermore, exits for CEOs below the age of 60 are classified as dismissal if either the press does not report the reason as death, poor health, or acceptance of another position or the press notes that the CEO is retiring but does not announce the retirement at least 6 months in advance (Parrino, 1997). If news reports explain the departure as due to reasons unrelated to the firm’s activities, it is not classified as a dismissal. This method identified 641 dismissals from a total of 2,416 departures in the sample (there were 1,769 voluntary departures and 6 that could not be classified).

Although coding of press reports is the most prevalent method to code CEO dismissal (Gao, Harford, & Li, 2012; Hubbard et al., 2017), on the basis of an anonymous reviewer’s comment, we conducted robustness tests using the Peters and Wagner (2014) age-based indi- cator for CEO dismissal. We code all CEO departures below the age of 56 as dismissals and the rest as voluntary exits (Peters & Wagner, 2014). The press- and age-based measures of CEO dismissal are strongly correlated (r = .75, p < .01), and both measures yielded similar support for our hypothesized relations. Because of concerns about using only age as an indi- cator of departure (Wiersema & Zhang, 2011), and to be consistent with much of the prior research in this area (Hilger et al., 2013), we report results using the press-based measure.

Our main predictor variables are Female, an indicator variable for the CEO being female, and prior firm performance. We present all analyses with only market-adjusted returns (Ret Mkt Adj, which is the difference between the firm’s return and the market’s return over the year preceding the departure date), as standard economic theory suggests that exogenous market performance gets filtered out in evaluating firm performance attributed to the CEO. Notably, results are quite similar with alternative measures of performance: raw returns (Ret 1 Year: the firm’s stock return during the year preceding the departure date) and industry- adjusted returns (Ret Ind Adj: the difference between the firm’s return and the industry’s return during the year preceding the departure date).3 These performance measures capture both absolute performance (Ret 1 Year) and relative performance (Ret Mkt Adj and Ret Ind Adj). As reported in the Supplementary Analyses section, results are robust to employing self-relative performance (which captures historical aspiration level; Harris & Bromiley, 2007) and accounting-based firm performance measures instead. We classify each firm’s industry according to the 12-industry classification in Fama and French (1997), and we obtain industry returns from Ken French’s data library.4

We include several control variables to capture the various factors that can influence the likelihood of CEO dismissal.5 First, we control for Firm Size, measured as the natural log of a firm’s total assets for the year prior to the departure date, to account for greater expectations

Gupta et al. / Gender and CEO Dismissal 567

of CEOs at bigger firms (Shen & Cannella, 2002). Second, we control for eight specific CEO attributes that could influence the likelihood of dismissal (Hubbard et al., 2017): CEO Duality (indicator variable for CEO also being chair of the board), CEO Ownership (the shares owned by the CEO scaled by total shares outstanding), CEO Origin (dummy variable that equals 1 if the CEO comes from inside the firm and 0 otherwise; Zhang & Rajagopalan, 2010), CEO Social Status (the number of other boards of listed firms the CEO serves on rela- tive to board size of focal firm; Flickinger, Wrage, Tuschke, & Bresser, 2016), CEO Age,6 CEO Functional Experience (dummy variable that equals 1 if the CEO has finance experi- ence and 0 otherwise; Gomulya & Boeker, 2014), Other-CEO Candidates (dummy variable that equals 1 if the firm also has a separate person in the COO or president position and 0 otherwise; Zhang, 2006), and CEO Ability (dummy variable that equals 1 if the CEO obtained a degree from a top 20 SAT school; T. King, Srivastav, & Williams, 2016). Third, we control for three specific board characteristics since the board hires and fires the CEO (Hilger et al., 2013): Board Size (the number of directors on the board), Independent Directors (the propor- tion of board members who are not executives of the firm), and Female Directors (the pro- portion of female board members excluding the CEO). In all, we include 12 covariates, plus firm performance, as controls in the reported regressions. All continuous variables are win- sorized at the 1% and 99% levels.

Estimation

We use a probit model to test the two hypotheses: (1) CEO dismissal is more likely for female CEOs than for male CEOs, and (2) dismissal among female CEOs is less influenced by firm performance relative to male CEOs. A probit model is appropriate here because the dependent variable, CEO Dismissal, is binary. Our results are robust to using a logit model, as we report in the Supplementary Analyses section. We cluster standard errors at the firm level, although as we discuss later, our results are robust to clustering at the industry or CEO level. Our analyses use both industry and year fixed-effects. STATA version 14 was used for all analyses.

Recent popular and academic discussions suggest that firms with female CEOs will dif- fer systematically from firms where the CEO position remains a “male bastion” (Ryan et al., 2016: 447). If correct, this could mean that the results from the full sample probit model discussed above are affected by selection bias. To alleviate self-selection concerns, we fol- low prior research (e.g., Bugeja, Matolcsy, & Spiropoulos, 2012; Geiler & Renneboog, 2015) in generating a propensity-matched sample of male- and female-led firms to test our predictions. These results, which also validate our predictions, are presented in alternative analyses below.

Results

Tables 1 and 2 present descriptive statistics and correlations for our variables. Table 1 presents descriptive statistics for the sample as a whole and for male- and female-led firms separately. Table 2 presents correlation coefficients for the sample as a whole. There are 21,772 firm-year observations in total, of which 617 are female-led and 641 are CEO dis- missals. Thus, female CEOs represent about 3% of our firm-year observations, consistent

568 Journal of Management / April 2020

with the notion that the representation of women at the CEO level remains quite low (Cook & Glass, 2014a). Dismissals are also about 3% of the total firm-year observations in the sample. Though the number of female dismissals is substantially lower than male dismissals (30 vs. 611), the percentage of women dismissed is significantly higher than men dismissed (5% vs. 3%), which is consistent with our expectations. Finally, female-led firms differ along a number of dimensions from their male-led counterparts. Female-led firms are smaller in size and have lower market-adjusted performance. Additionally, their boards are smaller and have a larger proportion of female and independent directors. Female CEOs are younger, less likely to be the chairperson of the board, own lower percentages of the firm, are more likely to progress from the inside, and sit on more boards compared to male CEOs. Female CEOs are also less likely to have graduated from a top school (our measure for CEO ability) and are less likely to have experience in the finance area (functional experience).

In Table 2, we find that most correlations are .25 or less in absolute terms. As expected, the correlation between board size and log (assets) is high (.62). As noted earlier, we report all regressions with market-adjusted returns as the performance measure, although results are similar with alternative performance measures. Finally, variance inflation factors are all less than 3, indicating that multicollinearity is not an issue in our data (Kennedy, 2003).

