Abstract
A growing literature examines whether women’s integration into management jobs erodes gender stereotypes and gender inequality. However, this literature neglects the other side of the status coin—women’s continued predominance in low-level support jobs. I theorize that what people see when they “look down” the occupational structural is more critical to the creation of status beliefs than what they see when they “look up,” and test this theory using matched employer-employee data from Japan. I find that, adjusting for job type and human capital, the gender pay gap is nearly three times greater in companies where subordinate jobs are female dominated. This theory provides new ways to understand the “stalled gender revolution” in the United States, Japan, and beyond.
Many scholars and business consultants agree that higher managerial representation for women is key to disrupting patterns of male supremacy in the workplace (see, for example, Cohen and Huffman 2007, 700; Huffman, Cohen, and Pearlman 2010, 273; Nemoto 2016; Kramer and Harris 2019, chap. 11; Shams and Tomaskovic-Devey 2019). Because managerial women erode gender essentialist beliefs, the argument goes, women’s representation in management also benefits the vastly larger ranks of nonmanagerial women workers by reducing bias and discrimination against all women (Ely 1995; Huffman, Cohen, and Pearlman 2010; Kramer and Harris 2019).
This perspective draws on status construction theory (SCT), which posits that where resources and power are unevenly distributed among groups, members of the better-resourced group will tend to be more highly valued and esteemed. This esteem for members of the better-resourced group not only leads people to bestow its members with higher rewards, but also gives their higher rewards the halo of merit and justice, further reinforcing the unequal conditions that first give rise to the belief (Ridgeway 1997, 2019). Inspired by this framework, a rapidly growing body of organizational scholarship investigates the effects of women’s gradual incursion into management on outcomes for nonmanagerial women in the United States and other postindustrial countries (see, for example, Ely 1995; Hultin and Szulkin 2003; Cohen and Huffman 2007; Huffman, Cohen, and Pearlman 2010; Kurtulus and Tomaskovic-Devey 2012; Stainback and Kwon 2012; Stainback, Kleiner, and Skaggs 2016; Abendroth et al. 2017; Halldén, Säve-Söderbergh, and Rosén 2018; Stojmenovska 2019).
This scholarship, however, has overlooked the other side of the status coin—women’s enduring overrepresentation in low-level service and administrative positions. Although women’s representation in historically male-typed professional and management positions has grown dramatically over the past five decades, men’s representation in historically female-typed jobs, such as those of secretaries or office assistants, has not increased accordingly (England 2010; Brynin and Perales 2016; on men’s hostility to feminized service jobs, see Koenig 2022). SCT implies that this persistent female overrepresentation in jobs with little power and resources may have marked effects on gender status beliefs and consequently on the ways in which all women are viewed and treated in the workplace. Yet the relationship between the gender composition of these categorically subordinate roles and other metrics of gender inequality receives little attention.
In this article, I draw on the cross-national literature on gender inequality and workplace demography to theorize why the composition of low-status jobs may be of even greater importance to the creation of status beliefs than the composition of managerial and professional jobs. I test the implications of this theory using data from Japan and show that gender gaps in pay and in the subjective experience of feeling valued by supervisors are both far larger where subordinate jobs are female dominated. I argue that the exclusive focus on women’s representation in management in previous scholarship can itself be seen as a form of pro-male bias, whereby the gender composition of stereotypically male-typed management jobs is implicitly assumed to matter more for workplace dynamics than the composition of female-typed subordinate jobs.
This new theory can illuminate the causes of what Paula England (2010) terms “the stalled gender revolution.” Although the direct relationship between women’s overrepresentation in low-status support roles and the stagnant gender wage gap is well documented (England 2010; England, Levine, and Mishel 2020), the theory developed here highlights how women’s predominance in support jobs may have important indirect consequences. In addition to its direct effect on wages, women’s overrepresentation in low-status jobs may entrench views of women as “mere” assistants, justifying devaluative treatment of women, regardless of the job they hold. This theory thus provides a new lens through which to understand why gendered beliefs in the workplace have remained intractable (Vial, Napier, and Brescoll 2016) and why gender inequality in pay has barely changed since the early 2000s (England, Levine, and Mishel 2020), despite women’s consistent gains in managerial representation over the past half century in the United States and beyond (World Bank 2021).
