Manufacturing Gender Inequality in the New Economy: High School Training for Work in Blue-Collar Communities

Abstract

Tensions between the demands of the knowledge-based economy and remaining, blue-collar jobs underlie renewed debates about whether schools should emphasize career and technical training or college-preparatory curricula. We add a gendered lens to this issue, given the male-dominated nature of blue-collar jobs and women’s greater returns to college. Using the ELS:2002, this study exploits spatial variation in school curricula and jobs to investigate local dynamics that shape gender stratification. Results suggest a link between high school training and jobs in blue-collar communities that structures patterns of gender inequality into early adulthood. Although high school training in blue-collar communities reduced both men’s and women’s odds of four-year college enrollment, it had gender-divergent labor market consequences. Men in blue-collar communities took more blue-collar courses, had higher rates of blue-collar employment, and earned similar wages relative to otherwise comparable men from non-blue-collar communities. Women were less likely to work and to be employed in professional occupations, and they suffered severe wage penalties relative to their male peers and women from non-blue-collar communities. These relationships were due partly to high schools in blue-collar communities offering more blue-collar and fewer advanced college-preparatory courses. This curricular tradeoff may benefit men, but it appears to disadvantage women.

The link between schools and the labor force is key to the functioning of every modern society, and it is a central mechanism in the production of social stratification. Since the late nineteenth century in the United States, heated debates have focused on whether training should emphasize vocational skills or promote a rigorous academic curriculum. In the aftermath of the Great Recession, the debate is evident in the tension between the demands for a globally competitive, innovative workforce and training for remaining pockets of blue-collar jobs.

A bachelor’s degree is required for most “good” jobs today (see Carnevale and Desrochers 2002; Hout 2012; Kalleberg 2011), but scholars and policymakers are increasingly reemphasizing career and technical education (CTE) and advocating mid-skill, sub- baccalaureate jobs as smart alternatives to those requiring a four-year college degree (Symonds, Schwartz, and Ferguson 2011). However, compared to most jobs that require a bachelor’s degree, sub-baccalaureate jobs are more highly gender-segregated, and those that pay best are male-dominated and concentrated within blue-collar fields (Carnevale, Rose, and Hanson 2012; Carnevale et al. 2011). Indeed, blue-collar jobs are the most touted exemplars of sub-baccalaureate “good” jobs, as exemplified in President Obama’s incendiary statement, “But I promise you, folks make a lot more—potentially—with the skilled trades and manufacturing than with an art history degree.”¹ Given that women receive greater returns to higher education (DiPrete and Buchmann 2006; Jacob 2002) and the gender-segregated nature of blue-collar jobs, an emphasis on blue-collar training that is not complemented by academically rigorous coursework may heighten gender inequalities in the labor market. This article investigates whether an emphasis on blue-collar high school training in communities with existing blue-collar jobs benefits men and women.

We take advantage of the local nature of schools to investigate the gendered effects of a tighter link between high school training and jobs in the community, focusing specifically on blue-collar communities—modern day “company towns.” Scholarship describing connections between schools and communities has a rich sociological tradition, but in the 1960s, research began to emphasize the individual-level predictors of educational and labor force outcomes and how schools function to serve national goals, downplaying the potentially important role of place (Arum 2000). Alongside this changing tide in academic research, policy initiatives, from the 1958 National Defense Education Act to Rising Above the Gathering Storm (National Academy of Sciences 2007) and Common Core State Standards, have emphasized the need for schools to meet national labor market demands and produce a globally competitive workforce. These initiatives led to more rigorous and standardized academic curricula across schools. At the same time, public and political discourse has spotlighted recent state- and local-level policies that bolster high school blue-collar training aligned with local blue-collar industries, with some concurrently relaxing math and foreign language academic requirements. These diverging curricular trends are not surprising, given research showing that the geographic distribution of jobs requiring higher and lower levels of human capital is more uneven today than ever (Moretti 2013). Despite this spatial patterning of jobs and the male-dominated nature of most decent-paying sub-baccalaureate occupations, renewed debates about whether high schools should promote vocational skills or rigorous college-preparatory coursework are largely gender and space neutral.

The current study applies a gendered and spatial perspective to this classic sociological issue. Using the Educational Longitudinal Study of 2002 (ELS:2002) linked to the U.S. Census 2000, we draw on local variation in jobs and high school curricula to investigate the gendered education and labor market consequences of linking high school training to jobs in the local economy. Specifically, we examine gender disparities in blue-collar and college-preparatory high school coursework, postsecondary matriculation, and early adulthood labor force outcomes across counties with higher and lower shares of workers employed in blue-collar jobs (i.e., construction and extraction; production; transportation and material moving; and installation, maintenance, and repair).

Taken together, our results reveal a link between high school training and local jobs in blue-collar communities that may facilitate a school-to-work link for young men but limits the postsecondary and labor force options for women. This study advances scholarship on high school stratification and the transition to college and work by uncovering the gendered, spatial inequalities that may emerge from a tight connection between high school training and local work, particularly in communities with highly gendered labor market opportunities. The local patterns of gender inequality we find suggest that place conditions the “female advantage” and highlight the importance of considering how schools and communities interact to shape gender stratification.

BACKGROUND

Studies rooted in status-attainment and reproduction paradigms have focused on how schools and schooling contribute to individual-level differences in the division of labor along class, gender, and racial/ethnic lines. Yet, an occupational division of labor also takes place at a spatial level, in which types of industries and jobs are unevenly distributed across communities (Hanson and Pratt 1992; Herod 2001). Research suggests that this spatial division of labor is gendered, such that men’s and women’s opportunities for good jobs vary across places that rely more or less heavily on gender-segregated industries (Gauchat, Kelly, and Wallace 2012; Massey 1984; McCall 2001). Studies that directly examine how local economic opportunities shape between- and within-gender inequalities in labor market outcomes (e.g., Cotter et al. 1998; Gauchat et al. 2012; McCall 2000, 2001) have not considered how local differences in high school training may foreshadow gender inequalities in the labor force. At the same time, the rich body of work on the educational predictors of labor force inequalities largely ignores the local nature of schools and labor markets. Moreover, research on gender disparities in education pays little attention to local labor market conditions, with only a few exceptions (Riegle-Crumb and Moore 2014; Werum 2002).

Returning to a sociological tradition that highlights the embedded nature of schools within communities (Arum 2000), we examine gender stratification as a function of place and investigate the role of high school training in shaping gender inequality across communities. The current study focuses on blue-collar communities and compares young men’s and women’s educational and labor market outcomes across communities with higher and lower concentrations of blue-collar workers. We pay particular attention to whether schools—through offering curricula more oriented toward blue-collar jobs and less oriented toward four-year college-going—benefit men and women equally or (re)produce gender stratification in the labor market. Indeed, schools are not only microcosms of society that train students for highly differentiated jobs demanded by the national economy; schools are also microcosms of the communities they serve and have historically geared training to the skills demands of local jobs in ways that reinforced the gender-segmented labor market (Rury 1991).

Blue-Collar Communities and Occupations

Blue-collar jobs have played a vital role for U.S. families, helping sustain a strong working-class and a powerful national economy through the mid-twentieth century. Thick descriptions of the local, complex social and economic dynamics of places that relied heavily on blue-collar jobs have focused primarily on racial/ethnic and class relationships (e.g., Heberle 1948; Kornblum 1974; Leggett 1968; Lewis 1987; Lynd and Lynd 1929; Royster 2003). Like research on gender-segregation in blue-collar jobs, studies examining gender dynamics within blue-collar communities have investigated women’s job preferences, work environments, and institutional discrimination (Connell 1990; Deaux and Ullman 1983; Rosen 1987). Additionally, quantitative work has documented larger gender disparities in earnings and wages in local economies with stronger manufacturing bases (Gauchat et al. 2012; McCall 2001). However, to our knowledge, only one study of gender in a blue-collar town examined gendered processes both before and after labor force entry; this ethnography—in sharp contrast to the aforementioned research documenting gender inequality in blue-collar occupations—sheds a positive light on women’s experiences in blue-collar work and identifies family socialization as a primary force underlying women “pioneers’” occupational choices (Walshok 1981). The current study departs from these prior studies by examining how potentially gendered educational opportunities in high schools within blue-collar communities shape gender stratification in education and work in early adulthood.