Table 3 presents the probit regression results for the first hypothesis that dismissal is more likely for female CEOs than for male CEOs. Probit regressions are nonlinear; thus, the eco- nomic significance of the regression coefficients is not straightforward to interpret (Bowen & Wiersema, 2004). We therefore include, for each model, average marginal effects next to the regression coefficients, which represent the change in probability associated with an increase in the dependent variable by 1 unit (we use STATA procedure MARGINS, DYDX).

Table 1

Descriptive Statistics

Whole Sample Male CEOs Female CEOs Female–Male

M SD M SD M SD Difference p value

CEO dismissal 0.03 0.17 0.03 0.17 0.05 0.22 0.02 .004 Log (assets) 7.77 1.70 7.77 1.70 7.56 1.75 −0.21 .002 CEO ownership % 0.02 0.05 0.02 0.05 0.01 0.04 −0.01 .000 CEO age 56.18 7.31 56.25 7.36 53.85 5.05 −2.40 .000 CEO duality 0.57 0.50 0.57 0.49 0.37 0.48 −0.20 .000 Candidate 0.47 0.50 0.47 0.50 0.40 0.49 −0.07 .000 Board size 9.39 2.50 9.40 2.51 9.05 2.11 −0.36 .000 CEO social status 0.17 0.09 0.17 0.09 0.20 0.10 0.03 .000 Female directors % 0.11 0.11 0.11 0.09 0.15 0.11 0.05 .000 Functional experience 0.05 0.21 0.05 0.21 0.04 0.19 −0.01 .255 Independent directors % 0.68 0.13 0.68 0.13 0.72 0.13 0.04 .000 CEO origin 0.67 0.47 0.66 0.47 0.76 0.43 0.09 .000 CEO ability 0.33 0.47 0.33 0.47 0.30 0.46 −0.03 .146 Ret Mkt Adj 0.07 0.41 0.07 0.41 0.04 0.41 −0.04 .027

Note: There are 21,772 firm-year observations. Unique firms = 2,390, female firm–year observations = 617, unique female firms = 130, CEO dismissals = 641, female CEO dismissals = 30. Ret Mkt Adj = market-adjusted returns.

Gupta et al. / Gender and CEO Dismissal 569

The table also includes the p values in parentheses below the coefficients. Model 1 contains the results from including all control variables. Model 2 adds Female as the predictor vari- able of interest. The coefficient on Female is positive and statistically significant (β = 0.19, t = 2.06, p = .04), and the average marginal effect indicates that a female CEO has about a 1 percentage point higher probability of being fired than a male CEO. Using Model 2 infor- mation, the predicted dismissal probability for female CEOs is approximately 4.2% and for male CEOs approximately 2.9%, indicating that female CEOs are about 45% more likely to be dismissed than male CEOs. Although not shown in the table, Female remains significantly related to dismissal when the other performance measures (raw returns, industry-adjusted returns, and self-relative returns) are used. Thus, Hypothesis 1 is supported.

Model 3 presents probit regression results testing the second hypothesis that dismissal prob- ability for female CEOs is less likely to be influenced by changes in performance than male CEOs. In this model, we add an interaction variable of Female and firm performance. Our inter- est here is in the regression coefficient and average marginal effects associated with this variable. The coefficient for the interaction term is positive and significant, as hypothesized (β = 0.96,

Table 2

Correlations

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1. CEO dismissal

1

2. Log (assets) −.01 (.381)

1

3. CEO ownership %

−.05 (.000)

−.19 (.000)

1

4. CEO age −.05 (.000)

.11 (.000)

.16 (.000)

1

5. CEO duality −.04 (.000)

.13 (.000)

.19 (.000)

.27 (.000)

1

6. Candidate .00 (.608)

.11 (.000)

.07 (.000)

.08 (.000)

.18 (.000)

1

7. Board size .00 (.991)

.62 (.000)

−.17 (.000)

.08 (.000)

.05 (.000)

.11 (.000)

1

8. CEO social status

−.01 (.041)

−.07 (.000)

.01 (.115)

.09 (.000)

.15 (.000)

.01 (.235)

−.29 (.000)

1

9. Female directors %

.03 (.000)

.25 (.000)

−.09 (.000)

.00 (.974)

.04 (.000)

−.02 (.000)

.23 (.000)

−.06 (.000)

1

10. Functional experience

.00 (.977)

.06 (.000)

−.06 (.000)

−.03 (.000)

−.01 (.152)

.00 (.568)

.07 (.000)

−.01 (.068)

.05 (.000)

1

11. Independent directors %

.01 (.093)

−.05 (.000)

−.12 (.000)

−.02 (.026)

−.11 (.000)

−.49 (.000)

.18 (.000)

−.14 (.000)

.10 (.000)

.02 (.002)

1

12. CEO origin .00 (.833)

.22 (.000)

−.10 (.000)

.01 (.130)

.02 (.008)

.09 (.000)

.20 (.000)

−.03 (.000)

.11 (.000)

.09 (.000)

−.11 (.000)

1

13. CEO ability .00 (.763)

.10 (.000)

.00 (.496)

−.01 (.106)

.04 (.000)

.04 (.000)

.07 (.000)

.08 (.000)

.08 (.000)

−.02 (.003)

−.04 (.000)

.00 (.494)

1

14. Ret Mkt Adj −.12 (.000)

−.06 (.000)

.03 (.000)

−.01 (.146)

.02 (.026)

.01 (.082)

−.04 (.000)

.03 (.000)

−.04 (.000)

−.01 (.347)

−.03 (.000)

−.02 (.001)

.00 (.546)

1

15. Female .02 (.004)

−.02 (.002)

−.03 (.000)

−.05 (.000)

−.07 (.000)

−.02 (.000)

−.02 (.001)

.05 (.000)

.08 (.000)

−.01 (.256)

.05 (.000)

.03 (.000)

−.01 (.146)

−.01 (.027)

1

Note: There are 21,772 firm-year observations. Unique firms = 2,390, female firm–year observations = 617, unique female firms = 130, CEO dismissals = 641, female CEO dismissals = 30. Parentheses contain p values. Ret Mkt Adj = market-adjusted returns.