WHY THE GENDER COMPOSITION OF SUBORDINATE JOBS MATTERS
Contrary to media hyperbole (for example, Feintzig 2020), office digitization has by no means eliminated secretarial work. Office and administrative support is the single largest occupational category in the United States, employing over 18.5 million workers in 2020, more than three times the number classified as managerial (Bureau of Labor Statistics 2021a). Despite the near disappearance of “secretary” as a job title, an army of “assistants,” “coordinators,” and even “managers” (Seeley 2018), “directors” (ZipRecruiter 2021), or “partners” (Facebook Careers 2021) continue to greet customers, answer phones and emails, make photocopies, manage calendars, order supplies, and maintain filing systems and databases (Truss et al. 2013) in the name of facilitating the “more important” work of others (Karlsson 2011; Seeley 2018).
The face of this administrative army is, and has long been, female. Women made up 77 percent of administrative support workers in the United States in 1975 (Wootten 1997) and 72 percent of administrative support workers in 2020 (Bureau of Labor Statistics 2021b). The more closely an administrative job hews to the secretarial model (Truss et al. 2013), the more female dominated it tends to be. In 2020, 93 percent of U.S. administrative assistants were women (Bureau of Labor Statistics 2021b).
Rosabeth Kanter’s groundbreaking study of gender in the white-collar workplace paid careful attention to the role of the secretary and noted how women secretaries’ categorically subordinate positions bolstered the power and authority of male managers (1993, 18, 84). Building on this study, status construction theory has focused on how power and resource differentials reinforce men’s workplace supremacy (Ridgeway 2019), acknowledging the role that female support employees play in workplace status dynamics (Ridgeway and Smith-Lovin 2002, 196–97).
Yet few studies investigate the circumstances of low-level administrative personnel (for exceptions, see Karlsson 2011; Truss et al. 2013; Costello 2015; Seeley 2018). Moreover, empirical work on organizational demography as a determinant of gender inequality focuses exclusively on the effects of women’s entry into and presence in male-typed managerial and professional jobs (Ely 1995; Cohen and Huffman 2007; Huffman, Cohen, and Pearlman 2010; Maume 2011; Kurtulus and Tomaskovic-Devey 2012; Penner, Toro-Tulla, and Huffman 2012; Srivastava and Sherman 2015; Stainback, Kleiner, and Skaggs 2016; Abendroth et al. 2017; Halldén, Säve-Söderbergh, and Rosén 2018; Stojmenovska 2019).
This cross-national literature, which includes cases from the United States (Kurtulus and Tomaskovic-Devey 2012), the United Kingdom (Stojmenovska 2019), Sweden (Halldén, Säve-Söderbergh, and Rosén 2018), Germany (Abendroth et al. 2017), South Korea (Stainback and Kwon 2012), and Japan (Nemoto 2016) has largely, if not universally, supported the predictions that women’s entry into management ameliorates other measures of gender inequality. However, where studies have found positive effects of women’s managerial presence, these effects are substantively small. For example, Fidan Kurtulus and Donald Tomaskovic-Devey (2012) find that a 1 percent increase in women’s share of management in large U.S. firms is followed by a less than 0.1 percent increase in women’s share of mid-level management over a six-year period. Matt Huffman, Philip Cohen, and Jessica Pearlman (2010) show that a 1 percent increase in women’s share of management is followed by change of 0.02 in the 0–1 segregation index of nonmanagerial jobs. Dragana Stojmenovska (2019), using British data, finds that a 1 percent increase in women’s share of management jobs is associated with a £17 (about $24) decrease in the gender wage gap in annual earnings. Why, given these small effect sizes, have scholars not turned their attention to other aspects of workplace demography, including the enduring overrepresentation of women in support roles?
In principle, the continued focus on the effects of women managerial workers is logical if women managers’ direct power and influence is the primary mechanism linking female managerial representation to better outcomes for other women (Hultin and Szulkin 2003). Women in management have far greater leeway to allocate resources in ways favorable to women than secretaries or other subordinate personnel do. In contrast to the symbolic processes described in SCT, then, the power mechanism justifies a narrower focus on women’s entry into management.
However, the power mechanism is poorly substantiated. First, qualitative studies cast doubt on claims that female managers foster subordinate women’s professional advancement (Costello 2015, 115–17). Second, even when female managers are motivated to reduce gender inequality, their capacity to act is constrained. Not only do they face more scrutiny and suspicion than male managers (Kanter 1993); they are also often concentrated on the bottom rungs of management hierarchies (Stojmenovska, Steinmetz, and Volker 2021). Finally, if managerial women’s power reduces gender inequality, we would expect to see the ameliorative effects of women leaders most strongly on gender inequality among their direct subordinates (Srivastava and Sherman 2015). Dyadic study designs, which capture the genders of both employees and their managers, are best suited to this task. However, dyadic studies have universally found null or even negative relationships between the presence of female managers, and outcomes for female subordinates (Maume 2011; Penner, Toro-Tulla, and Huffman 2012; Srivastava and Sherman 2015).