After almost a half-century of declines in blue-collar jobs, to what extent do blue-collar communities exist in today’s economy, and where are they located? Figure 1 shows a map of the concentration of blue-collar workers by county using occupational information from the U.S. Census 2000. Counties with higher concentrations of blue-collar workers are darker; counties with lower concentrations of blue-collar workers are lighter.² The Southeast and Midwest have the greatest concentrations of counties with high shares of blue-collar workers; the West and Northeast have the fewest communities with high concentrations of these workers. These regional patterns are consistent with recent research highlighting the geographic divergence between places with high concentrations of high-tech, creative jobs and former manufacturing towns (Moretti 2013), increasing educational segregation across the United States (Domina 2006), and spatial stratification in opportunities for upward mobility (Chetty et al. 2014). We investigate the case of economically isolated, blue-collar communities in which between one-third and one-half of all denizens work in blue-collar jobs. We anticipate that these local economies may have a gendered school-to-work pattern, given that almost 90 percent of blue-collar job-holders are men (Bose and Whaley 2000).

Figure 1.

Blue-Collar Communities in 2000m

Note: Authors’ calculations from U.S. Census 2000.

Our analysis of educational and work outcomes of men and women who attended high school in blue-collar communities draws on three main motivations. First, to the extent that schools offer courses that relate to local jobs, we expect this relationship to be especially strong within blue-collar communities. Blue-collar jobs are among the only jobs that require specialized skills and only a high school degree, and historically, the link between high school training and local industry was tightest in the Northeastern urban industrial centers—where blue-collar jobs dominated the local economy. This emphasis on local jobs resulted in early differences in academic and vocational course offerings and course-taking across regional economies (Rury 1991). Furthermore, the sub-baccalaureate labor market is largely local, with employers searching for workers locally and developing relationships with community educational providers (Grubb 1999).

Second, if schools in blue-collar communities offer greater numbers of blue-collar courses and fewer advanced academic courses relative to schools in non-blue-collar communities, this gendered curricula may restrict women’s educational and labor market opportunities. The majority of today’s better-paying sub-baccalaureate jobs are male-dominated (Carnevale et al. 2012; Carnevale et al. 2011), the returns to education are higher for women than for men (DiPrete and Buchmann 2006), and striking wage gaps exist among women with and without college degrees (McCall 2000, 2001). Research shows that women have fewer opportunities to obtain training for employment in blue-collar jobs, and the few women entering blue-collar jobs face resistance from male co-workers and are primarily relegated to low-paying, low-status positions (Bergmann 2011; O’Farrell 1999; Padavic and Reskin 2002; Rosen 1987). If the promise of decent-paying blue-collar work falls short for women, a curriculum that may foster a school-to-work link for men but does not prepare women for four-year college may result in gendered payoffs in the labor market.

Finally, our focus on the gendered consequences of linking high school training and local jobs presents a unique opportunity to reduce some of the selection bias concerns inherent in studies examining the effects of CTE and academic course-taking. Because students select into coursework based on unobservable characteristics that may be endogenous to the outcomes of interest, examining whether taking blue-collar or college-preparatory math courses shapes students’ college matriculation and occupational outcomes, for example, is problematic for causal inference. Our study focuses on how school curricular opportunities and students’ course-taking differ depending on the local labor market, and we expect that school course offerings and the local labor market both influence students’ course-taking and their postsecondary outcomes. Evidence discussed in the section on robustness checks supports our assumption that the choice of the local labor market or high school within the local labor market does not vary for young men and women (or for parents of sons versus daughters). Consequently, the gender-divergent education and work outcomes that we observe among young men and women who attended high school in blue-collar communities are likely not a function of gendered selection processes. This gives us greater confidence that our results tap the possible effects of CTE and academic coursework on young men’s and women’s postsecondary transitions to college and work under conditions when there are good sub-baccalaureate jobs.

Overall, a gender analysis of the link between blue-collar high school training and blue-collar jobs is timely and needed, given recent political discussions and legislative initiatives promoting this coupling. Some scholars describe optimism about blue-collar work, and in particular manufacturing, as a form of “nostalgia” (McCall 1998:401) and misplaced in our expanding knowledge-based economy (Moretti 2013), whereas others argue there are growing numbers of often well-paying blue-collar jobs that must be filled by trained workers. Legislators and scholars have charged that students are not acquiring the specialized training in high school or community college that sub-baccalaureate jobs demand. We examine whether young women’s education and labor market outcomes suffer when they attend high school in communities with a tight link between high school training and local blue-collar jobs at a point when national trends show that women are exceeding men in educational attainment (DiPrete and Buchmann 2013).

Blue-Collar Training in Blue-Collar Communities

Prior literature asserts that the development of school curricula and distribution of knowledge are not neutral processes (Ainsworth and Roscigno 2005; Bowles and Gintis 1976; Oakes 1985). For example, Apple (2004:28) states that “the knowledge made available (and not made available) to students” must be problematized through examining the “linkages between economic and political power and school curricula.” Historically, local economic interests and powerful business leaders shaped local schools’ curricular orientations. Tight linkages between local blue-collar jobs and blue-collar vocational training were forged as early as the late nineteenth century in industrial towns (Bartlett et al. 2002).

Yet, training for blue-collar jobs was historically aimed at and reserved for men. An early example of this is the federal government’s provision of vocational education funds in the early twentieth century, which emphasized blue-collar training for men but restricted women’s vocational options to often poorly funded home economics training (Werum 2002). In addition, men’s and women’s educational opportunities during this time “were shaped by the regional division of labor dictated by the North American economy” (Rury 1991:9). For example, Boston school boards built all-male mechanical arts schools and women’s schools centered on female-typical work, while students on the West coast appeared to have greater access to academic coursework (Rury 1991).

To what extent does a gendered curriculum related to local jobs exist in today’s blue-collar communities? Some work suggests that a relationship between local economies and school curriculum may persist today. As an explanation for differences in educational investments between urban and rural schools, including Advanced Placement (AP) course offerings, Roscigno, Tomaskovic-Devey, and Crowley (2006:2124) speculate that administrators may “invest resources in a manner consistent with the perceived needs of the local population and local labor markets.” In other words, schools may emphasize particular types of coursework as a way of preparing the next generation of local workers, as Ainsworth and Roscigno (2005:266) intimate but do not empirically examine. Oakes and Guiton (1995), in a case-study of high school tracking, note that a major goal of the district was to offer vocational courses that aligned with local economic demands. Thus, school decision-makers’ curricular investments may be circumscribed by the realities of local economic opportunities. In addition, blue-collar coursework may have higher status relative to academically rigorous curricula in blue-collar communities if a culture rooted in local “stratification histories” (Roscigno et al. 2006:2138) shapes school course offerings.

Business–school partnerships and economic stakeholders’ influence on school decision-makers may also shape what schools teach (Ray and Mickelson 1990; Shea, Kahane, and Sola 1990). In fact, many states in the South and Midwest are proposing or have passed legislation allowing local industries—many in blue-collar industries—to design high school courses that teach students job-related skills. For example, several states, including Michigan and Texas, have relaxed academic requirements to grant students greater flexibility in taking industry-designed courses, such as construction and welding classes. Similarly, the German model of apprenticeships and industry-certification programs are gaining popularity as avenues to connect students with these local blue-collar job opportunities, such as “Youth Apprenticeship Carolina” in South Carolina, “Urban Skilled Trades Connection” in Wisconsin, and the “Jump Start” program in Louisiana.

We might expect only male students’ blue-collar course-taking to be shaped by a curriculum that more heavily emphasizes blue-collar coursework. Trade and technical course-taking in general (Arum and Shavit 1995), and blue-collar course-taking specifically, has historically been and continues to be highly male-dominated (Ainsworth and Roscigno 2005; Werum 2002). Gender disproportionality in blue-collar communities may be especially pronounced for a variety of reasons. For example, institutional sorting processes—net of other factors—contribute to racial/ethnic, gender, and class differences in course-taking that tend to reflect the uneven representation of status-groups across occupations (Ainsworth and Roscigno 2005; Oakes 1985, 2005; Tyson 2011). If teachers and counselors aim to prepare students for job opportunities in the local labor market (Oakes and Guiton 1995), guiding males into male-stereotypical CTE coursework may be even more pronounced in blue-collar communities.

Local scripts about gender-appropriate work may also shape differences in blue-collar course-taking patterns by gender and across communities. For example, Morris (2012) found that male students living in a town that once relied on the coal industry associated masculinity with working in a manual labor job after high school. In addition, to the extent that high school students make educational and occupational decisions based on the educational and skills requirements of local jobs, as some studies suggest (Bozick 2009; Sewell and Orenstein 1965), male students in blue-collar communities may be most likely to invest in blue-collar coursework, given greater local blue-collar work opportunities and the highly gender-segregated nature of these jobs.

