Abstract
Firms have appointed a significant number of female chief marketing officers (CMOs) over the past decade. However, the question of how female CMOs differ from their male counterparts is yet to be explored. This research uses a multimethod approach to examine when and why female CMOs’ marketing decisions differ from those of male CMOs. In Study 1, the authors use secondary data to examine the effect of CMO gender on multiple marketing decisions and find that female CMOs make less risky decisions. Further, the authors find evidence that female CMOs’ risk-taking behavior is contingent on structural, organizational, and environmental factors (CEO gender, relative firm performance, and demand uncertainty). In Study 2, the authors employ the MarkStrat simulation, in which participants assume the role of CMO, to test the main finding from Study 1 in a controlled setting and provide evidence for the differential effect of gender on radical versus incremental new product introductions. In Study 3, the authors examine survey data to find evidence for the underlying mechanisms (overconfidence, failure avoidance orientation, and scrutiny) behind female CMOs’ lesser risk-taking behavior.
Keywords
Gender is one of the most salient and influential aspects of a person's identity. It affects a person's appearance, activities, financial decisions, and career paths. For example, owing to their overconfidence, men prefer more competitive environments (Niederle and Vesterlund 2007) and make more risky investment decisions (Jianakoplos and Bernasek 1998). Gender differences seem to affect not only individual decisions but also firm-level decisions. For example, Huang and Kisgen (2013) find that male CEOs engage in acquisitions and issue debt more often than female CEOs. These findings indicate that gender differences transmit to even the highest decision-making units in firms.
The number of women in top management teams (TMTs) has increased over the past two decades. Among Fortune 500 companies, the number of female CEOs has jumped significantly from just a single female CEO in 1998 to 41 in 2021 (Hinchliffe 2021). Not surprisingly, the empirical research on women's representation in TMTs has received attention only in the last decade. Extant research has studied gender differences at the CEO/CFO level (see Table W1 in Web Appendix A) and female representation in the TMT (Dezsö and Ross 2012). However, there is substantial variation in female representation across firms’ different functional areas. For instance, in our sample, women's representation in TMTs is less than 10% in technology and operations areas but nearly 40% in the marketing area. The number of CMO appointments was almost equitable, with 48% of women appointees in 2019 (Schultz 2019). Further, 31 of Forbes’s 50 most influential CMOs in 2019 were women (Rooney 2019). The trend of companies appointing more female CMOs is prevalent not only in developed countries but also in growth markets such as India (Singh and Zachariah 2017).
There is some anecdotal evidence/belief that female CMOs are better than male CMOs (Dooley 2020). In this context, we believe it is imperative to examine when and why gender differences affect CMOs’ marketing decisions, which, in turn, influence firm performance. Specifically, we answer the following research questions:
In this article, we investigate these research questions with the help of three studies. In Study 1, we use secondary data to examine the effect of CMO gender on marketing decisions (advertising spending, real earnings management [REM], and new product introductions [NPIs]; RQ1). In the same study, we examine the moderating effect of structural (CEO gender), organizational (relative firm performance), and environmental (demand uncertainty) factors. In Study 2, we examine the effect of the CMO's gender on radical and incremental product introductions and the role of risk taking as a mediator. In Study 3, we examine the underlying mechanisms (overconfidence, fear of failure, and workplace scrutiny) leading to differential risk taking between male and female executives. We summarize the three studies in Figure 1.
Summary of the Studies.
We make three contributions to the literature. First, we respond to the research gaps identified by You et al. (2020) and Whitler et al. (2020). Specifically, while previous research has examined the effect of CMO presence on firm outcomes, the characteristics of CMOs have yet to be thoroughly investigated (see Table 1). We address this gap by examining a highly salient demographic variable (gender). Moreover, while prior research on CMOs has examined only firm performance, we study more proximal marketing outcomes such as NPIs.
Literature Review.
Second, we contribute to the literature on executive gender (not just on the CMO). Prior studies rely primarily on male overconfidence to theoretically explain higher risk taking among male executives (see Table W1 in Web Appendix A). To the best of our knowledge, our research is the first to provide empirical evidence for fear of failure and scrutiny as mechanisms leading to lower risk taking among female executives (Study 3). Moreover, extant studies do not provide empirical evidence for risk taking as the mediating mechanism in the relationship between executive gender and decisions. Using MarkStrat simulation, we address this gap in Study 2. Importantly, we find that the indirect effect (through risk aversion) of CMO gender on radical NPIs is negative. However, the indirect effect on incremental product introductions is positive. These novel insights enhance our understanding of the role of risk-taking behavior in executive decisions.
Third, we identify the moderating conditions that influence female CMOs’ marketing decisions. Although, on average, female CMOs undertake less risky marketing decisions than their male counterparts, their risk aversion attenuates if their firm has a female CEO or demonstrates relatively better performance. In contrast, female CMOs’ risk aversion is exacerbated when there is higher demand uncertainty.
In the following section, we discuss the theoretical background of gender differences, followed by extant research on CMOs. Next, we develop the theoretical framework and hypotheses. We then present three empirical studies to test the hypotheses and provide evidence for the mechanisms. We conclude with a discussion of the theoretical and managerial implications of the findings.
Gender Differences
Theories of Gender Differences
The central premise of this study is that men and women have inherent differences in how they perceive and relate to their environment. We draw on agency–communion theory (Bakan 1966) and self-construal theory (Cross and Madson 1997) to understand gender differences. Agency–communion theory posits that individuals exist in two modalities: agency and communion (Bakan 1966). Agency refers to an individual's attempt for independence, to be defined as a separate entity from others. In contrast, communion considers an individual as part of a larger social unit. Whereas agency involves self-orientation, self-assertion, and self-enhancement, communion refers to group orientation, cooperation, and social acceptance (Bakan 1966). Though men and women possess both agentic and communion traits, men, on average, score higher on agentic traits, while women score higher on communion traits (Helgeson 1994).
The self-construal theory of gender differences argues that men and women structure their sense of self differently. Like the agency–communion theory, the self-construal theory suggests that men have an independent sense of self. In contrast, women have an interdependent self-construal (Cross and Madson 1997). Men, with an independent self-construal, focus on distinguishing themselves from others. Pursuing an independent self translates into psychological traits such as assertiveness, autonomy, and competition. In contrast, women have an expanded sense of self, which includes themselves and close others. Besides their well-being, self-enhancement for women involves the well-being of important others. Differences in individuals’ self-construal help explain gender differences in cognition, emotions, and decision making (Cross and Madson 1997).
The emergence of gender differences, as explained by agency–communion or self-construal theories, could result from evolutionary, biological, or social processes. From an evolutionary perspective, gender differences create a conducive environment for reproduction, which is essential for the species’ survival. For instance, this view explains aggressiveness and competitiveness as inherently masculine traits since evolutionary mating patterns reveal that males had to compete to be selected as potential mating partners by females (Geary 1999). Similarly, women can better process facial expressions because they were responsible for nurturing their offspring after childbirth (Alexander 2003). The biological perspective emphasizes the role of genetics and hormones in explaining gender differences. For example, research suggests that the X chromosome is partially responsible for gender differences in social skills and spatial ability (Skuse et al. 1997), and hormones such as androgens aid in developing masculine traits (Berenbaum 2002). 1 The social perspective views gender differences as a byproduct of systemic social conditioning, which begins in childhood. Children absorb gender-stereotyped norms and expectations from family, school, and society through rewards and punishments or observational learning (Bussey and Bandura 1999). These gender norms become part of an individual's mental schema, which they must abide by for cognitive consistency and which, over time, manifest in gender differences.