Support for the hypothesis that women’s managerial presence improves conditions for nonmanagerial women comes overwhelmingly from studies that examine gender inequality at higher levels of aggregation. These studies investigate whether gender inequality is lower across the board in firms with higher shares of women in management (Ely 1995; Cohen and Huffman 2007; Stainback, Kleiner, and Skaggs 2016; Abendroth et al. 2017), or in firms where women’s share of management has grown (Huffman, Cohen, and Pearlman 2010; Kurtulus and Tomaskovic-Devey 2012; Stojmenovska 2019). The disjuncture in the results of dyadic and aggregative studies suggests that, to the extent that women’s entry into management ranks mitigates gender inequality for nonmanagerial women, it is predominantly through symbolic channels operating at higher levels of analysis—in establishments, in organizations, industries, and society more broadly—as suggested by SCT.1 In turn, the importance of the symbolic divisions between the genders urges attention toward not just the top of the organizational pyramid but also the bottom.
Other strands of research also indicate that the composition of support jobs is likely to be a potent generator of status beliefs. Perhaps because assistants are seen as peripheral, academic studies that center their experiences are few and far between. However, the rare examples of such studies vividly highlight the persistent devaluation and marginalization of women in these roles (Truss et al. 2013; Costello 2015). An ocean of anecdotal evidence further suggests that this devaluation matters not just to the female assistants most directly affected by it, but also to women in professional and managerial roles. Female CEOs (McNally 2019), engineers (Niselow and Omarjee 2018), lawyers (Melaku 2019), medical doctors (Wible 2016), research scientists (Williams, Hall, and Philips 2014), photographers (Nittle 2017), professors (Laufenberg 2021), and politicians (Wheeler 2015) all report being mistaken for assistants. In the United States, these experiences are especially common among Black and Latina women (Williams, Hall, and Philips 2014).
The attention on female representation in elite (and male-typed) managerial and professional jobs and its effects on gender inequality is remarkably widespread, persistent, and impervious to countervailing evidence (also see Leicht 2022). Despite these examples from women’s daily lives, despite decades-old insights that female subordination shapes views of women in the workplace, despite subordinate women’s numerical predominance over managerial women across national contexts, and despite findings from across the postindustrial world that women’s entry into managerial positions has had small (or null) impacts on gender inequality for other women, the literature has perversely maintained this exclusive focus. I argue that this evidence enjoins us to take a new look at the construction of status and its effects on inequality in the workplace and beyond, with the spotlight on the bottom of the hierarchy rather than the top.
THE JAPANESE CASE
Building on the transnational literature on status, inequality, and workplace demography, I use a case study from Japan to test the implications of the theory that the composition of low-status jobs has important indirect effects on gender inequality. The selection of a Japanese case is a practical one. Contemporary American organizations sometimes attempt to sustain illusions of equality, concealing the subordinate or support nature of jobs behind inflated titles, such as “Director of First Impressions” rather than the prosaic “receptionist” (ZipRecruiter 2021). However, this sleight of hand is not typical of Japanese businesses, where hierarchy is generally accepted as a matter-of-fact feature of organizational life. Japanese organizations thus use worker classification systems that make the identification of subordinate workers straightforward.
Japanese firms’ particular delineations between subordinate jobs and other positions arise in the context of employers’ emphasis on long-term employment and internal labor markets, or ILMs (Brinton 1993; Ono 2007; Mun and Jung 2018). ILMs generate premiums for tenure and incentivize core workers to demonstrate potential and commitment by working long hours (Ono 2018). However, the dual expectation of long-term employment and long work hours make it all but impossible for women with children to fit the “ideal worker” model of Japanese firms (Brinton and Mun 2016; Ono 2018). Hence, over the past fifty years, firms have employed women predominantly, although not exclusively, in peripheral roles, either as irregular workers who lack job security and access to seniority wages (Mun 2016; JILPT 2021a), or as regular workers on a clerical track in which employees engage primarily in support work and are ineligible for promotion to management (Kanai 2013; Mun 2016).