If schools in blue-collar communities offer fewer advanced academic courses, both male and female students’ academic course-taking may be affected. The relationship between school course offerings and student course-taking may be direct if schools offer fewer academically rigorous courses or an emphasis on vocational courses translates into fewer advanced academic electives. From earlier sociological work on tracking (e.g., Gamoran and Mare 1989; Oakes 1985) and public and private schools (cf. Bryk, Lee, and Holland 1993; Coleman and Hoffer 1987) to more recent debates on the Common Core, the link between academically rigorous course offerings and student course-taking is well-established. This link may operate through the constrained curriculum hypothesis—the idea that students are less likely to take academically rigorous courses in schools that provide more alternatives to academically intensive courses (Lee, Smith, and Croninger 1997; Powell, Farrar, and Cohen 1985). In addition, the relative academic rigor of school curricula may cultivate different school climates (McDonough 1997) that shape young men’s and women’s educational choices (Legewie and DiPrete 2014).

Gender, Postsecondary Preparation, and Labor Force Outcomes

High school course-taking sets the stage for students’ postsecondary and labor force outcomes. Advanced academic coursework, and advanced math in particular, are powerful predictors of four-year college matriculation and bachelor’s degree completion (Adelman 2006). If young men and women in blue-collar communities are less likely to enroll in advanced academic courses, we would expect them to have lower four-year college enrollment rates relative to their peers in non-blue-collar communities.

At the same time, based on school-to-work scholarship, we might expect gender-divergent outcomes related to blue-collar course-taking. Research finds that people with vocational training have a higher probability of finding employment in a skilled job (Arum and Shavit 1995), especially when the training is career-specific (Bishop and Mane 2004; Shavit and Muller 1998) and matches students’ post-high school occupations (Grubb 1999). Therefore, high school training in blue-collar communities geared more toward blue-collar training and less toward college-preparation may not harm the labor market outcomes of men, whom we expect to enroll in these gender-typed, blue-collar courses at higher rates.

Women who attend high school in blue-collar communities may not enjoy this safety net and alternative route to decent wages if they do not invest in blue-collar coursework. At the same time, even young women who enroll in blue-collar courses may not reap the same benefits that men do, as research suggests that men (Grasso and Shea 1979)—and in particular white men—experience the greatest labor market returns to blue-collar training (Ainsworth and Roscigno 2005; Royster 2003). Some studies have highlighted supply-side factors, including women’s preferences for female-typical work (England 2010), as partial explanations for women’s lower representation in blue-collar occupations, whereas other research has emphasized the role of structural factors. Similar to the factors that restrict black men’s opportunities in blue-collar jobs (Royster 2003), studies suggest that women encounter limited access to critical informal networks and apprenticeships that lead to the best-paying blue-collar jobs (O’Farrell 1999; Padavic and Reskin 2002; Rosen 1987; Walshok 1981). Moreover, this research indicates that the few women who obtain employment in blue-collar jobs are typically concentrated in lower-status positions where they earn low wages and face male resistance (Bergmann 2011; Rosen 1987). Studies also suggest the spatial expression of these patterns. For example, Gauchat and colleagues (2012) find that only men experience earnings benefits in areas with higher concentrations of manufacturing jobs, and McCall (2001) finds that places characterized by a stronger blue-collar workforce may result in greater wage disparities among men and women, especially among people without a college degree.

Despite this research, current discourse encouraging blue-collar training ignores the gender-segregated nature of these occupations and implicitly assumes that blue-collar occupations are good employment options for men and women. To investigate this proposition for a Millennial cohort, we use the Current Population Survey (CPS) to calculate the gender gap in 2011 wages across major occupation groups among high school graduates age 25 to 28.³ Figure 2 shows the average hourly wages of men and women across white-collar, blue-collar, service, and other non-professional occupations.

Figure 2.

Average Hourly Wages in 2011 by Occupation for Men and Women, Ages 25 to 28 (wage gap shown above bars)

Note: N = 8,782. Authors’ calculations from the March 2012 Current Population Survey.

Wages for men and women are highest in white-collar professions and lowest in service jobs, with the largest gender pay-gap occurring within blue-collar occupations. These data are consistent with research showing that the economic stakes of earning a bachelor’s degree are especially high for women (DiPrete and Buchmann 2006). Thus, while blue-collar training may lead to relatively good wages for men, women in blue-collar communities may suffer labor market penalties if their schools do not offer an academically rigorous curriculum that prepares them for a four-year college degree.

RESEARCH QUESTIONS

We investigate whether the connection between high school training and local jobs in the context of blue-collar communities (re)produces gender inequalities in the labor market. Specifically, we use ELS:2002 to analyze the following questions:

• (1a)Do male and female students in communities with higher concentrations of blue-collar workers take greater numbers of blue-collar courses and have lower odds of taking advanced academic courses relative to their peers in communities with lower concentrations of blue-collar workers?
• (1b)To what extent are these relationships a function of differences in school course offerings?
• (2)Does high school training in blue-collar communities benefit both men and women in the labor market? We consider postsecondary and employment outcomes two years after high school and examine labor force outcomes eight years out of high school, when most young adults are in the labor force.

DATA AND METHODS

We use data from the Educational Longitudinal Study of 2002 (ELS:2002), a nationally representative study of 16,197 sophomores in high school, designed by the National Center for Education Statistics (NCES) (see Ingels et al. 2014). This cohort is tracked through early adulthood with follow-up surveys conducted in 2004, 2006, and 2012. These data offer high school transcripts and information on sociodemographic background, postsecondary education, and labor market outcomes. Course offering catalogs are also available for almost 90 percent of the schools attended by survey respondents. We restrict our analyses to high school sophomores attending public schools in the base-year survey (N = 12,770). Ancillary analyses referred to here are available upon request.

We obtained local labor market information by linking schools in ELS:2002 to U.S. Census 2000 county-level data (Minnesota Population Center 2011), the year that most participating students entered high school. We gathered district per-pupil expenditures from the 2000 to 2001 Common Core of Data, which the NCES collects annually for U.S. public schools and districts (see Chip and Cohen 2004).

Measures

Blue-collar communities

We measure the independent variable of interest using the percentage of employed civilians over age 16 in the county working in blue-collar occupations in the Census 2000. The Census 2000 classified occupational groups based on the 2000 Standard Occupational Classification (SOC) codes, and we use two-digit major occupation groups to characterize blue-collar occupations. Following reports from the Bureau of Labor Statistics (Earle et al. 2014) and academic research (e.g., Farley and Haaga 2005), we define blue-collar occupations as the following major occupation groups: construction and extraction; production; transportation and material moving; and installation, maintenance, and repair.⁴ For all analyses, this variable is transformed into quartiles, and we refer to counties in the 4th quartile as “blue-collar communities” and those in the 1st quartile as “non-blue-collar communities.”⁵

Blue-Collar and Advanced Math High School Training

School course offerings

We attempt to capture the blue-collar emphasis and academic rigor of schools with blue-collar, AP/IB (Advanced Placement/International Baccalaureate), and advanced math course offerings, measured as the number of a school’s course offerings falling within each category.⁶ We identified course offerings within the school course catalog by their Classification of Secondary School Courses (CSSC) code—a hierarchical, six-digit code that indicates a course’s main area of study, the subcategory of study, and specific name. The advanced math measure includes non-AP/IB academic math courses above Algebra 2, and the AP/IB measure includes all AP and IB courses. Among CTE courses, we consider courses “blue-collar” only if they fall under the following CSSC categories that align with our definition of blue-collar occupations: construction trades, mechanics and repairers, precision production, transportation and material moving, and industrial arts.⁷ Blue-collar and AP/IB course offerings are logged to account for their skewed distributions. Analyses controlling for course offering variables include a control for the total number of course offerings to account for variation in the number of courses schools offer.

Student course-taking

To assess the degree to which respondents’ high school coursework relates to high school course offerings and to blue-collar work, we construct measures of student blue-collar and academic course-taking using students’ high school transcripts and CSSC codes. Blue-collar courses are defined in the same way as the blue-collar course offerings variable. This variable is logged due to its skewed distribution. We measure advanced math course- taking with a dichotomous measure indicating whether a student attempted a math course beyond Algebra 2 by the end of high school, a college preparation indicator that is the strongest predictor of college enrollment and completion (Adelman 2006). We constructed this variable using the approach by Riegle-Crumb and Grodsky (2010).