570 Journal of Management / April 2020

Table 3

Incremental Effect of CEO Gender and Firm Performance on CEO Dismissal Likelihood (Full Sample)

(1) (2) (3)

β AME β AME β AME

Firm sizea 1.29 (.442)

0.08 (.441)

1.36 (.414)

0.08 (.414)

1.48 (.374)

0.09 (.374)

CEO ownership −5.59 (.000)

−0.34 (.000)

−5.57 (.000)

−0.34 (.000)

−5.55 (.000)

−0.34 (.000)

CEO agea −1.14 (.000)

−0.07 (.000)

−1.12 (.000)

−0.07 (.000)

−1.13 (.000)

−0.07 (.000)

CEO duality −0.09 (.032)

−0.01 (.032)

−0.08 (.042)

−0.01 (.042)

−0.09 (.037)

−0.01 (.037)

Other-CEO candidatesa 7.34 (.079)

0.45 (.079)

7.20 (.085)

0.44 (.086)

7.31 (.080)

0.44 (.081)

Board sizea −0.57 (.613)

−0.03 (.613)

−0.53 (.635)

−0.03 (.635)

−0.57 (.616)

−0.03 (.616)

CEO social status −0.43 (.064)

−0.03 (.065)

−0.47 (.040)

−0.03 (.040)

−0.46 (.045)

−0.03 (.046)

Female directors 0.67 (.001)

0.04 (.001)

0.63 (.002)

0.04 (.002)

0.65 (.001)

0.04 (.001)

CEO functional experiencea −1.19 (.888)

−0.07 (.888)

−0.56 (.947)

−0.03 (.947)

−0.40 (.962)

−0.02 (.962)

Independent directors 0.30 (.110)

0.02 (.110)

0.29 (.130)

0.02 (.130)

0.29 (.126)

0.02 (.126)

CEO origina −0.92 (.816)

−0.05 (.816)

−1.26 (.752)

−0.08 (.752)

−1.20 (.763)

−0.07 (.763)

CEO abilitya 2.39 (.535)

0.15 (.535)

2.48 (.519)

0.15 (.519)

2.53 (.512)

0.15 (.512)

Ret Mkt Adj −1.05 (.000)

−0.06 (.000)

−1.06 (.000)

−0.06 (.000)

−1.13 (.000)

−0.07 (.000)

Female 0.19 (.040)

0.01 (.075)

0.24 (.007)

0.02 (.023)

Female × Ret Mkt Adj 0.96 (.003)

0.06 (.003)

Constant −1.40 (.000)

−1.41 (.000)

−1.42 (.000)

Observations 21,772 21,772 21,772 Year fixed-effects Yes Yes Yes Industry fixed-effects Yes Yes Yes Firm cluster Yes Yes Yes Pseudo R2 .11 .11 .11 Akaike information criterion .24 .24 .24 Log likelihood −2,586 −2,584 −2,576 Chi-square 344.5 357.1 37.3 Chi-square against model 3 21.67 17.71

Note: Results are tabulated only for market-adjusted returns (Ret Mkt Adj), but other measures of returns give similar results (available from authors). Average marginal effects (AME), which capture the average of the marginal effects across all covariate values, are reported; p values are adjusted for firm-level clustering and are presented in parentheses. The Akaike information criterion estimates the relative quality of statistical models for a given set of data. aVariables scaled by dividing by 100.

Gupta et al. / Gender and CEO Dismissal 571

t = 3.01, p = .003). The chi-square statistic indicates that Model 3 represents a better fit for our data than Models 1 and 2. Furthermore, the Akaike information criterion statistic is lowest for Model 3, thereby corroborating the idea that Model 3 provides the best fit for the data.

Figure 1 provides the dismissal probabilities for male and female CEOs at various levels of performance along with confidence intervals at the 95% level (we use STATA procedures MARGINS and MARGINSPLOT, XDIMENSION). As expected, Figure 1 shows that at low levels of performance, the dismissal probabilities do not differ for male and female CEOs. However, as performance improves, the probability of dismissal decreases sharply for male CEOs but does not vary much for female CEOs, so that the male CEOs have a significantly lower dismissal probability than female CEOs at higher levels of performance. Thus, Hypothesis 2 is supported.

Furthermore, following an anonymous reviewer’s suggestion and prior research (e.g., Bowen, 2012; Lel & Miller, 2008; Norton, Wang, & Ai, 2004; Wiersema & Bowen, 2009), we plot the interaction effect in Figure 2a and corresponding z statistics in Figure 2b, both as a function of predicted probabilities (we use STATA procedure INTEFF). The interaction effect represents the marginal effect of firm performance for female CEOs in relation to their male counterparts on the probability of dismissal and is akin to the coefficient on the interac- tion variable in linear models. It is computed as the cross-partial derivative of the expected probability of dismissal with respect to performance (x1) and gender (x2) (specifically, we

compute ∆

∆

δ δ F u

x x

( )

1 2

). This derivative varies with the value of the independent variables; thus,

Figure 1 Dismissal Probabilities for Female and Male CEOs as a Function of

Firm Performance (Market-Adjusted Returns) Along With Confidence Levels (at the 95% Level)

572 Journal of Management / April 2020

Figure 2 Interaction Effects and Corresponding z Statistics on the Interaction Variable

Between the CEO Gender and Firm Performance Measure

Note: The figure shows scatterplots of (a) the interaction effect as a function of predicted probability of dismissal and (b) z statistics for the interaction effect.

Gupta et al. / Gender and CEO Dismissal 573

we plot its value against the predicted probability of dismissal for all data points. The graph in Figure 2a shows that irrespective of predicted probabilities, the interaction effect is largely positive, suggesting that for female CEOs, the drop in the likelihood of dismissal is less sen- sitive to improvements in performance than for male CEOs (supporting Hypothesis 2). The graph in Figure 2b shows the z statistic for most values is above the 5% threshold. Bowen (2012) advises extracting (what he calls) the secondary effect from the interaction effect. He observes that the interaction effect can be nonzero even when the coefficient on the interac- tion variable is zero; thus, it would be more meaningful to subtract from the interaction effect the portion obtained when the coefficient on the interaction is set to zero—the secondary effect. We find that the secondary effect is similar to the interaction effect, consistent with Hypothesis 2 (these results are available from the authors).

In additional analyses, we run two separate regressions in lieu of including the interaction variable: one regression when performance is above the median and another when perfor- mance is below the median (not reported). We find that when performance is above the median, the coefficient on Female is positive and statistically significant, suggesting that female CEOs are more likely to be dismissed than their male counterparts when performance is high. However, when performance is below the median, the coefficient on Female is sta- tistically insignificant, suggesting women are no more likely to be dismissed than men when performance is low. These results are consistent with Hypothesis 2 and the results we obtain from using the interaction variable.

Alternative Analyses

To alleviate selection concerns, we also test Hypotheses 1 and 2 using a matched-sample approach. Because matching on multiple covariates is more likely to produce unbiased esti- mates than matching on only one or two attributes (Armstrong, Ittner, & Larcker, 2012), we match on multiple covariates (described below) by creating a propensity score. Propensity score matching (PSM; Rosenbaum, 2000) is beneficial for our purpose because it seeks to obtain matched groups when random assignment is not possible.