Regular management-track jobs are male dominated in more than 80 percent of firms (Ministry of Health, Labor, and Welfare 2013, 9). Meanwhile, the clerical track is female dominated in 64 percent of firms (Ministry of Health, Labor, and Welfare 2013, 9), and 68 percent of irregular workers are women (JILPT 2021a). Thus the management, clerical, and irregular job categories institutionalize the female “secretarial ghetto” that is a ubiquitous, but less formalized, feature of office life in the United States and other postindustrial countries.
However, as in the United States, Japan’s dearth of women in management is the subject of far greater attention than women’s overrepresentation in irregular or clerical track jobs. Although the number of female irregular workers exceeds the number of managers of both genders by a factor of eleven (JILPT 2021a, 2021b),2 in the period between 2000 and 2021, mentions of “managerial work” and “women” (kanrishoku and josei) in the Asahi Shimbun, a leading Japanese newspaper, were twice as frequent as mentions of “irregular [employment]” and “women” (hiseiki and josei).
Elite discourse and policy follow this same pattern. In 2013, Prime Minister Shinzo Abe set a target of increasing women’s share of leadership positions to 30 percent by 2020 (The Economist 2014, 25). As of 2016, large companies have been required to set (voluntary) targets for women’s managerial representation and report women’s share of management jobs to a publicly accessible database. Prominent global consulting firms (Goldman Sachs 2019; McKinsey & Company 2021) champion the cause of female representation in management in Japanese firms, as do international investors (Mun and Jung 2018), often in the name of improving Japan’s economy and companies’ bottom lines. However, although there is ample room for improvement in women’s access to managerial jobs—Japan fell far short of Abe’s target, with women holding just 15 percent of managerial jobs in 2020—managerial workers are a scant 2 percent of the labor force (JILPT 2021b). As in the United States, then, faith that increasing women’s representation in management will revitalize Japan’s economy rests implicitly on the weakly supported assumptions that managerial women will create “trickle down” equality (see Leicht 2022), and that the composition of management is the most potent driver of status beliefs.
DATA
The data for this study come from a 2015 survey of employees nested within twelve large Japanese firms. The firms represent three industries—high-tech manufacturing, business services, and consumer services. All are members of the diversity subcommittee of the Japan Association of Corporate Executives, a prestigious business group.
Each firm selected at least two teams and sent the survey to every member of the team, including management track, clerical track, and irregular workers.3 The inclusion of both regular and irregular workers in nested teams and firms makes this survey unique and is crucial for exploring the question of interest because it has become increasingly common for firms to replace regular employees on the clerical track with irregular workers (Mun 2016). This data structure makes visible not only gender differences in managerial representation between firms, but also differences in gender representation at the bottom of the corporate hierarchy.
In total, 539 employees completed the survey for a response rate of 59 percent, exceeding the mean response rate of 52.7 percent in organizational surveys (Baruch and Holtom 2008). Respondents provided extensive information on their demographic background, their pay, job content, supervisory authority, and job classification. They also rated their workplaces on various subjective measures.
Descriptive data from the survey highlight a reason the composition of subordinate jobs may not only be influential in the creation of status beliefs, but also more important than the composition of management jobs. As Cecilia Ridgeway and Hazel Markus describe (2022, this issue), status beliefs emerge and are expressed in interaction. Further, deferential behaviors and attitudes are a key way in which organizational actors perform status differences (Maloney, Rogers, and Smith-Lovin 2022, this issue). In this sample, as in postindustrial economies more generally, subordinate jobs are more numerous than managerial jobs—25 percent of respondents are in subordinate roles versus 15 percent in managerial ones. Intuitively, then, we can expect that most respondents have greater opportunity to receive (or exact) deference from subordinate workers than to offer deference to managerial workers.
Respondents’ time-use reports, illustrated in figure 1, support this intuition. Including only time spent in interaction with coworkers, the median time respondents spent with managerial workers is 17 percent; 25 percent of workers spent no time at all with managers. In contrast, the median time spent with nonmanagerial workers is 80 percent. Whatever ways in which employees perform deference to managers, these performances are infrequent. Thus, contrary to the stance of the organizational demography literature, the gender composition of management is unlikely to be of highest importance in undermining or reinforcing gender status beliefs, while the gender composition of subordinate jobs is potentially of much greater impact.
Respondents’ Percentage of Interactional Work Time
Source: Author’s calculations.
Individual-level sample descriptives are presented in table 1. The analytic sample size of 528 is slightly smaller than the 539 total respondents because eleven respondents are missing data on income.
Individual-Level Descriptive Statistics
Although some Japanese firms “symbolically comply” to gender norms in other wealthy countries by hiring more women managers and leaving other aspects of gender inequality untouched (Mun and Jung 2018), both individual and firm-level statistics indicate this is not the case among the sample firms.