Gendered Postsecondary and Labor Force Outcomes in Blue-Collar Communities

We address our second research question—whether high school training in blue-collar communities benefits both men and women—by investigating postsecondary and employment outcomes in 2006, when most sample members were two years out of high school, and labor force outcomes reported in 2012, eight years after the cohort’s expected high school graduation year. We constructed a categorical measure indicating the highest level of education attempted for respondents who had enrolled in a postsecondary institution, and employment outcomes for respondents who had never enrolled in a postsecondary institution as of 2006. Categories are enrolled in a four-year institution, not working, employed in a job other than blue-collar, employed in a blue-collar job, and enrolled in a two-year institution. NCES coded respondents’ occupations using Occupational Information Network (O*NET) two- and six-digit codes. These codes align with SOC codes, which the Census 2000 used to define occupations (Levine et al. 2000). Thus, the codes used to categorize respondents in blue-collar occupations match those we used to define our measure of blue-collar communities described earlier. Due to small cell sizes, we collapse the out of the labor force and unemployed categories for 2006 and 2012 reports and refer to this population as “not working.”

We use the third follow-up of the ELS:2002, collected eight years after students’ expected high school graduation, to model respondents’ labor force outcomes in early adulthood. First, we predict students’ employment and occupation outcomes, measured with a categorical dependent variable indicating whether the respondent was not working and respondent’s occupation if employed in 2012 (the reference category is white-collar and includes respondents in professional, managerial, business, and finance occupations; other categories are service, blue-collar, other non-professional, and not working). Second, we predict hourly wages, which the NCES constructed using respondents’ 2011 reported annual earnings, the most recent year for which annual earnings were reported. We follow convention and log this dependent variable. A minority of respondents reported taking courses at a post-secondary institution during the third follow-up. We include this population in our analyses and control for current enrollment status, but estimates excluding this group yield nearly identical results.

Additional Covariates

All analyses control for the following sociodemographic factors: race/ethnicity, highest parental education, parental income, and family structure. Models control for the percentage of employed civilians over age 16 in the county who are (1) working in low-wage service jobs and (2) unemployed, constructed from Census 2000 county-level data. We control for students’ base-year math achievement test score and academic background, including 9th-grade cumulative GPA (four-point scale), 9th-grade math course level (see Riegle-Crumb and Grodsky 2010), and whether a student transferred schools between the base-year and first follow-up surveys. Analyses also control for school urbanicity, district per-pupil expenditures, percentage of students eligible for free- or reduced-price lunch, school percent minority, vocational and magnet school attendance, and an indicator of whether a respondent’s high school provided a course catalog. Weighted descriptive statistics are included in Tables A1 and A2 in the Appendix.

Analytic Plan

Our first research question investigates whether students in blue-collar communities take more blue-collar courses and less academically rigorous coursework, while paying close attention to the role of course offerings in attenuating any observed course-taking differences. We use Ordinary Least Squares (OLS) regression to predict the number of blue-collar courses (logged)⁸ that students take during high school. We use logistic regression to predict students’ probability of taking an advanced math course by the end of high school. Models are nested to assess whether differences in school course offerings reduce the observed associations between the local concentration of blue-collar workers and course-taking. Although we do not present models estimating the relationship between the local labor market and high school course offerings, ancillary school-level analyses indicate that high schools in blue-collar communities offer greater numbers of blue-collar courses and fewer advanced math and AP/IB courses, controlling for a host of appropriate school-level factors (e.g., urbanicity, per-pupil expenditures, percent free- or reduced-price lunch). Models predicting course-taking are restricted to public school students with high school transcript information (n = 11,610).

To address our second research question, we first examine whether the tight link between high school training and local blue-collar jobs in blue-collar communities benefits men and women by examining respondents’ labor market and postsecondary outcomes two years after expected high school graduation. We use multinomial logistic regressions to predict respondents’ post-high school destinations, and we nest models to assess whether school course offerings and student course-taking attenuate the relationship between the local labor market and respondents’ outcomes. Models predicting post-high school destination are restricted to public high school sophomores who graduated from high school and participated in the 2002 base-year and 2006 follow-up surveys (n = 10,080).

Our final models examine whether high school training in blue-collar communities has gender-divergent consequences in the labor force. We estimate men’s and women’s hourly wages in 2011 and employment status and occupation in 2012, after sample members have had the opportunity to complete postsecondary education. We use multinomial logistic regressions to predict respondents’ 2012 employment and occupation and Ordinary Least Squares (OLS) regression to predict logged hourly wages in 2011. Models predicting labor force outcomes are restricted to public high school sophomore students who graduated from high school and participated in the 2002 base-year and 2012 follow-up surveys (n = 9,190), and the wage analysis is restricted to high school graduates who reported non-zero wages (n = 8,510).

Because our sample selection filters may bias our estimates, we use a strategy rooted in the Heckman two-step selection correction logic in an attempt to address this problem. For our post-high school destinations and labor force analyses, we used probit models to estimate men’s and women’s likelihoods of being a high school graduate with a host of covariates listed in the descriptive statistics table. For our wage analyses, we used probit models to predict the likelihood of being a high school graduate and having non-zero wages; in these models, we also include measures of marital status and having children. From these probit models, we computed inverse Mills ratios (IMRs), or the hazard rate of not being included in the sample. Analyses for respondents’ post-high school destinations, occupations, and wages include the IMR as a regressor, but results are consistent without this adjustment.

Given the gendered nature of educational and occupational processes, especially gender-segregation within blue-collar jobs and training, we estimate all models separately by gender for ease of interpretation. In ancillary analyses, we estimate pooled models with interactions by gender to test for between-gender differences within blue-collar communities; we note these differences when significant. We report logit coefficients as average marginal effects (AME), which are not susceptible to changes in unobserved heterogeneity across logistic regression models (Mood 2010).

Our models incorporate appropriate student sample weights and adjust for clustering within schools. Analyses using Hierarchical Linear Modeling (HLM) produce substantively identical results. We use multiple imputation to handle missing data on independent variables. In an attempt to address time-invariant, unobserved heterogeneity between the states in which schools are located, all analyses include state fixed-effects. We report selected coefficients from our analyses within the text and present full models in the Appendix.

Robustness Checks

Given the large body of scholarship documenting a link between parents’ occupation and students’ education and work outcomes, we estimated models that control for parental occupation—specifically, whether respondents’ parents held blue-collar jobs. Our results are robust to this measure, and we do not include it due to its strong association with parental education. For analyses estimating post-high school outcomes, we tested models that include a control for whether a student attended high school in a county with a four-year college as a proxy for postsecondary access; accounting for this measure does not alter the substantive conclusions of our reported results. To examine the possibility that our results are related to young people moving out of their communities, we estimated models predicting early adulthood outcomes controlling for residential mobility between counties and distance moved between the base-year and early adulthood; results from those models are nearly identical to results reported here. In addition, we investigated the possibility that differential child-bearing rates in high school between women from blue-collar and non-blue-collar communities may be driving observed differences in outcomes; our results across outcomes are robust to the inclusion of family formation indicators during high school.

Given research documenting gender sorting across high schools within counties (Long and Conger 2013), we also explored the possibility that boys and girls (and their parents) in blue-collar communities select into schools with different curricular emphases. However, we found no statistically significant differences in the advanced academic and blue-collar course offerings of the schools that boys and girls in blue-collar communities attend. Additionally, results controlling for a crude proxy of school choice—the number of public and private high schools within counties—yield substantive conclusions identical to those presented here.

RESULTS

Blue-Collar and Advanced Math Course-Taking

Our first research question focuses on whether students’ high school training in blue-collar communities differs from that of their counterparts in non-blue-collar communities. In Table 1, we investigate whether students attending high schools in blue-collar communities (4th quartile, percent blue-collar workers in county) take greater numbers of blue-collar courses and have lower odds of taking advanced academic math relative to their peers in non-blue-collar communities (1st quartile, percent blue-collar workers in county). We pay particular attention to the role of high school course offerings in attenuating any observed associations.

Table 1.