Gender and Risk Taking
One of the critical gender differences that may percolate in marketing decisions is how men and women perceive risk and return. Byrnes, Miller, and Schafer (1999) conducted a meta-analysis of gender differences in risk taking by studying 150 papers across various measures of risk taking such as hypothetical choice tasks (e.g., choice dilemma/framing), self-reported behavior (e.g., driving, drug use, sexual activities) and observed behavior (e.g., gambling, risky experiments). The authors find significant gender differences in risk-taking decisions across most tasks, with men taking greater risks than women. Gender differences in risk taking are also evident in financial decision making. Charness and Gneezy (2012) study gender differences in financial risk taking and find consistent evidence across 15 experiments suggesting that men invest more in risky options than women. Jianakoplos and Bernasek (1998) find that single women hold smaller proportions of risky assets than single men. Gender differences also lead to differences in the TMT's decision making. For example, female CFOs issue less debt, have wider bounds on earnings estimates, and are more likely to exercise stock options early (Huang and Kisgen 2013). Similarly, female CEOs tend to make less risky financing and investment choices than male CEOs (Faccio, Marchica, and Mura 2016).
In our article, we further examine the mechanisms that manifest in differential risk-taking behaviors. Specifically, we anticipate women to be more risk averse than men because of overconfidence among men (Mohr 2014), along with a greater failure avoidance orientation (Nelson et al. 2013) and a higher fear of scrutiny (Brescoll, Dawson, and Uhlmann 2010) among women.
Overconfidence
While both men and women overestimate their ability to affect outcomes positively, men are significantly more overconfident than women (Barber and Odean 2001). This view aligns with agency–communion theory, where overconfidence is linked to unmitigated agency (Helgeson and Fritz 2000). Further, since men have independent self-construal, they are more likely to exaggerate or overestimate their abilities relative to their actual performance (Cross and Madson 1997). This is a means of self-enhancement to differentiate themselves from others and assert their superiority (Cross and Madson 1997). Such overconfidence may explain why men participate in competitions more than women (Niederle and Vesterlund 2007) and apply for more jobs or promotions, even when they are not fully qualified (Mohr 2014).
Failure avoidance orientation
The gender differences in attribution of success and failure can lead to differences in failure avoidance orientation. Evidence suggests that women have lower expectancies of success than men (Dweck and Licht 1980). Consequently, women are likely to consider success unexpected/temporary and may attribute it to external factors such as luck. In addition, women may attribute failure to intrinsic factors such as lack of innate ability. Conversely, men attribute failure to luck or task difficulty but attribute success to their own ability (Deaux and Farris 1977). Since women perceive failure to reflect their (lack of) innate ability, they may actively try to preempt failures. In addition, women's communal orientation makes them prone to internalize social biases (e.g., lower competence of women) (Beyer 1990). Therefore, women may have a greater failure avoidance orientation than men (Nelson et al. 2013).
Fear of scrutiny
Extant literature suggests that in addition to facing a glass ceiling that impedes their rise to top management positions, women also find themselves perched on a “glass cliff,” which means they are more likely to fall from their positions than men (Ryan and Alexander Haslam 2005). Brescoll, Dawson, and Uhlmann (2010) suggest that small mistakes in jobs are especially damaging for individuals who have achieved a top-level position in a gender-incongruent occupation. They argue that while competence is assumed for gender-congruent leaders, a small mistake creates ambiguity about the competence of the gender-incongruent leader and results in a loss of status, much more than for their gender-congruent counterparts. Kram and Hampton (1998) provide evidence of the “visibility-vulnerability spiral,” which they define as the heightened scrutiny women leaders face due to their minority distinctiveness. The authors argue that the minority distinctiveness of women leaders gives them less time and fewer opportunities to learn from their failures, as their actions are constantly under scrutiny.
CMO Literature
Extant literature has explored the impact of CMOs on two broad categories of outcomes. First, most studies examine how CMOs’ presence and characteristics (e.g., education) affect firms’ financial performance or value. For example, Germann, Ebbes, and Grewal (2015) argue that firms with a CMO report a 15% higher Tobin's q than those without a CMO. Further, Wang, Saboo, and Grewal (2015) find positive effects of the CMO's education and outsider status on cumulative abnormal stock returns around CMO succession events and a U-shaped relationship between CMO experience and firm value. Homburg et al. (2014) find that new ventures have a higher likelihood of funding if their CMOs have master of business administration (MBA) degrees from prestigious universities and greater marketing, industry, and start-up experiences. Kim et al. (2016) suggest that a CMO's equity compensation positively influences a firm's market value.
Second, relatively few studies examine the impact of the CMO on marketing decisions/outcomes. Whitler, Morgan, and Rego (2020) surveyed 390 CMOs about their responsibilities and identified brand and marketing strategy, marketing communications, marketing research, pricing, channel management, customer relationship management, and corporate strategy as part of CMOs’ core and expanded responsibilities. However, the effect of the CMO has been examined on only a few marketing outcomes. Brower and Nath (2018) find that CMO presence improves a firm's market orientation, but only when its TMT has high marketing experience. Kaur, Ramaswami, and Bommaraju (2021) find that the presence of a CMO discourages unplanned marketing expenditure cuts but has no impact on unplanned sales promotions.
Extant CMO research also highlights that there is no one right way to manage a firm and that CMOs’ decisions are embedded in the structural, organizational, and environmental context specific to each firm (Zeithaml and Zeithaml 1988). Research has examined structural factors relating to how firms organize their TMTs and how CMOs interact with other TMT members. For instance, Nath and Mahajan (2011) find that greater TMT marketing experience negatively impacts CMO power in the TMT. Nath and Bharadwaj (2020) examine C-suite dyads with CMOs and other functional heads (e.g., chief experience officers [CXOs], COOs) and find synergies that improve firm performance under varying circumstances. While investigating organizational factors, Boyd, Chandy, and Cunha (2010) find that the presence of a powerful customer limits the CMO's managerial discretion and adversely affects the firm value that a CMO can generate. They also report that firm-specific factors such as firm performance attenuate the negative effect. In addition, CMO research has examined the role of environmental factors. For example, Nath and Mahajan (2017) examine the role of industry growth and instability as antecedents of CMO turnover.
Theoretical Framework and Hypotheses
Theoretical Framework
Our study examines the differences between female and male CMOs in their risk-taking behavior. Researchers differ in the way they define risk taking. Some define it to capture a broad set of activities, and others define it more narrowly, to suit the specific context (Byrnes, Miller, and Schafer 1999). We use the broad definition given by Furby and Beyth-Marom (1992), in which risk taking is any behavior that could lead to more than one outcome, and some of these outcomes are undesirable from the risk taker's perspective. In other words, risk taking involves implementing decisions that could lead to negative consequences.
We use three risk-taking measures: advertising intensity, REM, and NPIs. Extant research suggests that advertising creates intangible market-based assets that positively affect sales, market share, and profitability. However, advertising effectiveness is hard to establish. Anecdotally, John Wanamaker is believed to have said, “Half the money I spend on advertising is wasted; the trouble is, I don’t know which half” (Bradt 2016). Sethuraman, Tellis, and Briesch (2011) find that short-term and long-term advertising elasticity has declined over time and is substantially lower than it was in the 1980s. Given declining advertising elasticity alongside advertising's effect on intangible assets, which are difficult to quantify, management often questions advertising's impact on revenues/profits (Luo and De Jong 2012) and asks whether it is justified to divert critical marketing resources to advertising. In addition, higher advertising intensity can amplify competition and result in combative advertising among competing firms (Chen et al. 2009). Thus, spending on advertising is risky because it could lead to more than one outcome (e.g., higher vs. lower returns or no returns), some of which (e.g., lower or negative returns, higher competition) could be undesirable.
Real earnings management (REM) refers to cutting discretionary expenses to boost the current period's earnings. REM obscures information about a firm's underlying performance and is associated with managers’ hoarding bad news and camouflaging poor performance. While REM may boost the current period's earnings, it could lead to significant negative returns in the long term (Mizik 2010). It is also associated with greater stock market underperformance (Kothari, Mizik, and Roychowdhury 2016). Thus, REM is a good measure of risk taking because it can have negative consequences. Lastly, while NPIs can be profitable, product failures are expensive since firms cannot recover the development and marketing costs incurred during the launch (Hitsch 2006). Thus, NPIs are a good measure of risk-taking behavior because multiple outcomes are possible (profits through successful products and losses through failed products) and one of the outcomes (product failure) is undesirable.