Including clerical track workers, nearly 80 percent of women in the sample have regular status, meaning they enjoy long-term job security and access to company welfare benefits. In contrast, at a national level, most women work in irregular positions, including part-time, term-limited, and dispatch jobs (JILPT 2021a). Further, adjusted for their shorter tenures, women have more job transfers within their firms, a key metric of firms’ commitment to developing female employees’ human capital in the context of ILMs.
Nonetheless, these metrics do not indicate that the firms have achieved a gender-equal environment. As in most Japanese firms, women are overrepresented in subordinate clerical track and irregular jobs and underrepresented in management.
Descriptive statistics on the firms in table 2 confirm the impression of the sample firms as unusually gender progressive in the broader context of large Japanese businesses. Nationwide, women’s average share of management positions in large firms, defined here as the section chief (kacho-level) and above, is around 5 percent (Ministry of Health, Labor, and Welfare 2016, 3); in 80 percent of firms, women’s share of management is less than 10 percent (Teikoku Databank 2018, 2). However, women’s share of management jobs exceeds 10 percent in seven of the twelve sample firms. Four firms have shares of women in management that are similar to the average for large firms, but three of these are manufacturing firms, where women’s typical share falls well below the cross-industry average. Thus, for large firms in their industry, even these firms are progressive.4 If women’s overrepresentation in subordinate roles is associated with greater gender inequality even in this environment where employers are unusually supportive of women’s advancement, this suggests that the relationship between these two factors may be even more marked among firms with lower commitments to women’s workplace advancement.
Firm-Level Measures of Job Distribution by Gender
ANALYTIC STRATEGY
Previous research has often used cross-sectional data to investigate whether gender wage gaps are lower in firms where women make up a larger share management (Hultin and Szulkin 2003; Cohen and Huffman 2007; Stainback and Kwon 2012; Stainback, Kleiner, and Skaggs 2016; Abendroth et al. 2017). In this study, I use this same methodology to ask whether gender wages gaps are wider where women compose a larger share of the subordinate workforce, as implied by the theory that the demography of subordinate jobs is critical for the construction of status beliefs, and for the distribution of resources in the workplace.
To do so, I model respondents’ earnings using two-level hierarchical linear modeling. This strategy accounts for clustering in the standard errors of income for individuals within firms (Snijders and Bosker 2012).5 All models use random effects for firms. The outcome variable is annual earnings, including salary and bonus. I do not log wages because earnings are not highly dispersed, reflecting the low ratio of highest to lowest earnings in Japanese firms relative to U.S. firms (Koike 1988). All models adjust for standard measures of human capital, including age, tenure, education, and work hours, as well as individual job classification (management track, clerical track, or irregular) and number of employees supervised.6
I assess the effect of interest by interacting respondent gender with a binary variable indicating whether women are overrepresented in subordinate roles in their firms.7 As a point of comparison, I also interact a binary variable of women’s representation in management with respondent gender.
If the gender composition of subordinate jobs is associated with gender inequality, it is also of interest how this association varies with job category. I therefore run a final model that interacts a categorical variable for gender and job type (divided into management track and nonmanagement tracks jobs of all kinds due to small sample size) with the binary variable for women’s overrepresentation in subordinate jobs. This allows us to determine whether women in both management track and in clerical track or irregular positions experience greater wage inequality in firms with female-dominated subordinate jobs.
Because the data used here are cross-sectional, I cannot make causal claims about the findings. I have argued that women’s overrepresentation in subordinate jobs may produce a devaluative organizational climate, where all women are seen as less worthy and capable and hence receive less pay. However, associations between gender inequality and subordinate job composition may be the product of other, unobserved characteristics of firms or workers. For example, HR practices may be different in firms where women are overrepresented in subordinate jobs; alternatively, women with high career ambitions may avoid firms where subordinate jobs are female dominated. However, studies of the relationship between women’s representation in management and gender inequality share these limitations. Even longitudinal studies cannot rule out time-variant differences between firms that drive both growing shares of women in management and reductions in gender inequality. Like previous research in the organizational demography tradition, this study does not conclusively demonstrate a causal relationship between the variables of interest. Instead, it establishes the promise of an analytical approach centered on the demography of subordinate jobs in the study of status processes and gender inequality.