Coefficients from Predicting Course-Taking, by Gender: Number of Blue-Collar Courses (OLS) and Advanced Math (Logistic Regressions, Reported as Average Marginal Effects)

	Male Students				Female Students
	# Blue-Collar (logged)		Advanced Math		# Blue-Collar (logged)		Advanced Math
	(1)	(2)	(1)	(2)	(1)	(2)	(1)	(2)
Percent Blue-Collar Workers in County (ref.: 1st Quartile)
2nd Quartile	.091 (.050)	.078 (.048)	−.001 (.028)	−.003 (.028)	.047 (.025)	.044 (.026)	−.032 (.028)	−.033 (.028)
3rd Quartile	.116* (.058)	.060 (.056)	−.077** (.028)	−.061* (.027)	−.003 (.022)	−.019 (.023)	−.072* (.030)	−.056 (.030)
4th Quartile	.240** (.069)	.166* (.070)	−.087** (.031)	−.079** (.031)	.043 (.036)	.015 (.036)	−.087* (.036)	−.060 (.036)
Course Offerings
Number of BC courses offered (logged)		.148*** (.022)		−.007 (.009)		.022* (.011)		−.008 (.011)
Number of AP/IB offered (logged)		−.048 (.031)		.009 (.012)		−.032 (.017)		.037** (.013)
Number of advanced math offered		.004 (.010)		.025*** (.005)		−.003 (.004)		.017** (.006)
9th-grade math level	.002 (.009)	−.002 (.009)	.077*** (.007)	.077*** (.007)	.000 (.004)	−.000 (.004)	.072*** (.008)	.073*** (.008)
Constant	.707*** (.199)	.706*** (.197)			.161 (.086)	.200 (.102)
R2	.18	.20			.11	.12
Observations	5,730		5,700		5,880		5,860

Note: Standard errors are in parentheses. All models control for total number of courses offered by the school, percent low-wage service jobs in county, county unemployment rate, race/ethnicity, student transfer status, parent education, family structure, family income, 9th-grade GPA, math test score, school percent free lunch, school percent minority, magnet school status, vocational school status, urbanicity, district per-pupil expenditures, state fixed-effects, and whether the school provided a course catalog.

^*p < .05;

^**p < .01;

^***p < .001 (two-tailed tests).

Model 1 estimates male and female students’ blue-collar course-taking and shows that, on average and net of base-year sociodemographic, academic, school, and local labor market controls, male students in blue-collar communities take about 24 percent more blue-collar courses in high school than do their male peers in non-blue-collar communities ( p < .01). In Model 2, we control for course offerings and observe that the number of blue-collar courses offered by schools is positively and statistically significantly associated with the number of blue-collar courses male students take. Adjusting for differences in course offerings across schools reduces the relationship between the local labor market and blue-collar course-taking by about one-third. This suggests that young men in blue-collar communities take more blue-collar courses partially because they attend schools that offer greater numbers of blue-collar courses. In contrast to the findings for male students, Models 1 and 2 for female students’ blue-collar course-taking show that female students in blue-collar communities do not take significantly greater numbers of blue-collar courses than their female peers in non-blue-collar communities, on average and net of background controls.⁹

Table 1 also presents the average marginal effects from logistic regressions estimating the likelihood of taking an advanced math course (above Algebra 2) by the end of high school, conditioning on background and local labor market controls. Model 1 predicts male students’ advanced math course-taking and shows that, on average, the probability that male students in blue-collar communities take advanced math by the end of high school is about 9 percentage points lower than that of their male peers in non-blue-collar communities. We control for school course offerings in Model 2, which slightly reduces the coefficient for males in blue-collar communities but does not explain their disadvantage. Turning now to female students’ advanced math course-taking, Model 1 shows that female students in blue-collar communities are also about 9 percentage points less likely to take an advanced math course compared to their female peers in non-blue-collar communities. Adjusting for differences in school course offerings in Model 2—specifically the academic rigor of schools’ math curriculum—reduces the magnitude of the gap in advanced math course-taking between female students in blue-collar and non-blue-collar communities by about 30 percent and renders it non-significant.

These findings indicate that blue-collar course offerings facilitate male high school students’ blue-collar course-taking in blue-collar communities, but they also suggest that male students tend to take coursework related to the skills and lower educational requirements of their local labor markets even independent of coursework opportunities at their schools. However, our results suggest that female students do not enroll in courses related to the male-dominated blue-collar jobs in their communities. Of concern is that young women in blue-collar communities are less likely to take an advanced math course by the end of high school—the strongest predictor of college-going and completion (Adelman 2006)—partially because their schools offer a less academically rigorous curriculum. These course-taking patterns suggest that young men and young women in blue-collar communities may be less prepared for college, but young men in blue-collar communities may be more prepared than their peers for blue-collar jobs after high school.

Two Years Out of High School

Turning to our second research question, we use multinomial logistic regressions to estimate young men’s and women’s likelihoods of attending a four-year university, being employed in a blue-collar job, being employed in a non-blue-collar job, attending a two-year college, and not working two years after expected high school graduation.

In Model 1 of Table 2, we observe that young men who attended high school in blue-collar communities have about a 9 percentage-point higher probability of being employed in a blue-collar job, and a 7 percentage-point lower probability of attending a four-year college, relative to men from non-blue-collar communities, net of base-year sociodemographic, academic, school, and other local labor market controls. Adjusting for differences in course offerings attenuates the gap in blue-collar employment between young men from blue-collar and non-blue-collar communities by nearly 30 percent, and it reduces the gap in four-year college attendance by 40 percent and renders it non-significant. In Model 3, even after controlling for course offerings and course-taking, young men who attended high school in blue-collar communities remain more likely to be employed in blue-collar jobs relative to their peers from non-blue-collar communities.

Table 2.

Average Marginal Effects from Multinomial Logistic Regressions Predicting Young Men’s Post-High School Destinations

	Model 1: Background					Model 2: Course Offerings					Model 3: Course-Taking
	Four- Year College	Not Work- ing	Other Job	BC Job	Two- Year College	Four- Year College	Not Work- ing	Other Job	BC Job	Two- Year College	Four- Year College	Not Work- ing	Other Job	BC Job	Two- Year College
Percent Blue-Collar Workers in County (Ref.: 1st Quartile)
2nd Quartile	−.006 (.024)	.017 (.010)	−.017 (.019)	.013 (.018)	−.007 (.020)	−.006 (.022)	.018 (.010)	−.018 (.019)	.016 (.019)	−.009 (.020)	−.002 (.022)	.018 (.010)	−.020 (.019)	.014 (.019)	−.010 (.020)
3rd Quartile	−.054* (.024)	.013 (.011)	−.006 (.022)	.067** (.021)	−.019 (.021)	−.042 (.025)	.012 (.011)	−.008 (.023)	.060** (.021)	−.022 (.021)	−.027 (.024)	.013 (.010)	−.014 (.023)	.056** (.021)	−.027 (.022)
4th Quartile	−.072** (.028)	.016 (.012)	−.026 (.023)	.089*** (.022)	−.006 (.024)	−.044 (.029)	.019 (.013)	−.029 (.025)	.064** (.021)	−.010 (.025)	−.017 (.029)	.019 (.013)	−.036 (.025)	.050* (.021)	−.016 (.025)
Course-Taking
Number of BC courses taken (logged)											−.065*** (.011)	−.008 (.006)	.018 (.010)	.057*** (.009)	−.001 (.010)
Took advanced math course											.132*** (.017)	.001 (.010)	−.043* (.018)	−.051*** (.015)	−.038* (.016)
Course Offerings
Number of BC courses offered (logged)						−.018 (.010)	.001 (.005)	.001 (.009)	.007 (.008)	.009 (.008)	−.009 (.010)	.002 (.005)	−.001 (.009)	−.002 (.008)	.011 (.008)
Number of AP/IB offered (logged)						.026* (.011)	.002 (.004)	−.006 (.010)	−.021** (.008)	−.001 (.009)	.015 (.010)	.002 (.004)	−.001 (.010)	−.017* (.008)	.002 (.009)
Number of advanced math offered						.009* (.004)	−.004 (.002)	−.004 (.004)	−.003 (.004)	.001 (.004)	.005 (.004)	−.004 (.002)	−.002 (.004)	−.001 (.004)	.003 (.004)
9th-grade math level	.034*** (.006)	−.008** (.003)	−.009 (.006)	−.008 (.005)	−.009 (.005)	.034*** (.006)	−.008** (.003)	−.008 (.006)	−.008 (.005)	−.009 (.005)	.026*** (.005)	−.008** (.003)	−.006 (.006)	−.004 (.005)	−.008 (.005)

Note: Standard errors are in parentheses. Observations = 4,660. All models control for total number of courses offered by the school, inverse Mills ratio for selection into sample, GED status, percent low-wage service jobs in county, county unemployment rate, race/ethnicity, student transfer status, parent education, family structure, family income, 9th-grade GPA, math test score, school percent free lunch, school percent minority, magnet school status, vocational school status, urbanicity, district per-pupil expenditures, state fixed-effects, and whether the school provided a course catalog.

^*p < .05;

^**p < .01;

^***p < .001 (two-tailed tests).