To create propensity scores, we regress (using probit) Female on seven predictors: CEO ownership, CEO duality, CEO social status, female representation on the board, market- adjusted returns, firm size, and percent of female directors in the industry. Of these, the first five were chosen because they differ systematically across the samples of male- and female- led firms and are at the same time also predictors of CEO dismissal; firm size was selected because it is a standard control in matched-sample studies; and the last predictor was selected because industry variations in female representation at the upper echelon are intuitively related with the likelihood of a firm selecting a female CEO (we thank an anonymous reviewer for this suggestion). The propensity score reflects the predicted probability of the CEO being female given these characteristics. We then use the propensity scores to match each female CEO firm with a male CEO firm with exact matches on year and industry and the most proximal propensity score. We subsequently examine our two hypotheses using this matched sample (see Table 4).7 Results are consistent across the different return measures, and we thus present the results for only market-adjusted returns (Ret Mkt Adj). Consistent with the results from the full sample, female CEOs have a significantly higher likelihood of being fired than male CEOs (β = 0.40, t = 2.74, p = .006; see Model 2), which provides

574 Journal of Management / April 2020

Table 4

Incremental Effect of CEO Gender and Firm Performance on CEO Dismissal Likelihood (Propensity Score–Matched Sample)

(1) (2) (3)

β AME β AME β AME

Firm sizea 1.20 (.842)

0.09 (.842)

1.14 (.848)

0.09 (.847)

1.84 (.762)

0.14 (.762)

CEO ownership −4.18 (.379)

−0.32 (.379)

−3.81 (.390)

−0.29 (.391)

−3.30 (.443)

−0.24 (.443)

CEO agea −1.59 (.221)

−0.12 (.230)

−1.14 (.437)

−0.09 (.443)

−1.35 (.374)

−0.10 (.385)

CEO duality −0.16 (.340)

−0.01 (.339)

−0.17 (.296)

−0.01 (.300)

−0.20 (.207)

−0.01 (.214)

Other-CEO candidates −0.25 (.191)

−0.02 (.193)

−0.28 (.148)

−0.02 (.149)

−0.33 (.086)

−0.02 (.009)

Board sizea 0.09 (.985)

0.01 (.985)

0.20 (.969)

0.02 (.969)

0.42 (.937)

0.03 (.937)

CEO social statusa 3.23 (.969)

0.25 (.969)

−3.63 (.723)

−2.30 (.724)

−5.93 (.946)

−0.44 (.946)

Female directors 1.45 (.038)

0.11 (.032)

1.46 (.044)

0.11 (.038)

1.47 (.048)

0.11 (.040)

CEO functional experiencea −0.19 (.997)

−0.01 (.997)

13.83 (.758)

0.01 (.758)

15.09 (.735)

1.12 (.735)

Independent directors −0.30 (.688)

−0.02 (.688)

−0.51 (.496)

−0.04 (.497)

−0.50 (.521)

−0.04 (.522)

CEO origin −0.37 (.008)

−0.02 (.008)

−0.44 (.004)

−0.03 (.004)

−0.43 (.006)

−0.03 (.007)

CEO ability −0.15 (.356)

−0.01 (.357)

−0.17 (.336)

−0.01 (–.338)

−0.16 (.379)

−0.01 (.381)

Ret Mkt Adj −0.36 (.237)

−0.03 (.236)

−0.41 (.185)

−0.03 (.186)

−1.27 (.017)

−0.09 (.019)

Female 0.40 (.006)

0.03 (.006)

0.52 (.009)

0.04 (.009)

Female × Ret Mkt Adj 1.16 (.049)

0.09 (.053)

Constant −4.03 (.001)

−4.54 (.000)

−4.46 (.001)

Observations 1,224 1,224 1,224 Year fixed-effects Yes Yes Yes Industry fixed-effects Yes Yes Yes Firm cluster Yes Yes Yes Pseudo R2 .12 .13 .15 Akaike information criterion .36 .36 .35 Log likelihood −16.0 −156.9 −154.0 Chi-square 1,274 822.4 712.0 Chi-square against model 3 11.98 5.70

Note: Unique firms = 606, firm-year observations = 1,224, female firm–year observations = 612, unique female firms = 130, CEO dismissals = 43, female CEO dismissals = 29. Results are tabulated only for market-adjusted returns (Ret Mkt Adj), but other measures of returns give similar results (available from authors). Average marginal effects (AME), which capture the average of the marginal effects across all covariate values, are reported; p values are adjusted for firm-level clustering and are presented in parentheses. aVariables scaled by dividing by 100.

Gupta et al. / Gender and CEO Dismissal 575

further support for Hypothesis 1. We also find a significant coefficient on the interaction between Female and Ret Mkt Adj (β = 1.16, t = 1.97, p = .049; see Model 3). Once again, the probability of male CEOs being dismissed is reduced when firm performance is better, but the probability of female CEOs being dismissed is consistent across firm performance levels. These results provide further support for Hypothesis 2.

Supplementary Analyses

We perform additional analyses to ensure the robustness of our results. First, while the results were estimated using probit regression models, the findings remain consistent with logit regression models. Second, we cluster standard errors at the firm level to account for possible correlations through time within the same firm, but our results are similar when we instead cluster standard errors at the industry level or the CEO level. Third, we also ensure that our results are robust to using only the subsample comprising firm-years where CEOs depart the firm (whether forced or voluntary, N = 2,413). Results are not presented here for parsimony but are available from the authors. Fourth, on the basis of an anonymous review- er’s suggestion, we test our hypotheses in a subsample that includes only a balanced panel of firms (N = 1,911) and find that results on the key variables of interest remain unchanged. Finally, we find that our findings and conclusions are similar when we use other performance measures such as raw firm returns, industry-adjusted firm returns, or self-relative firm returns as well as using the accounting-based firm performance measure return on equity, though we do not present these results for parsimony.

Discussion

The issue of CEO dismissal attracts significant attention both in academic research and in the popular press. Yet the ongoing conversation about CEO dismissal has (mostly) ignored a crucial new development in the corporate world: the growing presence of women in the CEO position. Building on prior research (e.g., Fredrickson et al., 1988; Hilger et al., 2013), we examine the role of CEO gender in dismissals. We theorize and find that female CEOs are much more likely to be dismissed than male CEOs. Perhaps even more importantly, we find that while the rate of male and female CEO dismissal is similar when the firm is performing poorly, female CEOs are significantly more likely to be dismissed than male CEOs when the firm is performing well. Taken together, our results contribute to a deeper understanding of what happens to men and women who make it to the top of the organizational hierarchy, revealing the higher risks women face of being dislodged from their leadership position. Notably, our results, based on a large data set of public corporations in the United States over the time period 2000 to 2014, are robust to different statistical procedures and econometric specifications, which strengthens confidence in our findings.