In addition, we study these marketing decisions because they (1) are within the purview of the CMO's responsibilities (Whitler et al. 2020), (2) are crucial to marketing (Mizik 2010), (3) can be objectively measured from secondary data sources (Kothari, Mizik, and Roychowdhury 2016; Warren and Sorescu 2017a, b), and (4) can impact the firm's financial performance (Mizik 2010; Saboo, Chakravarty, and Grewal 2016).
As we have noted, prior literature highlights that CMOs’ decisions are embedded within a structural, organizational, and environmental context specific to each firm. Thus, we identify one moderating variable within each contextual category. We examine the CEO's gender as an important structural factor, which may impact the CMO's decisions because (1) the CMO directly reports to the CEO and (2) prior research suggests that gender congruity plays an important role in workplace dynamics (Kunze and Miller 2017; Tate and Yang 2015). We choose relative firm performance as an organizational factor because extant research indicates that a manager's risk-taking behavior is influenced by a firm's relative performance (Han et al. 2019). We choose demand uncertainty as an environmental factor because demand generation falls directly under the purview of the CMO, and prior research suggests that uncertainty influences individual risk-taking behavior (Marquis and Joseph Reitz 1969). Therefore, we anticipate that CEO gender, relative firm performance, and demand uncertainty may moderate CMOs’ decisions.
Hypotheses
We expect female CMOs to be relatively more cautious when making marketing decisions for the following reasons. First, overconfident executives may overestimate their ability to affect future returns, thereby expecting a higher return (Larwood and Whittaker 1977). Alternatively, they may expect their return prediction to be certain, consequently underestimating the downside. Thus, overconfident executives may overestimate the net present value of projects and expand their pool of project choices by including some projects with negative net present value (Huang and Kisgen 2013). Since female CMOs are less overconfident than male CMOs, they are less likely to overestimate the expected returns from marketing decisions such as NPIs. Second, women are more likely to attribute marketing failures to their innate ability than to external conditions and, thus, may be more cautious in their marketing decisions. He, Inman, and Mittal (2008) find evidence that women prioritize loss avoidance over achieving gains. Thus, a female CMO may be more cautious in marketing decisions such as advertising spending. Third, higher scrutiny of female executives may make them more accountable for their decisions and increase the negative consequences in case of failure. For instance, female CMOs may be more skeptical of engaging in REM since it can be detrimental to their department's long-term performance, making them susceptible to greater scrutiny. In summary, female marketing executives are less likely to inflate the benefits of marketing investments, have a heightened need to avoid failures, and experience more significant consequences in case of marketing failures. Thus,
When a firm has a female CEO, we expect female CMOs to make more risky marketing decisions for the following reasons. First, we expect female CMOs to be more confident in the presence of female CEOs as compared with male CEOs due to positive spillovers in the presence of a female leader (Kunze and Miller 2017; Tate and Yang 2015). For example, Yan et al. (2022) find that female subordinates are more likely to assert their viewpoint under female leadership. Second, female CMOs are likely to exhibit less failure avoidance under a female CEO for the following reasons. Cohen and Swim (1995) find that when women perceive themselves as tokens in a team (i.e., when they are a numerical minority for improving diversity), they anticipate a more stereotypical performance evaluation. Since the stereotypical notion is that female leaders are less competent than their male counterparts, women leaders may have a heightened need to avoid failures. However, when the firm has a female CEO, female CMOs may no longer perceive themselves as token players, which may attenuate the need to “play it safe” or avoid failures. Third, we anticipate that female CMOs may have less fear of scrutiny when working with a female CEO. Specifically, extending the statistical discrimination model (Phelps 1972), which posits that superiors are better at evaluating productivity signals of employees of the same gender, Flabbi et al. (2019) suggest that female CEOs may be more precise in processing information or inferring the ability/productivity of female employees. This is reflected in a better assessment of female productivity and higher wages for women at the top of the wage distribution when employed by a female CEO compared with a male CEO (Flabbi et al. 2019). If female CEOs can better assess female CMOs’ decisions, it may reduce the latter's fear of scrutiny. Thus,
When a firm performs relatively better than other firms in the industry, we expect female CMOs to take more risky marketing decisions for the following reasons. First, we anticipate that the confidence gap between male and female CMOs would be narrower in well-performing firms since women adjust more strongly to performance feedback than men (Berlin and Dargnies 2016). Specifically, Berlin and Dargnies (2016) find that once individuals realize that their performance is above the median, men continue to be more confident than women, but the confidence gap between them reduces. Second, relatively higher firm performance may attenuate the downside loss from marketing decisions such as product failures. Firms with higher performance accumulate greater tangible and intangible resources, which gives them more flexibility in implementing growth strategies and insulates them from negative repercussions in case of potential failures (Han, Mittal, and Zhang 2017). Since women usually have greater failure avoidance orientation than men and respond to performance feedback more strongly, we anticipate that when a firm performs better, the fear of failure experienced by female CMOs may attenuate more than that of male CMOs. Third, when firms perform poorly, there is ambiguity about management's competence. In such a scenario, while both male and female CMOs have greater scrutiny, we anticipate female leaders to be more likely to face greater scrutiny than their male counterparts because of their minority distinctiveness (Kram and Hampton 1998) and gender incongruity in leadership roles (Brescoll, Dawson, and Uhlmann 2010). In contrast, when relative firm performance is higher, the ambiguity surrounding female CMOs’ competency is alleviated. Therefore, we anticipate that female CMOs face less scrutiny when relative firm performance is high.
When demand uncertainty is high, the link between marketing decisions and marketing performance is ambiguous. We anticipate that higher ambiguity leads to higher risk aversion among female CMOs for the following reasons. First, Barber and Odean (2001) suggest that male overconfidence is exaggerated when uncertainty is high. Similarly, Lenney (1981) finds that women have lower self-confidence than men when feedback from the environment is ambiguous. Since the decision–performance link is noisy under demand uncertainty, we anticipate that the confidence gap between female and male CMOs will widen. Second, in ambiguous situations, women are more uncertain than men about the appropriateness of their decisions (Schwartz and Clausen 1970), and thus women's lower success expectancies may worsen in such situations. Therefore, female CMOs’ failure avoidance is likely greater under high demand uncertainty. Third, when facing environmental ambiguity, other TMT members and the board of directors will likely rely on conventional ideas and heuristics to evaluate a CMO (Thebaud and Sharkey 2016). As a result, gender stereotypes are more frequently applied under conditions of high uncertainty (Ridgeway and Correll 2004). For example, when performance quality is ambiguous, female candidates are perceived as less competent than their male counterparts (Foschi 2000). Thus, female CMOs may face more scrutiny when demand uncertainty is higher.
Study 1
Sample
To test the hypotheses, we construct a panel data set from 2009 to 2019 by merging three proprietary databases, BoardEx, RavenPack, and Compustat. We use BoardEx to identify the demographic characteristics and job functions of executives. We obtain data on NPIs from RavenPack and gather firm-level accounting and financial data from Compustat. We first identify 1,840 firms (13,277 firm-years) from BoardEx, with at least one marketing executive in the TMT (described in detail next). We consider only these firms in our sample, as our hypotheses cannot be tested on firms with no marketing presence in the TMT. We then merge the TMT information of these firms with Compustat data to arrive at a sample used in models where advertising intensity and REM (7,482 firm-years) are the dependent variables. Further, we obtain data on NPIs (2,040 firm-years) from RavenPack. We provide detailed sampling steps in Web Appendix D. We note that our sample size is comparable to or higher than the sample sizes of previous CMO studies.