Table 3 presents these models for young women, conditioning on background and local labor market controls. Model 1 shows that, relative to young women who attended high school in non-blue-collar communities, female students from blue-collar communities are about 11 percentage points less likely to attend a four-year college. Accounting for differences in course offerings in Model 2—AP/IB course offerings in particular—reduces the gap in four-year college enrollment by about 35 percent. In Model 3, we introduce course-taking controls and see that young women who took an advanced math course in high school have a 15 percentage-point higher probability of attending a four-year college. Not surprisingly, accounting for the lower advanced math course-taking of female students in blue-collar communities further reduces their disadvantage. Although course offerings and course-taking together account for about 50 percent of the disparity in four-year college enrollment we observed in Model 1, young women in blue-collar communities remain less likely to attend a four-year college relative to their peers in non-blue-collar communities. With respect to gender inequality within blue-collar communities, ancillary analyses not shown here indicate that the gender gaps in blue-collar communities for not working and being employed in non-blue-collar jobs are greater than they are in non-blue-collar communities.

Table 3.

Average Marginal Effects from Multinomial Logistic Regressions Predicting Young Women’s Post-High School Destinations

	Model 1: Background					Model 2: Course Offerings					Model 3: Course-Taking
	Four- Year College	Not Work- ing	Other Job	BC Job	Two- Year College	Four- Year College	Not Work- ing	Other Job	BC Job	Two- Year College	Four- Year College	Not Work- ing	Other Job	BC Job	Two- Year College
Percent Blue-Collar Workers in County (Ref.: 1st Quartile)
2nd Quartile	−.060** (.021)	.004 (.011)	.019 (.022)	.005 (.006)	.032 (.019)	−.061** (.021)	.005 (.012)	.017 (.022)	.006 (.007)	.032 (.019)	−.049* (.020)	.003 (.012)	.012 (.021)	.006 (.007)	.028 (.019)
3rd Quartile	−.050* (.024)	−.001 (.011)	.038 (.024)	−.000 (.006)	.013 (.020)	−.041 (.024)	−.002 (.012)	.034 (.024)	−.001 (.006)	.010 (.020)	−.031 (.024)	−.003 (.012)	.029 (.025)	−.001 (.006)	.006 (.020)
4th Quartile	−.107*** (.027)	.018 (.015)	.031 (.027)	.012 (.009)	.045 (.024)	−.073** (.027)	.007 (.014)	.027 (.028)	.005 (.008)	.034 (.024)	−.055* (.026)	.006 (.015)	.019 (.028)	.004 (.008)	.026 (.023)
Course-Taking
Number of BC courses taken (logged)											−.059** (.020)	.014 (.009)	.019 (.021)	.009 (.005)	.018 (.017)
Took advanced math course											.150*** (.014)	−.013 (.009)	−.076*** (.017)	−.001 (.005)	−.061*** (.015)
Course Offerings
Number of BC courses offered (logged)						−.016 (.009)	.001 (.004)	.012 (.009)	.004 (.002)	−.002 (.008)	−.015 (.009)	.001 (.004)	.012 (.009)	.004 (.002)	−.002 (.008)
Number of AP/IB offered (logged)						.043*** (.010)	−.013** (.005)	−.013 (.011)	−.004 (.003)	−.014 (.010)	.031** (.010)	−.011* (.005)	−.007 (.011)	−.003 (.003)	−.009 (.010)
Number of advanced math offered						.002 (.004)	.001 (.002)	.003 (.004)	.001 (.001)	−.006 (.004)	−.001 (.004)	.001 (.002)	.004 (.004)	.001 (.001)	−.005 (.004)
9th-grade math level	.021*** (.006)	−.007** (.003)	−.012* (.006)	−.002 (.002)	.001 (.006)	.022*** (.006)	−.008** (.003)	−.010 (.006)	−.002 (.002)	−.001 (.006)	.015** (.005)	−.008** (.003)	−.007 (.006)	−.002 (.002)	.002 (.006)

Note: Standard errors are in parentheses. Observations = 5,190. All models control for total number of courses offered by the school, inverse Mills ratio for selection into sample, GED status, percent low-wage service jobs in county, county unemployment rate, race/ethnicity, student transfer status, parent education, family structure, family income, 9th-grade GPA, math test score, school percent free lunch, school percent minority, magnet school status, vocational school status, urbanicity, district per-pupil expenditures, state fixed-effects, and whether the school provided a course catalog.

^*p < .05;

^**p < .01;

^***p < .001 (two-tailed tests).

Our results indicate that high school course offerings and course-taking play a major role in young men’s and women’s lower probabilities of attending a four-year college relative to their peers in non-blue-collar communities. Young men in these communities enroll in greater numbers of blue-collar courses and are more likely to enter blue-collar jobs, and this is related to high school course offerings and their own course-taking. This suggests tight linkages among high school training, the local labor market, and working in blue-collar jobs for men but not for women. In summary, post-high school destinations of young men and women who attended high school in blue-collar communities are different from each other and from their peers in non-blue-collar communities, and differences in their high school training appear to play a role in producing these disparities.

Eight Years Out of High School

We now turn to investigating students’ labor force outcomes eight years post-high school, after they have had the opportunity to complete postsecondary education. Figure 3 presents average predicted probabilities, or the differences in the AME, for men’s and women’s occupations in 2012 for people who attended high school in blue-collar and non-blue-collar communities.¹⁰ With control variables at their observed values, the expected share of young men entering blue-collar occupations would be .28 if all young men had attended high school in blue-collar communities, compared to .19 if all young men had attended high school in non-blue-collar communities. Women in blue-collar communities are statistically significantly more likely to not be working (.23 versus .18), less likely to be employed in white-collar jobs (.33 versus .39), and slightly more likely to be employed in blue-collar jobs (.05 versus .03) relative to women in non-blue-collar communities. Analyses not shown indicate that these differences for women become non-significant once accounting for high school course offerings and course-taking, indicating a relationship between high school training and the labor force that persists into women’s mid-20s. With regard to gender gaps, we observe that women’s over-representation in low-wage service jobs and among those not working is statistically significantly greater in blue-collar communities compared to non-blue-collar communities.

Figure 3.

Average Predicted Probabilities for Occupation Eight Years Following Expected High School Graduation

Note: Calculated from models that include the inverse Mills ratio for selection into sample, postsecondary enrollment status, GED status, percent low-wage service jobs in county, county unemployment rate, race/ethnicity, student transfer status, parent education, family structure, family income, 9th-grade GPA, 9th-grade math level, math test score, school percent free lunch, school percent minority, magnet school status, vocational school status, urbanicity, district per-pupil expenditures, and state fixed-effects. Due to rounding, predicted probabilities may not add to 1.

^aIndicates gender gap in 4th quartile is statistically significantly different from gender gap in 1st quartile (p < .01).

^*p < .05; ^**p < .01; ^***p < .001 (statistically significant within-gender difference; two-tailed tests).

Given differences in the postsecondary and occupational outcomes of men and women between blue-collar and non-blue-collar communities, we might expect to see gender disparities in wages. Table 4 shows the estimated 2011 average hourly wages for men and women who attended high school in blue-collar (4th quartile) and non-blue-collar (1st quartile) communities, when they are about 25 years old. We estimate predicted hourly wages from gender-stratified OLS regressions predicting logged hourly wages for each labor market quartile and controlling for base-year sociodemographic, academic, and other controls (full models are shown in Table S5 in the online supplement). For presentation purposes, we show predicted wages for men and women within the 1st and 4th quartiles only.

Table 4.

Estimated Hourly Wages and Gender Wage Gap from OLS Regressions

	Men	Women	Gender Wage Gap
Percent Blue-Collar Workers in County
1st Quartile	15.27	14.93	98%
4th Quartile	14.49	12.27*a	85%
Observations	3,900	4,620

Note: Calculated from models that include inverse Mills ratio for selection into sample, GED status, percent low-wage service jobs in county, county unemployment rate, race/ethnicity, student transfer status, parent education, family structure, family income, 9th-grade GPA, 9th-grade math level, math test score, school percent free lunch, school percent minority, magnet school status, vocational school status, urbanicity, district per-pupil expenditures, and state fixed-effects.

^aIndicates gender gap in 4th quartile is statistically significantly different from gender gap in 1st quartile (p < .01).

^*p < .05;

^**p < .01;

^***p < .001 (statistically significant within-gender difference; two-tailed tests).