Our research reveals that female CEOs are approximately 45% more likely to be dis- missed than male CEOs, which is indicative of significant gender bias in CEO dismissal. Although it has long been recognized that gender stereotypes “give rise to biased judgments and decisions, impeding women’s advancement” in the organization (Heilman, 2012: 115), our results suggest that gender stereotypes may also result in women being pushed out of their leadership roles even after they reach the highest position in the firm. It is telling that

576 Journal of Management / April 2020

existing models of CEO dismissal (e.g., Fredrickson et al., 1988; Hilger et al., 2013) do not consider CEO gender as an explanatory variable, perhaps because those models were con- ceived when women were largely absent from the corner office. We hope our research will draw greater attention to the gendered nature of CEO dismissal and motivate interest in understanding the drivers of the higher likelihood of dismissal among female CEOs. From a practical standpoint, our research provides useful information about a substantive hidden bias in CEO dismissal, which should be relevant to firm boards and those involved in improv- ing corporate governance practices. The board’s authority to dismiss a CEO comes with the responsibility to exercise its prerogative in a fair manner, and our results, which indicate significant gender differences in CEO dismissal, suggest bias that favors male CEOs over female CEOs. Given that most boards are “full of aging white men” (Arfken, Bellar, & Helms, 2004: 180), it may be useful to include gender-sensitivity training focused at reducing bias in the board’s assessment of senior managers for current and aspiring board members.

Our findings regarding the interaction between CEO gender and firm performance are informative and intriguing. Specifically, results indicate that male and female CEOs face similar risk of being fired when firm performance is poor, but dismissal probability decreases significantly with performance improvements only for male CEOs. Such results indicate that dismissal is sensitive to performance for male CEOs, as traditional agency-theory explana- tions suggest (Crossland & Chen, 2013), but not for female CEOs, contributing to ongoing discussions about the performance sensitivity of CEO dismissals (Park, Kim, & Sung, 2014). Although firm performance is widely recognized as a predictor of CEO dismissal (Finkelstein et al., 2009), there is little understanding of the factors that decouple the performance-dis- missal link (Graffin, Boivie, & Carpenter, 2013). This lack of understanding is unfortunate because empirical investigations indicate that in many situations, firm performance “explains only a small portion of variance in CEO dismissals” (Hubbard et al., 2017: 2257), and many CEOs are dismissed even in the absence of performance troubles (Ertugrul & Krishnan, 2011). We consider the gender differences in CEO firings in the absence of performance problems quite disconcerting, as they suggest bias in the dismissal of female CEOs relative to male CEOs.

We also advance the emerging literature on the greater precariousness of women’s leader- ship position, nominally referred to as the glass cliff phenomenon (Ryan et al., 2016). Our findings that female CEOs are more likely to be forced out and that female CEO dismissal is less sensitive to performance factors provides support for the glass cliff thesis that women face higher risks and perils in their leadership positions than men. While much research in this area has focused on the challenges that women face in rising to the top (Cook & Glass, 2014a), our research draws attention to the comparatively higher rate at which female leaders fall, or are pushed, from the top.

Limitations and Directions for Future Research

Our research illuminates intriguing gender differences in CEO dismissal in a rigorous fashion, yet it has some limitations that need to be acknowledged for their potential to open new avenues for future inquiry. First, although our press-based measure of CEO dismissal is consistent with much prior research (e.g., Guo & Masulis, 2015; Shen & Cannella, 2002), we acknowledge ongoing “concerns about the measurement of dismissal” in the literature

Gupta et al. / Gender and CEO Dismissal 577

(Hilger et al., 2013: 24). Since firms usually do not explicitly disclose CEO departure as fir- ing, it is difficult to know with complete confidence why a CEO is leaving the firm. Consequently, CEO dismissal tends to be captured through indirect measures, such as news articles (Fee, Hadlock, Huang, & Pierce, 2018). We attempt to address concerns about the accuracy of press-based measures by using an alternative age only–based measure of CEO dismissal (Peters & Wagner, 2014) and find similar results. Nevertheless, we acknowledge that existing measures of CEO dismissal are imperfect and believe that additional research examining the construct validity of various dismissal measures would be useful. Such research would be especially valuable if, perhaps using qualitative techniques such as tran- scripts of board meetings, scholars could ascertain whether their coding of CEO dismissal is accurate. We understand that access to corporate elites and their confidential deliberations is very difficult to obtain, but Tuggle, Schnatterly, and Johnson (2010) provide a model for how such data may be obtained.

Second, research comparing male- and female-led firms is vulnerable to selection bias as firms that choose a female CEO differ from firms that choose a male CEO (Huang & Kisgen, 2013). We attempt to account for such differences with our control variables and by conduct- ing a propensity-matched sample analysis that matched male- and female-led firms on observable characteristics. Although our results were consistent using various analytic tech- niques, we acknowledge that since we do not have an experimental design with random assignment, it is possible that firms may differ on some unobserved characteristics that influ- ence our pattern of results (Ketokivi & McIntosh, 2017). Nonetheless, our analytic approach is consistent with prior research (e.g., Bugeja et al., 2012) and has been recommended by many researchers (Caliendo & Kopeinig, 2008; Shipman, Swanquist, & Whited, 2016). Furthermore, although we acknowledge that it is impossible to rule out selection bias without random assignment, we believe that the robustness of our results to alternative analytical tools (e.g., PSM) suggests that selection bias is not a viable explanation for our pattern of results.

Third, while informative in revealing a gender dimension to CEO dismissal that had been overlooked in the academic literature, our research does not directly capture the reasons for differential incidence of dismissal among male and female CEOs. For example, it is possible that male and female CEOs differ on human or social variables (e.g., productivity or con- nectedness) that are not fully accounted for by the control variables we include. In this vein, some research suggests that male and female managers differ in important ways (e.g., over- confidence; Huang & Kisgen, 2013), although others argue that women and men at the upper echelon level are more alike than different (R. B. Adams & Funk, 2012). An interesting extension of our work would be to examine whether male- and female-led firms differ on critical characteristics (e.g., Do male and female CEOs run firms that differ on productivity?) that may be relevant to our research question.

Conclusion

The rate of CEO dismissal has increased substantially over time, drawing considerable attention to who gets dismissed and why (Wiersema & Zhang, 2011). We find that consistent with our theorizing, female CEOs are more likely to be dismissed than male CEOs and that higher levels of firm performance protect male, but not female, CEOs from dismissal. Such

578 Journal of Management / April 2020

results provide strong evidence that gender plays a significant role in CEO dismissal. More broadly, our results challenge conventional wisdom that CEO dismissal results largely from firm performance issues (Hubbard et al., 2017), as well as advance knowledge about the challenges women face in CEO positions (Joshi, Neely, Emrich, Griffiths, & George, 2015). The empirical support we find in various regression models across different econometric specifications enhances confidence in the robustness of our findings. We hope our research increases awareness of differential treatment meted out to women in senior management positions compared to their male counterparts and spurs researchers and practitioners to eliminate gender bias from CEO dismissal and other corporate governance functions.