RESULTS
Model 1 in table 3 shows the relationship between individual-level variables and earnings. Measures of human capital follow expected patterns, and, also as anticipated, wages vary significantly by job classification. For example, clerical track employees earn around ¥1.6 million ($16,000) less than management-track employees on an annual basis. Irregular jobs are not significantly associated with lower wages after adjustments for tenure because irregular workers by definition do not stay with the firm long enough for their low returns to tenure to accrue. Supervisory authority is strongly and positively associated with earnings. However, significant gender gaps in pay remain after adjustments for human capital and job characteristics. Women are predicted to earn ¥1.4 million ($14,000, or 17 percent) less than men, net of adjustment variables. Although vertical segregation and shorter tenure are frequently cited as reasons for Japan’s high levels of gender inequality, direct effects of these measures explain just over half of the gender pay gap in this study context.
Regression of Human Capital and Gender Representation on Annual Earnings
Model 2 tests whether, as I have hypothesized, gender inequality is greater where subordinate jobs are female dominated. The model reveals a large and statistically significant negative interaction between female gender and women’s share of subordinate roles. This interaction is visualized in figure 2. For women in firms where subordinate jobs are gender balanced or dominated by men, women’s wage disadvantage shrinks to around ¥0.8 million ($8,000), or 11 percent; in contrast, women’s wage disadvantage is more than three times larger, ¥2.8 million ($28,000), or 31 percent, in the firms where subordinate jobs are female dominated. In other words, in the firms where women are “the face” of subordinate jobs, women earn considerably less than men, net of human capital and job category.
Predicted Female Wage Disadvantage by Gender Composition of Subordinate Jobs (Binary Specification)
Source: Author’s calculations.
This large interaction effect is robust to a number of different modeling variations. Changing the cut point for female-dominated subordinate jobs from 60 to 70 percent increases the predicted size of the effect (model not shown). And, the effect is also apparent in a linear specification of women’s share of subordinate jobs in model 3. This model shows that, for every percentage increase in women’s share of subordinate jobs, the gender wage gap increases by ¥31,000 ($310). Figure 3 illustrates this effect.
Predicted Female Wage Disadvantage by Gender Composition of Subordinate Jobs (Linear Specification)
Source: Author’s calculations.
I also examine the relationship between gender inequality and women’s representation in management jobs, using a binary measure of female managerial representation with a cut-off point of 15 percent, following Kanter’s estimation of the percentage below which minority group members are perceived as tokens. No evidence indicates that women’s representation in management is associated with the gender wage gap, other than through its direct effects on women managers themselves.
Model 5 in table 4 investigates how gender inequality varies by job category as well as by composition of the subordinate workforce. Figure 4 presents these results. As the error bars show, estimates are imprecise due to small sample size but we can nonetheless observe some interesting patterns. Consistent with previous research that has found that female share of a job predicts low wages (England, Allison, and Wu 2007), both women and men in subordinate jobs earn considerably less than management-track men in firms where subordinate jobs are female dominated. Thus subordinate jobs are devalued more heavily, regardless of who holds them, in firms with female-dominated subordinate workforces. We also see a heavy penalty for management-track women in firms with female-dominated subordinate jobs. But, most significantly, the biggest losers in the firms with female-dominated subordinate tracks are women who themselves hold subordinate positions. Not only do these women experience the largest penalties relative to management-track men; in addition, relative to men in subordinate jobs, the gender penalty they face is also greater relative to that of their counterparts in firms where subordinate jobs are male dominated or gender balanced. Although this difference is not statistically significant, it suggests an amplification effect whereby if women significantly outnumber men in subordinate jobs, these women experience double jeopardy of lower occupational wages and steeper gender devaluation.
Regression of Job Type and Gender Representation on Annual Earnings
Predicted Wage Disadvantage Compared to Management Track Men by Gender Composition of Subordinate Jobs
Source: Author’s calculations.
Because these data are cross-sectional, the observed interactions between gender composition of subordinate jobs and gender inequality in pay are not necessarily causal. However, subjective measures of workplace environment offer a rare opportunity to further explore whether a causal relationship is plausible. Respondents to the survey were asked to answer the prompt, “My supervisors ______ value my contributions on the job,” on a 1–5 scale ranging from never to always. If subjective valuations of women and men link women’s dominance in subordinate jobs to women’s lower pay in these firms, we would expect a larger gender gap in feeling valued in the firms where subordinate jobs are female dominated. Table 5 shows the likelihood of feeling valued by gender, job type, and composition of subordinate jobs. Although women are less likely than men to say that their supervisors usually or always value their contributions in both types of firms, the difference with male peers is far more striking in firms where subordinate jobs are female dominated. In these firms, women on the management track are 19 percent less likely to feel valued, and women in subordinate jobs 27 less likely to feel valued, while in other firms, their counterparts are just 5 and 6 percent less likely than men to feel valued.