Looking at within-gender differences between these local labor markets, we observe that, on average and net of controls, men who attended high schools in blue-collar communities are expected to earn less than their male counterparts in non-blue-collar communities ($14.49 compared to $15.27), but this difference is not statistically significant. In contrast, women from blue-collar communities are estimated to earn more than $2.50 less per hour than young women from non-blue-collar communities ($12.27 compared to $14.93 per hour). Ancillary analyses indicate that accounting for the less rigorous academic course offerings of schools in blue-collar communities reduces the wage penalty that women in blue-collar communities experience by 50 percent and renders the difference non-significant. Accounting for high school math course-taking, highest degree earned, and occupation drives the wage gap nearly to zero.

Most striking is that the gender gap in wages between men and women who attended high school in blue-collar communities is over two dollars ($14.49 versus $12.27), compared to a 30 cents difference ($15.27 versus $14.93) between men and women from non-blue-collar communities. Ancillary analyses of the interacted pooled model indicate that the gender wage-gap within blue-collar communities is statistically significantly larger than in non-blue-collar communities; this gap is reduced by about one-third after controlling for the early adulthood occupation differences we observed earlier. Taken together, these results suggest that gender inequality in the labor force is heightened among young men and women who attended high school in blue-collar communities.

In summary, our results suggest a relationship between high school training and local blue-collar work that structures gendered spatial inequalities in the transition to college and the labor force. Results indicate that women in blue-collar communities are less likely to attend a four-year college than women in non-blue-collar communities. This disparity foreshadows their lower representation in white-collar jobs and higher probability of not working or working in blue-collar jobs in early adulthood. We also found that women’s over-representation among young adults not working and in low-wage service jobs—a persistent impediment to gender equality—is greatest among students attending high school in blue-collar communities. Finally, the post-secondary matriculation and labor force penalties young women in blue-collar communities face are partially explained by their schools’ weaker academic curricula and their lower odds of taking advanced math.

Similar to their female peers, men from blue-collar communities are less likely to enroll in a four-year college than their male counterparts attending high school in non-blue-collar communities, largely due to their schools’ less rigorous academic curricula and their higher enrollment rates in blue-collar coursework. At the same time, they are more likely to be employed in blue-collar jobs two years out of high school than both their female peers and their male counterparts in non-blue-collar communities. Eight years out of high school, they had about a 10 percentage-point higher probability of being in blue-collar jobs compared to men in non-blue-collar communities. These relationships operated partially through differences in school course offerings and young men’s greater investments in blue-collar coursework. Finally, unlike their female peers, young men in blue-collar communities did not appear to suffer severe wage penalties relative to their counterparts in non-blue-collar communities, resulting in a greater gender pay-gap among men and women attending high school in blue-collar communities.

DISCUSSION

The ways our nation’s schools adapt to a rapidly changing economy have important implications for social stratification. Indeed, the educational gradient in life course outcomes is increasing while jobs and occupational demands are rapidly changing. Some scholars and policymakers continue to argue that the best path to economic success is for individuals, schools, and communities to invest in higher education and develop the cognitive skills required by jobs in the knowledge-based economy. In contrast to this camp and the wave of reforms aimed at curricular upgrading since A Nation at Risk (National Commission on Excellence in Education 1983), many states are reemphasizing career and technical training. Moreover, a cadre of academics argues against the “college-for-all” prescription, pointing to the rising cost of tuition and evidence of growing mid-skill, sub-baccalaureate jobs. Both camps tend to ignore the local nature of schools and the uneven distribution of sub-baccalaureate jobs across local economies (Moretti 2013). The debate has also been gender-neutral, even though well-paying sub-baccalaureate work lies primarily in male-dominated, blue-collar occupations.

This study adds a gendered and spatial perspective to current debates about old questions surrounding the type of high school training that meritocratically prepares all students for the labor market. We uncovered the gender-divergent educational and work consequences of linking high school training to local sub-baccalaureate jobs in the context of today’s blue-collar communities—places within the United States where these highly gender-segregated jobs still exist and that inspired a classical, primarily ethnographic literature on the local dynamics of social stratification. We found that high school training is stratified along a spatial axis of inequality, and the link between schools and communities in places with prevalent blue-collar jobs disproportionately penalizes women.

Our study’s results are consistent with research linking local labor markets with lower demands for female labor (Cotter et al. 1998) and heavier concentrations of manufacturing jobs to greater gender inequality (Gauchat et al. 2012; McCall 2001). By conceptualizing schools as embedded within communities, this study elucidates the role of gendered school curricula within these local labor markets in the production of gender inequality. This pattern was evident in industrial cities in the early twentieth century, when women’s high school training options were confined to lower-status female-typical work, while men were trained for blue-collar jobs (Rury 1991). The results reveal a similar story almost a century later. Schools in communities with existing blue-collar jobs emphasize blue-collar training for young men but deemphasize academic course-work that encourages four-year college completion. Our findings highlight the labor market costs of a less academically rigorous curriculum for women. Where Chetty and colleagues (2016) show geographic variation in income mobility for men and women, our findings suggest a potential mechanism in school curricula and its link to the local labor market that may contribute to gendered economic opportunities across place.

Our results have important implications for the large body of scholarship on school-to-work links. Research shows that the strength of school-to-work links varies across countries (e.g., Kerckhoff 2000; Shavit and Muller 1998). School-to-work links are more weakly coupled in the United States relative to most Western European countries, but our results suggest geographic variation in the strength of school-to-work connections within the United States. We find that this variation is partly a function of the coupling between high school coursework and opportunities for local jobs. We found the strongest school-to-blue-collar work link for young men attending high school in blue-collar communities. These young men also earned wages that were slightly lower but not statistically different from those earned by young men from non-blue-collar communities who were more likely to attend a four-year college. This finding is perhaps not surprising, given current four-year degree persistence rates and research suggesting labor market advantages of career-specific high school vocational training (Bishop and Mane 2004; Shavit and Muller 1998) that aligns with students’ post-high school occupation (Grubb 1999).

These results may provide some support for President Obama’s touted manufacturing innovation hubs, based on the German model, and policymakers’ promotion of blue-collar jobs as pragmatic alternatives to the kinds of jobs often obtained after attempting to complete a four-year degree. Rather than shouldering the substantial cost of higher education, these young men were in occupations with average earnings that can rival and surpass many occupations filled by workers holding baccalaureate degrees. Our findings for men may support current discourse challenging the college-for-all ethos, suggesting a local setup in which young men—whose educational attainment has inspired research and alarmed the public for over a decade—do quite well without a four-year degree. However, our results are specific to men attending high school in communities with greater opportunities for blue-collar work. The results observed for blue-collar course-taking in blue-collar communities cannot be generalized to other occupation-specific coursework or different local occupational structures. It is also important to temper strong conclusions drawn from early adulthood labor force patterns (about age 25); at this stage, bachelor’s degree holders have just begun working and may have higher earnings as they age. Furthermore, research finds that the early adulthood employment advantages of taking more high school vocational courses and fewer general courses are followed in later adulthood by a reduced ability to respond to changes in the economy and higher risk of unemployment (Hanushek, Woessmann, and Zhang 2011). Given that these men were less likely to attend a four-year college, they may have few options to obtain other well-paying jobs, weather labor market vicissitudes, or deal with physical health impairments associated with manual labor and aging.

Whereas high school training in blue-collar communities may foster—or at least not inhibit—a school-to-work link for men in the short-term, we found a weaker link for women. We found no difference in the likelihood of blue-collar employment two years out of high school between women attending high school in blue-collar and non-blue-collar communities; this difference was only two percentage points (5 percent compared to 3 percent) eight years out of high school. Interestingly, whereas previous research found a positive link between trade and technical course-taking and women’s employment in skilled manual jobs in the 1980s (for a cohort at the end of the Baby Boom) (Arum and Shavit 1995), the relationship between blue-collar course-taking and blue-collar employment two years out of high school for this Millennial cohort of women is weak and non-significant. Although beyond the scope of the current study, previous work suggests that gendered informal networks and hiring discrimination may limit the remaining blue-collar job options for women, and studies propose that affirmative action policies may provide a partial solution (Bergmann 2011; Padavic and Reskin 2002). In any case, our study points to a gendered spatial dimension of school-to-work transitions and suggests that gender inequalities in the transition to decent-paying work, especially among people who do not pursue a four-year degree, may be heightened in communities with heavier concentrations of male-dominated, sub-baccalaureate jobs.

It is worth noting that previous research suggests differential labor market returns to blue-collar training and blue-collar jobs at the intersection of gender and race/ethnicity, with white men reaping the greatest economic rewards (Ainsworth and Roscigno 2005; McCall 2001; Royster 2003). Small sample sizes precluded a full analysis of these issues, although our exploratory analyses suggest that the disparity in blue-collar employment between black and white men was greatest among young men attending high school in blue-collar communities. Future research should prioritize an investigation of racial and social class inequalities within blue-collar communities.