Independent Variable: CMO Gender
We use BoardEx to identify the CMO's presence and gender. Not all marketing heads may have the title of chief marketing officer (CMO). However, in line with previous research, we use the term CMO to indicate the head of marketing (Germann, Ebbes, and Grewal 2015; Nath and Mahajan 2008).
Extant marketing research uses two sources to identify the presence of a CMO: Execucomp and 10-K reports. The Execucomp database provides compensation information for the highest-paid executives. By regulation, all publicly listed companies need to disclose the compensation information for the top five highest-paid executives. However, some companies choose to present compensation information for more than five executives, and the same company may report compensation for a different number of executives across the years. Thus, there is bias induced in identifying the presence of a CMO. For example, if a company chooses to report only the top five highest-paid executives, and the CMO is the seventh-highest-paid, one may conclude that the firm does not have a CMO. Moreover, if the CMO salary increases to the top five in subsequent periods, one may conclude that the firm now has a CMO but did not have one in the past.
To avoid such issues, scholars have relied on 10-K reports, which contain information on executive officers (e.g., Germann, Ebbes, and Grewal 2015), and have treated the list of executive officers as equivalent to TMT. An executive officer is one who has policy making authority. However, firms are free to use discretion in determining what constitutes policy making. As Nath and Bharadwaj (2020) note, this is subjective and not a straightforward matter for firms. A consequence of this is that it may induce artificial variation in listing a CMO on 10-K reports. For example, Maggie Lower, who worked at TrueBlue (based on BoardEx data and verified on her LinkedIn page) as the CMO from 2018 to 2020, was not listed on the 10-K (or annual reports/proxy statement) in any of those years. As another example, Deborah Bussière, Global CMO at Broadridge from 2017 to 2021, was not listed in the 10-K from 2017 to 2019 but was listed in 2020. Thus, a researcher using 10-Ks would code that Broadridge did not have a CMO between 2017 and 2019 but had one in 2020 (for other examples, see Table W5 in Web Appendix E). Another major challenge with 10-K reports is that gathering information for many firm-year combinations is highly laborious. Thus, not surprisingly, the sample sizes of previous CMO studies have comprised only 125–175 firms, which generally tend to be large companies.
We rely on BoardEx to overcome the challenges of reporting uncertainty, sample size, and sample selection bias toward large firms. It gathers data from various public sources, including regulatory filings, annual reports, proxy statements, company websites, and press and regulatory news wires. It also tracks and uses stock exchange and national/state registries. BoardEx is used in finance and management literature to study TMTs (e.g., Fu, Tang, and Chen 2020). In marketing, Tavassoli, Sorescu, and Chandy (2014) use BoardEx to identify the company boards on which executives serve as directors.
BoardEx provides information on the board of directors and executives. However, it does not define/identify who constitutes the TMT. Thus, we took the following three steps to identify TMT and the presence of CMO, if any. First, in line with prior literature (Feng, Morgan, and Rego 2015), we define the following job titles as belonging to the TMT: CEO, COO, CFO, CMO, chief or head of various functions, president, executive vice president (VP), senior VP, and VP. Second, we observe that, unlike other functions that are singularly defined (such as finance, human resources, and legal), marketing comprises multiple subfunctions. Following prior literature, we include 32 subfunctions (see Web Appendix D) under the umbrella of “marketing” (Feng, Morgan, and Rego 2015). A few examples of the subfunctions are advertising, innovation, growth, product development, brand management, and channels. This procedure helps us identify, for each job title, whether the title is a TMT role and/or a marketing role. We consider a title to be marketing TMT if the title falls under both TMT (e.g., president) and marketing (e.g., advertising). Next, we use Feng, Morgan, and Rego’s (2015) hierarchy of titles (chief, head, president, executive VP, senior VP, and VP, respectively) to identify the marketing head. As noted previously, in line with prior research, we use the term CMO to describe the marketing head. In the final step, we identify the CMO’s gender and code CMO_Gender as 1 if the gender is female and 0 otherwise.
Dependent Variables
We assess the impact of gender differences on advertising intensity, REM, and NPIs.
Advertising intensity
We rely on Compustat to gather data on advertising expenses and total revenue. Advertising intensity is the ratio of advertising expenses to total revenue.
REM
We gather data to measure REM from Compustat. To measure abnormal changes in advertising expenses and profitability (measured by return on assets [ROA]), we need an estimate of normal or expected levels of these variables. Following prior literature (Kothari, Mizik, and Roychowdhury 2016; Mizik 2010), we use the following fixed effects first-order autoregressive panel data models to estimate the expected levels of ROA and advertising spending:
We estimate forecast errors to measure the deviation of the variable from the expected value. We code a dummy variable, REMAD, using the forecast errors to identify myopic firms. Specifically, if
NPIs
We gather data on NPIs from RavenPack News Analytics. The database collects information pertaining to several thousand companies, organizations, people, events, currencies, and commodities. RavenPack uses sophisticated text analytics tools to collate a structured data set from Dow Jones Newswires, press releases, and the web. Warren and Sorescu (2017a, b) use RavenPack to gather new product launch information. We follow the same approach, which we briefly explain next.
RavenPack categorizes firm-specific news using unique tags such as acquisitions, campaign ads, product release, product recall, and many more, which are called events. The database provides two crucial pieces of information for each event: the event's relevance to the firm and the novelty of the news. A relevance score ranging from 0 to 100 indicates how strongly a firm is connected to the underlying news. For example, a relevance score of 100 means that the entity is the most prominent in the news and plays a vital role in the context. Following previous work in marketing (Warren and Sorescu 2017a, b), we have chosen events with a relevance score of 100. In addition to the relevance score, RavenPack also provides a novelty score for each event. An event novelty score of 100 indicates that the news story is the first in the chain of news items covering an event. Setting an event novelty score of 100 ensures that the count of NPIs is free from duplication because of multiple news items covering the same product release event. Thus, we ensure that news items on each product release relate to the firm (relevance score of 100) and are not duplicated (event novelty score of 100). We measure a firm's NPIs as the count of unique news items pertaining to product releases aggregated annually.
Moderating Variables
CEO gender
We identify the CEO's gender using BoardEx and code it as a dummy variable that takes a value of 1 if the CEO is female and 0 if the CEO is male.
Relative firm performance
We evaluate the focal firm's relative performance compared with competing firms in the same industry (as defined by the two-digit Standard Industrial Classification [SIC] code). Using ROA to measure firm performance, we define relative firm performance as the difference between the focal firm's ROA and the average industry ROA.
Demand uncertainty
We follow Keats and Hitt’s (1988) model to measure demand uncertainty. First, we aggregate the revenues of all firms operating in the same industry each year to calculate annual industry revenues. Next, we employ rolling regressions over a three-year window to measure demand uncertainty for each industry by using the following model:
Control Variables
In our models, we control for several firm-level and TMT-level variables that influence our dependent variables. Specifically, we control for differences in firm size, ROA, leverage, advertising intensity, research and development (R&D) intensity, and organizational slack at the firm level (Han, Mittal, and Zhang 2017; Nath and Mahajan 2011; Sorescu and Spanjol 2008). At the TMT level, we control for differences in overall TMT size, marketing TMT size (i.e., number of marketing executives in the TMT), the proportion of women in the TMT, CFO gender, the average age of TMT executives, the average tenure of TMT executives, the average ratio of the TMT's equity to total compensation, and the CMO’s firm-specific tenure and industry experience (Nath and Bharadwaj 2020; Nath and Mahajan 2011).
Model-Free Evidence
We present the model-free results in Table 2. We find that female CMOs engage in earnings management in fewer firm-year combinations (mean difference = −.061, p < .01). Similarly, we find that female CMOs launch fewer new products (mean difference = −.330, p < .05). However, we find that the advertising intensity is higher among firms that have a female CMO (mean difference = .002, p < .01).
Model-Free Evidence.
*p < .10. **p < .05. ***p < .01.