Likelihood of Feeling Valued by Firm Characteristics, Job Characteristics, and Sex
These perceptions of feeling valued by supervisors are not simply proxies for dissatisfaction with pay. Men in subordinate jobs in the firms where such jobs are female dominated report feeling valued at a high rate even though their wages are low relative to management-track men. Further, respondents were also asked to respond to the prompt “Relative to my contributions on the job, my pay is ______,” with one of the following choices: very low, somewhat low, appropriate, somewhat high, very high. Answers to this question correlate only weakly (Pearson’s r = 0.12) with responses to the question on how often supervisors value their contributions. That both objective and subjective devaluation of women occurs in the firms where subordinate jobs are female dominated, but that subjective devaluation is only weakly associated with perceptions of low pay increases confidence that devaluation causes low wages for women in these firms.
DISCUSSION AND CONCLUSION
In the United States and around the world, the gender revolution has stalled—wage gaps between men and women are no longer rapidly shrinking (England, Levine, and Mishel 2020), and at current rates it will take 135 years for gender gaps to close worldwide (Haynes 2021). As Paula England argues (2010), this is a result of the failure of gender integration in historically female jobs. Although women have moved into professional and managerial jobs, this movement is not enough to counterbalance women’s continued predominance in the more heavily populated world of low-paid service and administrative work. However, as I argue, the implications of women’s concentration in low-paid, low-status work may be even more profound than previously recognized. This is because women’s overrepresentation in these roles has symbolic and cultural implications as well as structural ones. Where all, or nearly all, assistants are women, all women risk being perceived as “mere” assistants.
A voluminous literature draws on status construction theory to examine the impact of women’s entry into management on gender inequality. However, I theorize here that this emphasis on the top of the occupational structure is misplaced. As Kanter cogently argues, many of the negative stereotypes of women, such as emotionalism or excessive focus on details, are examples of women’s behavioral responses to the dependent status of the secretarial role (Kanter 1993, 73–97). Where these jobs remain female-typed, we can expect these stereotypes to persist.
Status construction theory further posits that status beliefs emerge in interaction (Ridgeway and Markus 2022, this issue). In the corporate pyramid, there are commonly more workers at the bottom of the hierarchy than at the top. For example, in Japan, clerical workers outnumber managers by a factor of ten (JILPT 2021b), and in the United States by a factor of three (Bureau of Labor Statistics 2021a). As demonstrated empirically here, this means that workers spend only a relatively small share of their time interacting with managers. In an environment where most subordinate jobs are held by women, and where subordinate jobs, and interactions with subordinate jobs holders, are more numerous, when employees think “woman,” they are far likelier to think “assistant” than “CFO,” even if management teams are gender balanced or female dominated.
My analyses also bear out the insight that group-level representation in low-status roles is closely related to both attitudes toward, and treatment of, group members. As I show, where (nearly) all assistants are women, women feel less valued than men, and receive lower pay, regardless of their job category. The effects are severe for non-subordinate women, whose wage gap with men is more than three times greater in firms where subordinate jobs are female dominated. However, women in subordinate jobs bear the highest cost. Their enormous wage disadvantage in firms where support jobs are female dominated is a result of both higher gender penalties and higher job penalties for subordinate jobs in these firms.
The effect sizes in these analyses dwarf those found in the literature on women in management. Dragana Stojmenovska (2019), using British data, finds that a 1 percent increase in women’s share of management jobs is associated with a $24 decrease in the gender wage gap in annual earnings.8 In contrast, in the current study, a 1 percent decrease in women’s share of subordinate jobs is associated with a $310 decrease in the gender earnings gap, an effect more than twelve times as large. These large effects imply not only that the demographic makeup of subordinate jobs affects status beliefs, but also that the composition of low-status jobs matters more for status beliefs than the composition of high-status jobs.
Although the data used in this case study are cross-sectional, and unobserved firm-level characteristics may explain both the gender composition of subordinate jobs and the size of the gender pay gap, the findings highlight the imperative for and promise of further research on the relationships between subordinate job composition, status beliefs, and inequality across the postindustrial world. The theoretical insights generated here have the potential not only to illuminate the causal dynamics of the “stalled gender revolution,” but also to deepen our understanding of other axes of workplace inequality, such as race or ethnicity.