Overall, our study suggests a gendered, spatial layer of complexity to the college-for-all debate that has been largely ignored by both advocates and critics. High school business vocational coursework once provided women with a sub-baccalaureate avenue to decent-paying clerical-related jobs (Arum and Shavit 1995). However, the mass automation of non-routine manual tasks (Autor and Dorn 2013) has led to an even greater polarization of higher- and lower-skill jobs for women than for men (Black and Spitz-Oener 2010), with most well-paying sub-baccalaureate jobs today concentrated within male-dominated, blue-collar fields. Furthermore, the highest-paying mid-skill jobs for women offer lower wages than the highest-paying jobs for men with a high school degree (Carnevale et al. 2011). High school training in blue-collar communities appears to facilitate a successful school-to-blue-collar work transition for men but appears to harm women’s labor market prospects—not because these schools offer greater numbers of blue-collar courses ipso facto, but because they do so at the expense of offering academically rigorous courses that prepare women for bachelor’s degrees and professional work. Elevating schools’ academic climates (DiPrete and Buchmann 2013) and encouraging four-year degree completion may be the most effective way to promote long-term labor market success for both men and women.

This study is limited in its ability to draw causal connections between local labor markets, school curricula, and gendered education and labor market outcomes. It is possible, for example, that we did not fully capture differences between families who select into blue-collar communities and those who do not. If so, these differences may be related to both the local labor market and students’ education and labor market outcomes. A number of factors give us confidence that these selection issues do not drive our results. Ancillary analyses indicate that our findings are robust to controls for parents’ employment in blue-collar occupations, a potential source of selection into these communities. Additionally, our results identify a likely institutional mechanism—school course offerings and course-taking—that contributes to gender disparities in post-high school destinations across communities. Finally, we found gender differences in labor market outcomes within blue-collar communities among otherwise similar families that differed only on whether their son or daughter was in the sample. Future research may gain leverage by examining the gendered effects of changes in job opportunities in local economies over time due to economic shocks or other sources of exogenous variation. Researchers should also exploit new state policies that relax academic high school graduation requirements and emphasize career and technical education coursework to better estimate the causal impacts of school curricula on gender inequality in education and the labor force.

Conclusion

Recent research suggests a paradox in which “our global economy is becoming increasingly local” (Moretti 2012:248)—that is, instead of growing more similar, local worker profiles and economies within the United States have grown further apart in the global economy. Our study suggests that high school training is stratified across these increasingly polarized local economies, even as major reforms attempt to standardize curricula across the nation and produce a globally competitive workforce. In the case of high school training in blue-collar communities, we find that the male-dominated nature of blue-collar work adds a layer of complexity to college-for-all debates. Results suggest that high school training in blue-collar communities may not penalize young men in the short-term, as a large share obtain employment in decent-paying blue-collar jobs. At the same time, our study’s results caution that school investments in blue-collar training that are not complemented by academically rigorous coursework may leave women behind in the global and local economy. Taken together, these findings call for a concern about gender equity as we continue to debate the educational pathways that lead to well-paying, stable jobs in the New Economy. Given geographic differences in social mobility (Chetty et al. 2014) and growing inequality between places (Domina 2006; Lichter, Parisi, and Taquino 2012; Moretti 2012), it is important that scholars continue to study how schools and communities interact to produce stratification by gender and other axes of inequality.

Biographies

April Sutton is a Frank H.T. Rhodes Postdoctoral Fellow at Cornell University. Her research fields of interest are education, social stratification and mobility, and life course transitions. The objective of her work across these areas is to investigate the circumstances under which schools and communities widen or narrow gender and racial/ethnic inequalities during critical life transitions, from adolescence through early adulthood. Her current work examines how local labor markets and schools interact to shape stratification in postsecondary and labor force outcomes.

Amanda Bosky is a graduate student in sociology and a Population Research Center trainee at the University of Texas at Austin, and she holds a JD from Loyola University Chicago School of Law. Her primary research interests are education and stratification in work and occupations, with a particular focus on the educational predictors of labor market inequalities across the life course. Her current work investigates how individuals’ skills in adolescence are related to their labor force attachment at midlife.

Chandra Muller is Alma Cowden Madden Centennial Professor of Liberal Arts in the sociology department and faculty affiliate in the Population Research Center at the University of Texas at Austin. Her research focuses on the effects of skills and education on stratification in the transition to adulthood and through midlife. She is currently leading a study that is following up the High School and Beyond sophomore and senior cohorts in midlife to track the long-term effects of high school during a period of economic transition.

APPENDIX

Table A1.

Weighted Means and Proportions for Key Variables, by Sex and Blue-Collar Quartile

	Men				Women
	1st Quartile	2nd Quartile	3rd Quartile	4th Quartile	1st Quartile	2nd Quartile	3rd Quartile	4th Quartile
Student Course-Taking (N = 11,610)
Math above Algebra 2 by 12th grade	.51	.45	.40	.38	.58	.48	.46	.44
Number of blue-collar courses (SD)	1.12 (1.85)	1.41 (2.18)	1.50 (2.24)	2.12 (2.89)	.21 (.63)	.31 (.82)	.20 (.70)	.35 (1.00)
Logged (SD)	.50 (.63)	.60 (.71)	.62 (.70)	.82 (.81)	.12 (.31)	.17 (.39)	.11 (.31)	.18 (.43)
Post-High School Destination (N = 10,080)
Attending four-year college (ref.)	.44	.38	.38	.33	.57	.44	.45	.37
Not working	.04	.05	.04	.04	.04	.05	.05	.08
Non-blue-collar job	.23	.22	.23	.19	.20	.25	.27	.28
Blue-collar job	.10	.13	.18	.27	.01	.02	.01	.03
Attending two-year college	.19	.21	.17	.17	.18	.25	.22	.24
Occupation Eight Years after Expected HS Graduation (N = 9,190)
White-collar (ref.)	.32	.31	.28	.27	.41	.32	.34	.31
Low-wage service	.11	.09	.08	.08	.14	.17	.16	.18
Blue-collar	.17	.22	.26	.34	.03	.02	.02	.06
Not working	.15	.11	.14	.11	.16	.20	.22	.24
Other non-professional	.25	.27	.24	.20	.26	.29	.26	.21
Hourly Wages Eight Years after Expected HS Graduation (N = 8,510)
Wages (SD)	17.62 (11.81)	17.38 (10.20)	17.58 (10.91)	16.51 (10.18)	16.54 (9.12)	15.35 (9.04)	14.29 (8.01)	13.75 (7.95)
Logged (SD)	2.72 (.53)	2.73 (.50)	2.72 (.53)	2.67 (.51)	2.68 (.48)	2.60 (.51)	2.54 (.48)	2.50 (.51)
Parent Education (N = 12,770)
No postsecondary degree (ref.)	.45	.51	.54	.60	.45	.53	.55	.62
Two-year degree	.11	.12	.11	.15	.10	.11	.11	.11
Four-year degree or above	.44	.37	.35	.25	.45	.36	.34	.27

Table A2.

Weighted Descriptive Statistics for Additional Variables (N = 12,770)

	Mean/Proportion	SD
Local Labor Market Measures
Percent blue-collar workers in county
1st Quartile (ref.)	.25
2nd Quartile	.26
3rd Quartile	.26
4th Quartile	.23
Percent low-wage service workers in county	13.19	2.46
Percent unemployed in county	6.10	2.36
School Measures
Course offerings
Number of BC courses offered (logged)	2.14	.98
Number of AP/IB courses offered (logged)	2.21	.96
Number of advanced math courses offered	3.48	2.14
Number of courses offered by the school	195.64	113.37
School provided course catalog	.94
School percent receiving free/reduced lunch	22.79	18.64
School percent minority	35.50	30.88
Magnet school	.13
Vocational school	.09
Urbanicity
Urban (ref.)	.38
Suburban	.51
Rural	.21
District per-pupil expenditures	8619.17	2377.66
Academic Background
9th-grade GPA	2.50	.92
9th-grade math level	3.88	1.40
10th-grade math achievement test score	36.85	11.92
Sociodemographic Background
Race/Ethnicity
White (ref.)	.59
Black	.15
Asian	.04
Hispanic	.17
Other	.05
Lives with both biological parents	.56
Family income	8.85	2.37
Parent education
No postsecondary degree (ref.)	.53
Two-year degree	.12
Four-year degree or above	.35
Additional Covariates
Enrolled in postsecondary courses	.25
Transfer student	.08
GED status	.04