Notes: Results presented are using the two-sample t-test.
We also examine the mean differences between female and male CMOs for a few key variables. We observe that firms with a female CMO are relatively larger in size (mean difference = .333, p < .01) and have a higher market value (mean difference [million USD] = 6,230.85, p < .01). There are no statistically significant differences in ROA or Tobin's q in firms with a female CMO than in firms with a male CMO. In addition, female CMOs have slightly less firm (mean difference = −.309, p < .01) and industry (mean difference = −.285, p < .01) experience as compared with their male counterparts.
Addressing Endogeneity
The model-free results might suffer from endogeneity if firms systematically choose their CMO’s gender. For example, high-performing firms may aim for better gender representation in their TMT and prefer a female CMO. Moreover, high-performing firms are also likely to introduce more new products. Therefore, omitted variables may confound the effect of CMO gender on marketing variables.
We rely on the instrumental variable approach to address the possibility of endogeneity. In line with recent marketing research (Malshe, Colicev, and Mittal 2020; Shi, Grewal, and Sridhar 2021), we employ the proportion of female CMOs in peer-of-peer (or second-degree peer) firms as the instrumental variable. To construct the instrumental variable, we obtain our sample firms’ primary and secondary SIC codes from Compustat segments data. Next, we make pair-wise comparisons of the firms’ SIC codes to identify their first and second-degree peers (Shi, Grewal, and Sridhar 2021). Firms that operate in at least one common industry as the focal firm constitute the focal firm's immediate or first-degree peers. Since firms operate in multiple industries, the peers of the focal firm may operate in industries where the focal firm does not operate. Peer-of-peer or second-degree peers refer to firms that are not the focal firm's peers but peers of the focal firm's peers.
The proportion of female CMOs in second-degree peer firms must satisfy relevance and exclusion restriction criteria to be a valid instrument. The instrument will satisfy the relevance criterion if it is correlated with the endogenous variable, CMO_Gender. We anticipate that the proportion of female CMOs in second-degree peer firms would impact the likelihood of having a female CMO in the focal firm since the focal firm's immediate peers directly interact with second-degree peers. These within-group interactions would lead the second-degree peers to influence the decisions of the immediate peers, which would affect the focal firm's decisions (Shi, Grewal, and Sridhar 2021). We run a probit regression of CMO_Gender on the instrumental and control variables (Table 3) to empirically test the relevance criterion. The instrument is a significant predictor of CMO_Gender, establishing instrument relevance (β = .591, p < .05).
First-Stage Regression (Probit Regression).
*p < .10. **p < .05. ***p < .01.
Notes: Standard errors are in parentheses.
Exclusion restriction means that the instrument does not directly influence the focal firm's marketing decisions. This assumption cannot be empirically verified but needs to be conceptually justified. Extant literature elucidates the advantage of instruments based on second-degree peers versus immediate industry peers (Shi, Grewal, and Sridhar 2021). Employing instruments from first-degree peers may not satisfy the exclusion restriction criterion, as peers can influence the focal firm's marketing decisions. For instance, firms often focus on industry benchmarks while determining advertising intensity and NPIs. Moreover, correlated unobservables, such as economic shocks, can affect the peers and the focal firm. However, second-degree peers that do not operate in the focal firm's industry are unlikely to influence the firm's advertising spending and NPI decisions. In other words, since second-degree peers do not belong to the same industry as the focal firm, the proportion of female CMOs among second-degree peers can influence the focal firm's decisions only through the appointment of a female CMO in the focal firm.
Empirical Models
We specify the following empirical model to estimate the main effect of CMO gender on marketing decisions:
In the next empirical model, we incorporate moderating variables as follows:
Because we are examining marketing decisions that comprise a continuous (advertising intensity), a binary (REM), and a count (NPIs) variable, we need different estimation procedures to account for endogeneity. First, we estimate the main and moderating effects for advertising intensity (continuous outcome) using the traditional control function approach, where we add generalized residuals from the first-stage probit regression as a control in the second-stage linear regression. Second, we model REM's main and moderating effects (binary outcome) using bivariate probit regression, which jointly estimates the first- and second-stage equations (Wooldridge 2010, section 15.8.5). Third, as Wooldridge recommends, 2 we use IVPOIS regression (generalized method of moments estimator of Poisson regression that corrects for endogeneity) to model the main effects of NPIs (count outcome). It generates consistent estimators for both continuous and discrete endogenous variables, estimates any exponential model, and offers the most robust procedure for handling endogeneity with count outcomes. However, IVPOIS regression does not estimate count models with interaction terms. Therefore, to estimate the NPI model with moderators, we resort to the control function approach, where we insert generalized residuals from the first-stage regression as a control in the second-stage Poisson regression.
Results
In Table 4, we assess how the CMO's gender affects marketing decisions within the firms. We find no statistically significant differences in advertising intensity between female and male CMOs. Next, we study how female CMOs affect earnings management practices within the firms. We find strong evidence suggesting that firms with female CMOs have lower advertising-related earnings management (βCMO_Gender = −1.275, p < .01). Together, these two results imply that while advertising spending does not differ between male and female CMOs, the latter are less likely to engage in myopic practices. Since earnings management helps immediate firm performance but harms future financial performance (Mizik 2010), female CMOs may be more cautious about engaging in such practices.
Impact of CMO Gender on Marketing Decisions.
*p < .10. **p < .05. ***p < .01.
The ad intensity model is estimated using the control function approach with bootstrap SEs. R2 is reported for the ad intensity model.
The REMAD model is estimated using bivariate probit regression. We report Wald chi-square for the REMAD model.
The NPI model is estimated using IVPOIS regression.
Notes: Standard errors are in parentheses.
Moreover, we find that female CMOs are associated with fewer NPIs (βCMO_Gender = −4.100, p < .01). Because introducing new products is risky, this result suggests that, on average, women CMOs exercise greater caution when introducing new products. Indeed, a more rationalized or cautious approach to commercializing products may be a positive development, as prior research suggests that the product failure rate is 30%–50% (Castellion and Markham 2013). These product failures are expensive because firms’ development and marketing costs incurred during the launch cannot be recovered (Hitsch 2006). Greater caution in introducing new products to the market may economize on such commercialization costs.
Next, we discuss the moderating impact of CEO gender, relative firm performance, and demand uncertainty on marketing decisions (Table 5). In line with H2, we find that female CMOs undertake more risky marketing decisions under a female CEO. Specifically, we find that female CMOs are associated with more NPIs under a female CEO (β2 = .731, p < .01; see Figure 2). While we find a marginally significant moderating effect of CEO gender on advertising intensity (β2 = .006, p < .10; see Figure 3), the moderating effect on REMAD is not statistically significant.
NPI: CMO Gender × CEO Gender.
Ad Intensity: CMO Gender × CEO Gender.
Moderating Impact of CEO Gender, Relative Firm Performance, and Demand Uncertainty.
*p < .10. **p < .05. ***p < .01.
The ad intensity and NPI models with moderators are estimated using the control function approach with bootstrap SE. R2 is reported for the ad intensity model and pseudo-R2 for the NPI model.
The REMAD model with moderators is estimated using bivariate probit regression. Wald chi-square is reported for the REMAD model. The relative firm performance moderator cannot be tested for REMAD as the dependent variable is computed using ROA.
Notes: Standard errors are in parentheses.
Concerning H3, we find partial evidence suggesting that female CMOs undertake more risky marketing decisions when relative firm performance is higher. We find that female CMOs introduce more new products when relative firm performance is higher (β3 = 1.068, p < .05; see Figure 4). However, the moderating impact of relative firm performance on advertising intensity and earnings management is statistically insignificant. 3
NPI: CMO Gender × Relative Performance.