Notes 1. We exclude voluntary turnovers from our sample. However, including such turnovers does not materially

change our results. 2. We thank Dirk Jenter and Florian Peters for help with obtaining CEO dismissal data. 3. The correlations between the various return-based performance measures range from .65 to .93 (all statisti-

cally significant), which reveals high intercorrelations among the different ways of measuring firm performance. All analyses in this paper are presented using market-adjusted returns, but the results are similar when we use alternative measures.

4. Accessible at http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html. 5. Given recent discussions about potential “overcontrolling” problems (Bernerth & Aguinis, 2016; Carlson &

Wu, 2012; Spector & Brannick, 2011), we also ran the analyses with fewer controls and found no meaningful change in the results.

6. CEO age is a continuous variable in our analyses. Following prior research (e.g., Mobbs, 2013; Zhang, 2006), we also ran robustness tests with CEO age as a dummy variable and found that results remain largely unchanged.

7. As expected, the difference between characteristics of female-led and male-led firms are much smaller after PSM. Although tabulated summary statistics are not presented here for parsimony, these results are available from the authors.

References Adams, R. B. 2016. Women on boards: The superheroes of tomorrow? The Leadership Quarterly, 27: 371-386. Adams, R. B., & Funk, P. 2012. Beyond the glass ceiling: Does gender matter? Management Science, 58:

219-235. Adams, S. M., Gupta, A., & Leeth, J. D. 2009. Are female executives over-represented in precarious leadership

positions? British Journal of Management, 20: 1-12. Alexander, J. A., Fennell, M. L., & Halpern, M. T. 1993. Leadership instability in hospitals: The influence of board-

CEO relations and organizational growth and decline. Administrative Science Quarterly, 38: 74-99. Arfken, D. E., Bellar, S. L., & Helms, M. M. 2004. The ultimate glass ceiling revisited: The presence of women on

corporate boards. Journal of Business Ethics, 50: 177-186. Argote, L., & Greve, H. R. 2007. A behavioral theory of the firm—40 years and counting: Introduction and impact.

Organization Science, 18: 337-349. Armstrong, C. S., Ittner, C. D., & Larcker, D. F. 2012. Corporate governance, compensation consultants, and CEO

pay levels. Review of Accounting Studies, 17: 322-351. Auster, E. R., & Prasad, A. 2016. Why do women still not make it to the top? Dominant organizational ideologies

and biases by promotion committees limit opportunities to destination positions. Sex Roles, 75: 177-196. Barrett, G. V., & Morris, S. B. 1993. The American Psychological Association’s amicus curiae brief in Price

Waterhouse v. Hopkins: The values of science versus the values of the law. Law and Human Behavior, 17: 201-215.

Bernerth, J. B., & Aguinis, H. 2016. A critical review and best-practice recommendations for control variable usage. Personnel Psychology, 69: 229-283.

Gupta et al. / Gender and CEO Dismissal 579

Bigelow, L., Lundmark, L., Parks, J. M., & Wuebker, R. 2014. Skirting the issues: Experimental evidence of gender bias in IPO prospectus evaluations. Journal of Management, 40: 1732-1759.

Bowen, H. P. 2012. Testing moderating hypotheses in limited dependent variable and other nonlinear models: Secondary versus total interactions. Journal of Management, 38: 860-889.

Bowen, H. P., & Wiersema, M. F. 2004. Modeling limited dependent variables: Methods and guidelines for researchers in strategic management. In D. Ketchen & D. Bergh (Eds.), Research methodology in strategy and management: 87-134. Bingley, England: Emerald Group.

Bratton, K. A. 2005. Critical mass theory revisited: The behavior and success of token women in state legislatures. Politics & Gender, 1: 97-125.

Bugeja, M., Matolcsy, Z. P., & Spiropoulos, H. 2012. Is there a gender gap in CEO compensation? Journal of Corporate Finance, 18: 849-859.

Caliendo, M., & Kopeinig, S. 2008. Some practical guidance for the implementation of propensity score matching. Journal of Economic Surveys, 22: 31-72.

Carlson, K. D., & Wu, J. 2012. The illusion of statistical control: Control variable practice in management research. Organizational Research Methods, 15: 413-435.

Cook, A., & Glass, C. 2014a. Above the glass ceiling: When are women and racial/ethnic minorities promoted to CEO? Strategic Management Journal, 35: 1080-1089.

Cook, A., & Glass, C. 2014b. Women and top leadership positions: Towards an institutional analysis. Gender, Work and Organization, 21: 91-103.

Crossland, C., & Chen, G. 2013. Executive accountability around the world: Sources of cross-national variation in firm performance–CEO dismissal sensitivity. Strategic Organization, 11: 78-109.

Dedman, E., & Lin, S. W. J. 2002. Shareholder wealth effects of CEO departures: Evidence from the UK. Journal of Corporate Finance, 8: 81-104.

Eagly, A. H. 2007. Female leadership advantage and disadvantage: Resolving the contradictions. Psychology of Women Quarterly, 31: 1-12.

Eagly, A. H., & Karau, S. J. 2002. Role congruity theory of prejudice toward female leaders. Psychological Review, 109: 573-579.

Eagly, A. H., & Mladinic, A. 1989. Gender stereotypes and attitudes toward women and men. Personality and Social Psychology Bulletin, 15: 543-558.

Elsaid, E., & Ursel, N. D. 2018. Re-examining the glass cliff hypothesis using survival analysis: The case of female CEO tenure. British Academy of Management, 29: 156-170.

Elsesser, K. M. 2016. Gender bias against female leaders: A review. In M. Connerley & J. Wu (Eds.), Handbook on well-being of working women: 161-173. Dordrecht, The Netherlands: Springer.