Management consultants routinely recommend increasing women’s representation in management as a tool of workplace transformation. Of course women and men deserve equal opportunities for workplace advancement, but I argue that these initiatives targeting managerial representation are unlikely to achieve these lofty goals. What, then, are organizations to do? Affirmative action for male secretaries is unlikely to prove useful. Few men apply to secretarial or other support jobs. Furthermore, any affirmative action for men runs the risk of further disadvantaging women to whom even a poorly paid secretarial job is a financial lifeline. However, Rosabeth Kanter’s early and groundbreaking study of “Indsco” offers a useful starting point. Specifically, Kanter advocated for job redesign to give the secretarial workforce greater scope to develop skills and build careers (1993, 267–81). To degender the workplace, employers will need to delegate highly valued tasks to low-level workers, and more equitably distribute the labor of undervalued tasks across levels of the organizational hierarchy.
Restarting the “stalled gender revolution” will require scholars to push back against status beliefs that predispose us to assume that the composition of male-typed managerial jobs is necessarily the most important determinate of status beliefs, and organizations to reconfigure subordinate and support jobs to be rewarding and powerful enough to appeal to both men and women. Although undoing gendered patterns at the top of an organization may help all women to a degree, undoing gendered patterns at the bottom, as Kanter recommended more than forty years ago, may help more women, more consequentially.
FOOTNOTES
↵1. If power is the mechanism, it may narrowly benefit the relatively small cadre of management-adjacent women professionals rather than women as a whole.
↵2. The number of female workers in clerical occupations exceeds the total number of managerial workers by a factor of six (JILPT 2021b).
↵3. Because these data were collected to understand foreign workers’ workplace integration (see Holbrow 2020), the teams were selected based on the presence of white-collar foreign workers; team selection is thus unrelated to gender composition of the teams.
↵4. Shares of women in management in the firms are strongly correlated with women’s share of management in the surveyed teams (Pearson’s r = 0.77), indicating that the surveyed teams are representative of their firms as a whole.
↵5. Within firms, variation is minimal in wages at the team level, so the models do not include random effects for teams.
↵6. One potential concern with this modeling strategy is that the small number of level two units (firms) may bias the point estimates and standard errors of the cross-level interactions, leading to a higher rate of type one errors (Snijders and Bosker 2012). To mitigate these concerns, I use Restricted Maximum Likelihood (REML) estimation, with a Kenward-Roger adjustment (Kenward and Roger 1997, 2009). Although traditional maximum likelihood estimation requires twenty to thirty clusters to generate unbiased estimates, REML with the Kenward-Roger adjustment produces unbiased estimates with as few as ten clusters (McNeish and Stapleton 2016; McNeish and Harring 2017).
↵7. Data on firm-wide share of women in subordinate roles is not available, but the comparison of women’s share of management in the sample and the firms as a whole shows a strong correlation between the sample-level and firm-level representation, indicating that sample-derived data are an adequate proxy.
↵8. This is calculated by multiplying the effect size, which is displayed as weekly earnings, by fifty-two.
- © 2022 Russell Sage Foundation. Holbrow, Hilary J. 2022. “When All Assistants Are Women, Are All Women Assistants? Gender Inequality and the Gender Composition of Support Roles.” RSF: The Russell Sage Foundation Journal of the Social Sciences 8(7): 28–47. DOI: 10.7758/RSF.2022.8.7.02. The author gratefully acknowledges financial support from the Fulbright Foundation as well the invaluable introductions provided by Akinari Horii and Jun Kurihara at the Canon Institute for Global Studies (CIGS), and by Shinju Fujihira at Harvard. This research would not have been possible without them and the support of numerous staff at the Canon Institute, the Japan Association of Corporate Executives (JACE), and the companies that participated in the survey. I am particularly indebted to Yasuko Nagatsuka of CIGS, Natsuko Kasahara of JACE, and to Keiko Shibatani. Their intelligence and professionalism brought the project from idea to reality. Hazel Markus, Suzanne Nichols, Cecilia Ridgeway, Jocelyn Viterna, the Russell Sage Foundation conference participants, and the anonymous reviewers provided insightful feedback. Direct correspondence to: Hilary Holbrow, at hilholbr{at}iu.edu, Department of East Asian Languages and Cultures, Hamilton Lugar School of Global and International Studies, Indiana University, 355 N. Eagleson Ave., Bloomington, IN 47405, United States.
Open Access Policy: RSF: The Russell Sage Foundation Journal of the Social Sciences is an open access journal. This article is published under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.
REFERENCES
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.