Finally, in support of H4, we find that female CMOs undertake less risky marketing decisions when demand uncertainty is high. Specifically, we find that female CMOs are associated with lower REMAD (β4 = −1.449, p < .05; see Figure 5). We also find a marginally significant negative effect on NPIs (β4 = −5.340, p < .10; see Figure 6). However, the moderating effect of demand uncertainty is statistically insignificant for advertising intensity.
REMAD: CMO Gender × Demand Uncertainty.
NPI: CMO Gender × Demand Uncertainty.
Overall, we find that, on average, female CMOs are likely to launch fewer products, and this effect is moderated by CEO gender, relative firm performance, and demand uncertainty. 4 We also find that female CMOs are less likely to engage in REM, and this effect is moderated by demand uncertainty. There is no effect of CMO gender on advertising intensity. However, a female CMO is likely to have higher advertising intensity in the presence of a female CEO.
As a robustness check for the result with advertising intensity as the dependent variable, we examine the impact of female CMOs on strategic emphasis, that is the relative emphasis a firm places on value appropriation as compared with value creation activities (Han, Mittal, and Zhang 2017). We measure strategic emphasis as (advertising expenses − R&D expenses)/total assets. We find no statistically significant differences in strategic emphasis between female and male CMOs. The results suggest that female CMOs may not differ in their budgetary allocation relating to advertising and R&D expenditures compared with their male counterparts.
Studies 2 and 3
Overview
We conduct two additional studies to examine the underlying mechanisms that can explain gender differences in marketing decisions. Study 2 has two objectives: (1) to test the link between gender and marketing decisions in an experimental setup (i.e., with fewer concerns for endogeneity) and (2) to examine the mediating role of risk appetite in the relationship between gender and marketing decisions. We rely on MarkStrat, a well-established marketing simulation, to achieve these objectives. A total of 185 teams participated in the simulation. We examine whether a team lead's risk appetite mediates the relationship between their gender and marketing outcomes such as NPIs. In Study 3, we use a survey to examine whether confidence, failure avoidance orientation, and fear of scrutiny mediate the relationship between gender and risk-taking behavior.
Study 2
MarkStrat overview
MarkStrat simulation imitates the competitive business environment and has been used previously to examine TMT decisions. For example, Kilduff, Angelmar, and Mehra (2000) use the MarkStrat simulation to study how demographic diversity affects cognitive diversity within teams and its impact on firm performance. There is widespread consensus that MarkStrat simulation captures a real-world competitive business environment (Kinnear and Klammer 1987). In this realistic simulation, participants must formulate and implement a long-term marketing strategy and make multiple short-term tactical decisions in a competitive environment. Thus, this simulation is used widely in both MBA programs and executive training programs.
Each team participating in MarkStrat represents the firm's TMT and is responsible for formulating and implementing a cohesive marketing strategy for its firm. Specifically, each team must decide which consumer segments to target and how to position its products; how to design, launch, modify, or withdraw products; how to engage in production and inventory planning; how to make marketing-mix decisions (pricing, advertising, etc.) for each brand; how to allocate resources to its commercial team and select sales channels; and how to order market research studies to improve marketing intelligence (Larréché and Gatignon 2020).
Sample
Nine hundred thirty MBA students participated in the simulation for six rounds (each round corresponding to one financial year). Participants were grouped into teams of five to six. Each team was assigned a firm within an industry, and each industry comprised five firms. That is, each firm competed with four other firms in its industry. Our sample comprises 185 teams from 37 industries.
The participants were preassigned into teams by the business school for the core courses. Thus, the assignment of teams is exogenous. Further, the team allocation ensures that teams have good diversity (age, education, experience, and gender). For example, most teams have a uniform gender composition, with roughly 40% women. This enables us to delineate the impact of female versus male CMOs while controlling for any additional effects of gender and other forms of diversity on marketing decisions.
Procedure
Prior to the first decision round in MarkStrat, each team designated a CMO 5 responsible for managing overall profitability, coordinating, deploying a unified marketing strategy, and resolving intrateam conflicts. Essentially, the CMO acted as the team lead. We measure the gender and risk appetite of the designated CMOs and relate these to NPIs. We measure risk appetite by asking participants to rate themselves on a scale of 1–5 (low–high) on their risk appetite. Out of 185 designated CMOs, only 169 CMOs responded to the survey. Therefore, our final sample comprises 169 teams, out of which 105 teams designated a male CMO (62%) and 64 chose a female CMO (38%). We create a dummy variable called CMO_Gender, which takes the value of 0 for male CMO and 1 for female CMO.
Our dependent variable is the total count of new brands that a firm introduces over six simulation rounds (BRDTOT). Each firm starts with two brands in the existing market in the first period. New brands can be launched in an existing or new product market. The MarkStrat manual (Larréché and Gatignon 2020) describes products in the existing market (BRDEXIS) as those that have existed for several years and have an established growing market and several competing brands at different price points. In contrast, products launched in a new market (BRDNEW) are described as new types of product that are yet to emerge and may require significant investments to develop and launch. Therefore, brands launched in the existing market can be considered incremental innovations, whereas those launched in the new market can be considered radical innovations. BRDEXIS is the cumulative number of brands launched in the existing markets, and BRDNEW is the number of brands launched in the new markets over six periods.
Mediation analysis
We examine whether (1) the CMO’s gender has an impact on NPIs, and (2) risk appetite (risk aversion) mediates the effect of the CMO’s gender on new product (brand) introductions. We find that the total effect of CMO_Gender on BRDTOT is negative and significant (β = −.270, p < .10). This implies that female CMOs launch fewer products than male CMOs. This finding is in line with Study 1, using secondary data, where we find a similar effect. Like the BRDTOT results, the total effect of CMO_Gender on BRDNEW is negative and significant (β = −.257, p < .05). However, the total effect of CMO_Gender on BRDEXIS is negative but statistically insignificant (β = −.0124, p = n.s.).
We study the proposed mediation using the bootstrapping approach suggested by Preacher and Hayes (2004). The mediation results are summarized in Figure 7. The indirect effect of CMO_Gender on BRDTOT through risk appetite is statistically insignificant. This could be because of the contradictory effect that risk appetite has on BRDEXIS (brands launched in the existing market) and BRDNEW (brands launched in the new market). We find that the indirect effect of CMO_Gender on BRDNEW through risk appetite is negative (indirect effect = −.067) and significant (the 95% CI of the indirect effect [−.1510, −.0104] does not contain zero). This implies that female CMOs have a lower risk appetite, which leads to the introduction of fewer radical innovations. In contrast, we find a positive (indirect effect = +.078) and significant (95% CI: [.0135, .1630]) indirect effect of CMO_Gender on BRDEXIS through risk appetite (i.e., due to their greater risk aversion, female CMOs introduce more incremental innovations). We believe these results provide evidence for the underlying mechanism of risk aversion and demonstrate the differential impact of executive gender on radical versus incremental product introductions.
Mediation Analysis: Effect of CMO Gender on Innovation Through Risk-Appetite.
Robustness check
We also examine the effect of CMO gender on radical versus incremental innovations using secondary data. For this objective, we rely on the Product Launch Analytics database, which collates information on truly innovative products in the consumer packaged goods industry (see Web Appendix G). Product Launch Analytics assigns innovation ratings to each of its products, differentiating a truly innovative product that is world-first or region-first in a category from other innovative standout products. Using prior literature, we categorize these products as radical and incremental innovations (Sorescu and Spanjol 2008). Specifically, we define radical innovations as the count of all products in Product Launch Analytics that have been assigned an innovation rating (such as formulation, positioning, packaging, new markets, technology, and merchandising) aggregated annually. Similarly, we define incremental innovation as the count of products labeled as other innovations in their innovation ratings aggregated annually. The sum of radical and incremental innovations is defined as a firm's total innovation.
After merging Product Launch Analytics data with information on CMO gender and the control variables, we retain data comprising 68 firms, with 302 observations. Using the traditional control function approach, we examine the impact of female CMOs on the ratio of radical to total innovation. We find that female CMOs are associated with a lower ratio of radical to total innovation as compared with their male counterparts (β = −.680, p < .05) (see Table W8 in Web Appendix G). These results provide external validity to the findings of the MarkStrat experiment.