Ertugrul, M., & Krishnan, K. 2011. Can CEO dismissals be proactive? Journal of Corporate Finance, 17: 134-151. Fama, E. F., & French, K. R. 1997. Industry costs of equity. Journal of Financial Economics, 43: 153-193. Farrell, K. A., & Whidbee, D. A. 2002. Monitoring by the financial press and forced CEO turnover. Journal of

Banking & Finance, 26: 2249-2276. Favaro, K., Karlsson, P., & Neilson, G. L. 2014. Study of CEOs, governance, and success: The value of getting CEO

succession right. Strategy and Business, 79: 1-16. Fee, C. E., Hadlock, C. J., Huang, J., & Pierce, J. R. 2018. Robust models of CEO turnover: New evidence on rela-

tive performance evaluation. Review of Corporate Finance Studies, 7: 70-100. Finkelstein, S., Hambrick, D. C., & Cannella, A. A. 2009. Strategic leadership: Theory and research on executives,

top management teams, and boards. New York: Oxford University Press. Fiordelisi, F., & Ricci, O. 2014. Corporate culture and CEO turnover. Journal of Corporate Finance, 28: 66-82. Flickinger, M., Wrage, M., Tuschke, A., & Bresser, R. 2016. How CEOs protect themselves against dismissal: A

social status perspective. Strategic Management Journal, 37: 1107-1117. Folkman, Z. 2012. A study in leadership: Women do it better than men. In K. Hurley & P. Shumway (Eds.), Real

women, real leaders: Surviving and succeeding in the business world: 165-169. Hoboken, NJ: Wiley. Fredrickson, J. W., Hambrick, D. C., & Baumrin, S. 1988. A model of CEO dismissal. Academy of Management

Review, 13: 255-270. Gao, H., Harford, J., & Li, K. 2012. CEO pay cuts and forced turnover: Their causes and consequences. Journal of

Corporate Finance, 18: 291-310. Geiler, P., & Renneboog, L. 2015. Are female top managers really paid less? Journal of Corporate Finance, 35:

345-369.

580 Journal of Management / April 2020

Glass, C., & Cook, A. 2016. Leading at the top: Understanding women’s challenges above the glass ceiling. The Leadership Quarterly, 27: 51-63.

Gomulya, D., & Boeker, W. 2014. How firms respond to financial restatement: CEO successors and external reac- tions. Academy of Management Journal, 57: 1759-1785.

Graffin, S. D., Boivie, S., & Carpenter, M. A. 2013. Examining CEO succession and the role of heuristics in early- stage CEO evaluation. Strategic Management Journal, 34: 383-403.

Grant, J. 1988. Women as managers: What they can offer to organizations. Organizational Dynamics, 16: 56-63. Guo, L., & Masulis, R. W. 2015. Board structure and monitoring: New evidence from CEO turnovers. Review of

Financial Studies, 28: 2770-2811. Gupta, V. K., Mortal, S. C., & Guo, X. 2018. Revisiting the gender gap in CEO compensation. Strategic Management

Journal, 39: 2036-2050. Haines, E. L., Deaux, K., & Lofaro, N. 2016. The times they are a-changing . . . or are they not? A comparison of

gender stereotypes, 1983–2014. Psychology of Women Quarterly, 40: 353-363. Haleblian, J., & Rajagopalan, N. 2006. A cognitive model of CEO dismissal: Understanding the influence of board

perceptions, attributions and efficacy beliefs. Journal of Management Studies, 43: 1009-1026. Harris, J., & Bromiley, P. 2007. Incentives to cheat: The influence of executive compensation and firm performance

on financial misrepresentation. Organization Science, 18: 350-367. Heilman, M. E. 1983. Sex bias in work settings: The lack-of-fit model. Research in Organizational Behavior, 5:

269-298. Heilman, M. E. 1997. Sex discrimination and the affirmative action remedy: The role of sex stereotypes. Journal of

Business Ethics, 16: 877-889. Heilman, M. E. 2001. Description and prescription: How gender stereotypes prevent women’s ascent up the organi-

zational ladder. Journal of Social Issues, 57: 657-674. Heilman, M. E. 2012. Gender stereotypes and workplace bias. Research in Organizational Behavior, 32: 113-135. Heilman, M. E., & Haynes, M. C. 2008. Subjectivity in the appraisal process: A facilitator of gender bias in work

settings. In E. Borgida & S. Fiske (Eds.), Beyond common sense: Psychological science in the courtroom: 127- 155. Malden, MA: Blackwell.

Hilger, S., Mankel, S., & Richter, A. 2013. The use and effectiveness of top executive dismissal. The Leadership Quarterly, 24: 9-28.

Huang, J., & Kisgen, D. J. 2013. Gender and corporate finance: Are male executives overconfident relative to female executives? Journal of Financial Economics, 108: 822-839.

Hubbard, T. D., Christensen, D. M., & Graffin, S. D. 2017. Higher highs and lower lows: The role of corporate social responsibility in CEO dismissal. Strategic Management Journal, 38: 2255-2265.

Huson, M. R., Parrino, R., & Starks, L. T. 2001. Internal monitoring mechanisms and CEO turnover: A long-term perspective. Journal of Finance, 56: 2265-2297.

Jenter, D., & Kanaan, F. 2015. CEO turnover and relative performance evaluation. Journal of Finance, 70: 2155-2184. Joshi, A., Neely, B., Emrich, C., Griffiths, D., & George, G. 2015. Gender research in AMJ: An overview of five

decades of empirical research and calls to action. Academy of Management Journal, 58: 1459-1475. Kanter, R. M. 1977. Some effects of proportions on group life: Skewed sex ratios and responses to token women.

American Journal of Sociology, 82: 965-990. Kark, R., & Eagly, A. H. 2010. Gender and leadership: Negotiating the labyrinth. In J. Chrisler & D. McCreary

(Eds.), Handbook of gender research in psychology: 443-468. New York: Springer. Kato, T., & Long, C. 2006. CEO turnover, firm performance, and enterprise reform in China: Evidence from micro

data. Journal of Comparative Economics, 34: 796-817. Kennedy, P. 2003. A guide to econometrics (5th ed.). Cambridge, MA: MIT Press. Ketokivi, M., & McIntosh, C. N. 2017. Addressing the endogeneity dilemma in operations management research:

Theoretical, empirical, and pragmatic considerations. Journal of Operations Management, 52: 1-14. King, E. B., Hebl, M. R., George, J. M., & Matusik, S. F. 2010. Understanding tokenism: Antecedents and conse-

quences of a psychological climate of gender inequity. Journal of Management, 36: 482-510. King, T., Srivastav, A., & Williams, J. 2016. What’s in an education? Implications of CEO education for bank per-

formance. Journal of Corporate Finance, 37: 287-308. Koenig, A. M., Eagly, A. H., Mitchell, A. A., & Ristikari, T. 2011. Are leader stereotypes masculine? A meta-

analysis of three research paradigms. Journal of Applied Psychology, 137: 616-642. Krishnan, H. A. 2009. What causes turnover among women on top management teams? Journal of Business

Research, 62: 1181-1186.

Gupta et al. / Gender and CEO Dismissal 581

Landy, F. J. 2008. Stereotypes, bias, and personnel decisions: Strange and stranger. Industrial and Organizational Psychology, 1: 379-392.

Lee, P. M., & James, E. H. 2007. She’-e-os: Gender effects and investor reactions to the announcements of top executive appointments. Strategic Management Journal, 28: 227-241.

Lehn, K. M., & Zhao, M. 2006. CEO turnover after acquisitions: Are bad bidders fired? Journal of Finance, 61: 1759-1811.

Lel, U., & Miller, D. P. 2008. International cross-listing, firm performance, and top management turnover: A test of the bonding hypothesis. Journal of Finance, 63: 1897-1937.

Leslie, J. B., & Van Velsor, E. 1996. A look at derailment today: North America and Europe. Greensboro, NC: Center for Creative Leadership.

Leslie, L. M., Manchester, C. F., & Dahm, P. C. 2017. Why and when does the gender gap reverse? Diversity goals and the pay premium for high potential women. Academy of Management Journal, 60: 402-432.

Leung, S. 2014. Why do female CEOs get fired more often than male ones? Boston Globe, October 22. Retrieved from https://www.bostonglobe.com/magazine/2014/10/22/why-female-ceos-get-fired-more-often-than-male- ones/h5xII63gdClkhVn8innf4J/story.html. Accessed October 17, 2018.

Li, W., Lu, Y., Makino, S., & Lau, C. M. 2017. National power distance, status incongruence, and CEO dismissal. Journal of World Business, 52: 809-818.

Loden, M. 1985. Feminine leadership or how to succeed in business without being one of the boys. New York: Times Books.

Martin, J. A., & Combs, J. G. 2011. Better sooner than later: What triggers early CEO dismissal? Academy of Management Perspectives, 25(2): 82-83.

Mobbs, S. 2013. CEOs under fire: The effects of competition from inside directors on forced CEO turnover and CEO compensation. Journal of Financial and Quantitative Analysis, 48: 669-698.

Murphy, K. J. 1999. Executive compensation. Handbook of Labor Economics, 3: 2485-2563. Norton, E. C., Wang, H., & Ai, C. 2004. Computing interaction effects and standard errors in logit and probit mod-

els. Stata Journal, 4: 154-167. Oliver, A. G., Krause, R., Busenbark, J. R., & Kalm, M. 2018. BS in the boardroom: Benevolent sexism and board

chair orientations. Strategic Management Journal, 39: 113-130. Park, J. H., Kim, C., & Sung, Y. D. 2014. Whom to dismiss? CEO celebrity and management dismissal. Journal of

Business Research, 67: 2346-2355. Parrino, R. 1997. CEO turnover and outside succession: A cross-sectional analysis. Journal of Financial Economics,

46: 165-197. Peters, F. S., & Wagner, A. F. 2014. The executive turnover risk premium. Journal of Finance, 69: 1529-1563. Reingold, J. 2016. Why top women are disappearing from corporate America. Fortune, September 9. Retrieved

from http://fortune.com/women-corporate-america/. Accessed October 17, 2018. Rosenbaum, P. R. 2000. Propensity score. Wiley StatsRef: Statistics Reference Online. Hoboken, NJ: Wiley. Rosette, A. S., & Tost, L. P. 2010. Agentic women and communal leadership: How role prescriptions confer advan-

tage to top women leaders. Journal of Applied Psychology, 95: 221-235. Rudman, L. A., & Glick, P. 2001. Prescriptive gender stereotypes and backlash toward agentic women. Journal of

Social Issues, 57: 743-762. Ryan, M. K., & Haslam, S. A. 2005. The glass cliff: Evidence that women are over-represented in precarious leader-

ship positions. British Journal of Management, 16: 81-90. Ryan, M. K., & Haslam, S. A. 2007. The glass cliff: Exploring the dynamics surrounding the appointment of women

to precarious leadership positions. Academy of Management Review, 32: 549-572. Ryan, M. K., Haslam, S. A., Hersby, M. D., & Bongiorno, R. 2011. Think crisis–think female: The glass cliff and

contextual variation in the think manager–think male stereotype. Journal of Applied Psychology, 96: 470-484. Ryan, M. K., Haslam, S. A., & Kulich, C. 2010. Politics and the glass cliff: Evidence that women are preferentially

selected to contest hard-to-win seats. Psychology of Women Quarterly, 34: 56-64. Ryan, M. K., Haslam, S. A., Morgenroth, T., Rink, F., Stoker, J., & Peters, K. 2016. Getting on top of the glass cliff:

Reviewing a decade of evidence, explanations, and impact. The Leadership Quarterly, 27: 446-455. Sandberg, S. 2013. Lean in: Women, work, and the will to lead. New York: Knopf Doubleday. Schein, V. E. 2001. A global look at psychological barriers to women’s progress in management. Journal of Social

Issues, 57: 675-688.

582 Journal of Management / April 2020

Shen, W., & Cannella, A. A. 2002. Revisiting the performance consequences of CEO succession: The impacts of successor type, post-succession senior executive turnover, and departing CEO tenure. Academy of Management Journal, 45: 717-733.

Shipman, J. E., Swanquist, Q. T., & Whited, R. L. 2016. Propensity score matching in accounting research. Accounting Review, 92: 213-244.

Spector, P. E., & Brannick, M. T. 2011. Methodological urban legends: The misuse of statistical control variables. Organizational Research Methods, 14: 287-305.

Tuggle, C. S., Schnatterly, K., & Johnson, R. A. 2010. Attention patterns in the boardroom: How board composi- tion and processes affect discussion of entrepreneurial issues. Academy of Management Journal, 53: 550-571.

Underdahl, S. C., Walker, L. S., & Woehr, D. J. 2014. Gender and perceptions of leadership effectiveness: A meta- analysis of contextual moderators. Journal of Applied Psychology, 99: 1129-1145.

Wiersema, M. F., & Bowen, H. P. 2009. The use of limited dependent variable techniques in strategy research: Issues and methods. Strategic Management Journal, 30: 679-692.

Wiersema, M. F., & Zhang, Y. 2011. CEO dismissal: The role of investment analysts. Strategic Management Journal, 32: 1161-1182.

Wynen, J., Beeck, S., & Ruebens, S. 2015. The nexus between gender and perceived career opportunities. Public Personnel Management, 44: 375-400.

Yukl, G., & Chavez, C. 2002. Influence tactics and leader effectiveness. In L. Neider & C. Schriesheim (Eds.), Leadership: 139-165. Greenwich, CT: Information Age.

Zhang, Y. 2006. The presence of a separate COO/president and its impact on strategic change and CEO dismissal. Strategic Management Journal, 27: 283-300.

Zhang, Y. 2008. Information asymmetry and the dismissal of newly appointed CEOs: An empirical investigation. Strategic Management Journal, 29: 859-872.

Zhang, Y., & Qu, H. 2016. The impact of CEO succession with gender change on firm performance and successor early departure: Evidence from China’s publicly listed companies in 1997–2010. Academy of Management Journal, 59: 1845-1868.

Zhang, Y., & Rajagopalan, N. 2010. Once an outsider, always an outsider? CEO origin, strategic change, and firm performance. Strategic Management Journal, 31: 334-346.