Study 3
Overview
In Study 3, we employed a large sample of 930 participants (the same set of participants from Study 2, but the survey is now for all the participants and not just the team leads) to examine the underlying factors which drive gender differences. Specifically, we study whether overconfidence, failure avoidance orientation, and the level of scrutiny mediate the relationship between gender and risk-taking behavior.
We conducted the survey in a core marketing course in a reputed business school. The participants’ age ranged from 22 years to 41 years, with an average age of 26.79 years. They came from diverse backgrounds, such as consulting, finance, technology, marketing, health care, defense, and others. The participants had an average work experience of 4.6 years, with a minimum of 1.7 years and a maximum of 20+ years. Therefore, all participants had prior exposure to leadership roles and team dynamics in a professional setting.
Out of 930 participants, 730 graduate students responded to the survey. Among the respondents, 62% were male and 38% female. We asked the students to imagine themselves as CMOs while answering the survey questions.
Measures
Risk appetite (aversion)
We measure an individual's risk appetite using two methods. First, we asked the participants to rate their risk appetite on a scale of 1 (“low”) to 5 (“high”). Second, we relied on an objective measure of risk aversion used in the literature (e.g., Lim, Ahearne, and Ham 2009; Opper, Nee, and Holm 2017). We asked participants to choose between two lotteries (A and B) with different relative risks to elicit their risk preferences. We modified the lotteries to resemble potential payoffs from two NPIs. Product A has a high payoff of Rs. 300 crores (3 billion Indian rupees) if it is successful and a low payoff of Rs. 240 crores if unsuccessful. In contrast, Product B has a potentially high payoff of Rs. 580 crores and a potentially low payoff of Rs. 15 crores. We then asked participants to choose between Product A and Product B, successively, for five rounds. In each successive round, the probability of a high payoff (initially 10%) increased by 20%, and the probability of a low payoff (initially 90%) decreased by 20%. We provide details in Web Appendix H.
To measure risk aversion, we code the decision round where the participant switches from Product A to Product B. In other words, we count the number of rounds it takes for a participant to switch from a relatively safer option (Product A) to a relatively riskier option (Product B) (Opper, Nee, and Holm 2017). The game terminates once a participant chooses Product B to eliminate irrational or random choices. Therefore, if a participant chooses Product B in the third round, we code their risk aversion as 3. Similarly, if a participant chooses Product B in the fifth round, we code their risk aversion as 5. A participant can play the game for a maximum of five rounds. If a participant still chooses Product A in the fifth round, we code the decision as 6. Thus, the risk aversion scale ranges from 1 to 6.
Overconfidence
We measure overconfidence between male and female participants based on two questions. First, we asked participants to rate their confidence on a scale of 1–5. Second, we asked participants if they think they are or can be more effective team leaders than the other team members (yes/no).
Fear of failure
We asked participants to rate their fear of failure on a scale of 1–5.
Fear of scrutiny
We measure fear of scrutiny using five items adapted from Conroy, Willow, and Metzler (2002). We asked participants to imagine themselves as a company's CMO and then rate the fear of scrutiny they are likely to face. We use the following items (a five-point scale ranging from “strongly disagree” to “strongly agree”) to measure fear of scrutiny: “If I fail …” “I will be heavily criticized by my colleagues,” “I will lose the trust of my colleagues,” “My doubters will feel that they were right about me,” “My value will decrease in the organization,” and “My future in the organization will be uncertain.” The fear of scrutiny measure is calculated as the average of the five-item score.
Results
The results are summarized in Table 6. We find consistent evidence using (1) a self-rating scale and (2) an objective measure using NPI decisions, which suggest that women are more risk averse than men. Specifically, when we asked the participants to rate themselves on their risk appetite, the female participants rated themselves lower than the male participants (β = −.255, p < .01). Second, we find that the female participants have a higher risk aversion score than their male counterparts (β = .231, p < .05). This implies that women require a higher probability of a favorable outcome to launch a risky product as compared with men.
Independent Samples T-Test Results.
*p < .10 **p < .05. ***p < .01.
Number of observations for the risk aversion scale: male = 363, female = 223, total = 586.
Notes: Number of observations by gender: male = 451, female = 279, total = 730.
Next, we find mixed evidence suggesting that men may have higher confidence than women. Although we find no statistically significant differences in mean self-rating scores between male and female participants (β = −.082, p = n.s.), we find that a smaller percentage of women think they are or can be more effective team leaders than their other team members (β = −.122, p < .01). Further, when we asked the participants to rate themselves on their fear of failure, women displayed a higher fear-of-failure score than men (β = .222, p < .01). In addition, women scored higher on the fear-of-scrutiny scale compared with men (β = .178, p < .01). In summary, we find that women are more cautious or risk averse than men. We also have mixed evidence suggesting that men are relatively more overconfident than women. Further, women have greater fear of failure and greater fear of scrutiny compared with men.
Mediation analysis
We examine whether confidence, fear of failure, and fear of scrutiny mediate the effect of an individual's gender on their risk appetite (see Figure 8). We study this mediation using Preacher and Hayes’s (2004) bootstrapping approach. We find no evidence suggesting that an individual's confidence mediates the relationship between their gender and risk appetite. The result is expected, since we find no differential effect of gender on confidence (self-rating) scores. However, we find that fear of failure and fear of scrutiny partially mediate the relationship between an individual's gender and risk appetite. The indirect effect of gender on a person's risk appetite through fear of failure is negative (indirect effect = −.0319) and significant (95% CI: [−.0597, −.0084]). Similarly, the indirect effect of fear of scrutiny is negative (indirect effect = −.0160) and significant (95% CI: [−.0373, −.0014]). The mediation results imply that women have higher fear of failure and greater fear of scrutiny, both of which lower their risk appetite.
Mediation Analysis-Mechanisms Underlying Risk Appetite.
Discussion
Synthesis of Findings
Using a multimethod approach (secondary data, experiment, and survey), we examine when and why female CMOs’ marketing decisions differ from those of their male counterparts. We find strong evidence that firms with female CMOs, on average, launch fewer new products (Studies 1 and 2). Further, we find that the indirect effect of CMO gender (through risk taking) on NPIs is negative for radical innovations but positive for incremental innovations (Study 2). We also find that female CMOs are less likely to engage in REM. However, we do not find that the CMO gender influences advertising intensity. This could be because the advertising budget allocation may depend on other factors (e.g., past spending, competitor spending) and not on the CMO's gender.
Regarding moderators, we find evidence for demand uncertainty on NPIs and REMAD and CEO gender on NPIs and ad intensity. However, we find support for the moderating role of relative firm performance only for NPIs. In other words, we do not find strong support for the argument (H3) that female CMOs will display more risk-taking behavior when relative firm performance is higher.
In addition, we examine underlying mechanisms of risk-taking behavior in Study 3 and find that the gender–risk appetite relationship is not mediated by confidence. It is important to note that this result departs from the prior studies on executive gender, which rely on confidence as the theoretical mechanism to explain why female executives are less risk taking. Instead, we find that fear of failure and fear of scrutiny in the workplace mediate the relationship between gender and risk-taking behavior. We next discuss the implications of these findings.
Theoretical Contributions
We make five theoretical contributions. First, our work contributes to the growing body of research in understanding the role of CMOs. This research stream has gained momentum with Nath and Mahajan's (2008, 2011) articles that examine the antecedents and consequences of having a CMO. Germann, Ebbes, and Grewal (2015) reexamine the relationship between CMO and firm performance and find strong evidence that CMO presence increases firm performance. However, as You et al. (2020) note, empirical findings are rich on CEO characteristics but not on characteristics of other TMT members, including CMOs. For example, prior research on CEOs examines the effect of CEO background, demographics, and experience on their decisions and finds that the CEOs’ characteristics influence firm-level decisions. By investigating whether the CMO's gender matters (a salient difference from other functional areas that do not have an equitable gender balance), we expand the scope of CMO research. We hope our research further stimulates marketing scholars’ interest in investigating other characteristics of the CMO (e.g., inside hire, personality, experience).
Second, we identify new mechanisms for lower risk-taking behavior among female TMT members. Previous studies on CEOs find that male CEOs are more risk taking. Those studies offer male overconfidence as the reason for higher risk taking among male CEOs. Our study examines the issue from female executives’ perspectives. While we find empirical evidence for higher fear of failure and greater fear of scrutiny among female CMOs, we do not find unequivocal evidence to suggest that female executives are less confident. In other words, the confidence levels of people working in executive positions may not differ (Tinsley and Ely 2018). However, they may still be influenced by gender biases, such as scrutiny in the workplace. It is an important insight as, unlike overconfidence, which could be a personality trait and not necessarily influenced by organizational policies, scrutiny in the workplace can be proactively managed.
Third, we contribute to the literature on radical and incremental innovations. Specifically, while previous studies find that radical innovations are associated with increases in profits (e.g., Sorescu and Spanjol 2008), there is limited work on understanding the antecedents to radical versus incremental innovations in marketing. For example, Sorescu, Chandy, and Prabhu (2003) find that dominant firms introduce more radical innovations than nondominant firms. Similarly, Moorman et al. (2012) find that public firms are more likely to ratchet radical innovations. However, how TMT characteristics influence radical innovations has not been examined yet. Based on the results from Study 2 and the Product Launch Analytics data set (details in Web Appendix G), we find that the CMO's gender can influence decisions to launch radical and incremental products.
Fourth, we examine the interactions between CMOs and TMT members. Nath and Bharadwaj (2020) study how other TMT members (e.g., chief sales officers, chief technology officers) influence the effectiveness of the CMO. We examine the interaction between the CMO and the CEO to expand this stream of research. Specifically, we contribute by examining how the CEO and CMO's gender similarity influences the CMO's risk-taking behavior. We hope our research stimulates future studies examining how the CEO and CFO characteristics can influence the CMO's decisions. For example, it would be interesting to investigate whether a CMO hired by a CEO is willing to engage in more earnings management practices.
Fifth, we propose an alternative method to identify marketing presence in the TMT. The operationalization of who constitutes the TMT is not trivial and has varied in previous research (Krause, Roh, and Whitler 2022; Nath and Bharadwaj 2020). A few studies use Execucomp, which lists the top five to ten executives in terms of salary, to identify CMO presence. Specifically, 8.1% of the published TMT studies (across the disciplines) use Execucomp to define the TMT (Krause, Roh, and Whitler 2022). However, using Execucomp induces a bias due to variations in reporting policy across the firms as well as the years within a firm. Nath and Bharadwaj (2020, p. 691) note that “latter operationalization, which results in selecting a relatively smaller number of CXOs is likely to bias estimates of CXO presence or other CXO characteristics.” To overcome this, most marketing research on CMOs relies on 10-K reports (Germann, Ebbes, and Grewal 2015; Nath and Mahajan 2017) and treats the list of executives given in those reports as the TMT. Among all TMT studies (not just within marketing), only 2.5% rely on 10-K reports (Krause, Roh, and Whitler 2022). We document that using 10-K reports to identify a CMO presence can pose similar challenges as with Execucomp. Moreover, given the manual labor involved in collating data through 10-K reports, previous studies have limited their sample size commonly to 120–150 firms.
We propose an alternative to the above two methods of identifying the CMO presence and characteristics by using BoardEx data and the hierarchical levels within the titles (Chief, President, Executive VP, Senior VP, and VP) as identified in previous marketing research (Feng, Morgan, and Rego 2015). While Feng, Morgan, and Rego (2015) use the hierarchy of titles to measure marketing department power, we rely on the same procedure to identify the marketing head. Since BoardEx covers more firms and titles, our sample includes more than 1,250 firms across 11 years. Further, if a researcher is interested in understanding the relationship between the characteristics of the marketing head and the marketing outcomes (not necessarily firm-level outcomes), the relevant person may or may not be listed in the annual reports/Execucomp but may still be taking all the important decisions pertaining to marketing (e.g., advertising content). Therefore, as future research examines more nuances surrounding the head of marketing, our procedure provides a more reliable and scalable approach.
Managerial Implications
We have three important managerial implications. First, our results demonstrate that female CMOs exercise caution in their marketing decisions, which can benefit firms. For instance, a cautious approach to NPIs may save product development costs and insulate firms from potential product failures. In addition, female CMOs are more attuned to their environment and suitably adapt their decisions depending on environmental cues, such as relative performance and demand uncertainty. For example, they are more cautious in times of greater demand uncertainty and more risk taking when their firms perform well. Therefore, female CMOs suitably modify their risk-taking behavior depending on the situational context. Thus, we recommend that boards appoint more female CMOs not just to increase diversity but also to benefit shareholders. In fact, when we regress firm value (Tobin's q and market cap) on CMO_Gender along with other control variables and the generalized residuals, we find that the presence of female CMOs is associated with higher Tobin's q (β = 2.22, p < .01) and market cap (β = 3.91, p < .01).
Second, we find evidence that suggests a spillover of female representation in senior leadership positions on subordinates. We find that female CMOs exhibit more risk-taking behavior in the presence of female CEOs, possibly because of higher confidence, less fear of failure, and less fear of scrutiny. The presence of a female CEO challenges stereotypical biases and weakens the perception that only men occupy and thrive in senior leadership roles. Further, extant research suggests that female CEOs may be more adept at inferring the productivity signals of female subordinates (Flabbi et al. 2019). From a managerial perspective, our findings support that hiring a female executive has spillovers for women in subordinate roles.
Third, there is no dearth of articles or advice for women to be more confident, possibly because the “standard” to which they are being compared is overconfident men (Gallop and Chamorro-Premuzic 2021). However, we did not find a difference in the confidence levels of the MBA students who participated in the survey (Study 3). Hence, when these people reach the upper echelons of their organizations, their confidence levels may not differ, though they may still face differential levels of workplace scrutiny. Thus, organizations should proactively take measures to reduce unwanted scrutiny, which may hamper decision making.
Limitations and Future Directions
First, we examine the effect of one moderating variable within each category of structural, organizational, and environmental factors. However, there could be other factors within each category that can moderate the CMO’s decisions. For example, future research could examine the moderating impact of marketing department power (structural), branding strategy (organizational), and industry competition (environmental). Further, while we examine the downstream impact of female CMOs on marketing decisions, we find limited research that examines the antecedents determining female CMO appointments. Second, since it is challenging to collect survey data from senior executives, we rely on surveys conducted on MBA students in Study 3 to delineate the mechanisms that explain risk taking. Even though we ask the students to imagine themselves as CMOs, we qualify that the findings from Study 3 only approximate senior executives’ decisions. Third, the secondary data we employ in Study 1 is specific to the United States, whereas the experiment and surveys in Studies 2 and 3 are from a developing nation. While the results from Study 1 (secondary data) and Study 2 (experiment) are congruent, it is imperative to consider that there may be cultural differences in gender equality across nations. Future research could examine how the role of the CMO varies across cultures and uncover its implications for female CMOs.
Supplemental Material
sj-pdf-1-mrj-10.1177_00222437231156902 - Supplemental material for Female Chief Marketing Officers: When and Why Do Their Marketing Decisions Differ from Their Male Counterparts’?
Supplemental material, sj-pdf-1-mrj-10.1177_00222437231156902 for Female Chief Marketing Officers: When and Why Do Their Marketing Decisions Differ from Their Male Counterparts’? by Rajita Varma, Raghu Bommaraju and Siddharth S. Singh in Journal of Marketing Research
Footnotes
Associate Editor
Alok Saboo
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
