Occupational Career Change and Gender Wage Inequality

Abstract

This article examines the relationship between the changing occupational careers of female wage earners and gender wage inequality. Using Current Population Survey-Merged Outgoing Rotation Group data, it assesses the effect on the gender wage gap of changes in the composition and price both of care-providing occupations that are culturally associated with female labor and of managerial and professional occupations that are not part of the care economy, over the period 1979 to 2015. It finds that the rapid entry of female workers into high-wage managerial occupations, and their exit from low-wage private household work, contributed to gender wage convergence. However, the wage-equalizing effects of occupational shifts and related behavioral changes diminish over time, and wage convergence ceases after 2007. It also finds that female workers continue to be disadvantaged by wage dispersion and that most of the remaining gender wage gap arises within occupations. The concluding sections discuss the findings and their implications for closing the wage gap.

Keywords

earnings career paths gender

The gender wage ratio in the United States fluctuated around 0.60 throughout the early postwar decades, rose substantially during the 1980s, and then increased more slowly in the 1990s and early 2000s. In the lifetime human capital model that underpins conventional analyses of earnings differences, this gender disparity in the market returns to work is explained by productivity differences presumed to arise from women's shorter and more intermittent work careers and limited investment in job- or occupation-specific skills (e.g., Mincer & Polachek, 1974; O'Neill, 2003; Polachek, 2012). Consistent with the logic of this lifetime human capital model, its proponents attribute the narrowing of the gender pay gap in recent decades to the increasing duration and continuity of women's work experience and their greater willingness to invest in specific skills.

Human capital theory has long been criticized for attributing earnings differences to the characteristics and decisions of individuals while neglecting the productivity-determining characteristics of jobs and the allocative processes that match individuals with jobs (e.g., Granovetter, 1981). Viewed from a human capital perspective, shifts in the gender wage ratio are explained by corresponding changes in the relative experience and qualifications of the female workforce. The wage-related characteristics of the jobs that female workers enter, and the forces that affect job rewards independently of the qualifications of jobholders, rarely enter into analyses of group inequality (Leicht, 2008; Morris & Western, 1999). The present study addresses this deficiency by examining how changes in women's occupational career paths, and in the relative earnings of the occupations to which those career paths lead, impact gender wage inequality. Even at the most detailed level, occupational categories do not fully capture the job differences that affect wages. Nonetheless, most individuals make their job choices within the context of occupational careers, and the relationship between women's changing occupational careers and their relative earnings illustrates how changes in both job allocation processes and job rewards affect the gender wage gap.

In pursuing this inquiry, the study builds upon previous analyses of the 1980s wage convergence and its subsequent slowdown (Blau & Kahn, 1997, 2006; Cha & Weeden, 2014; Datta Gupta, Oaxaca, & Smith, 2006). These analysts use the formal wage decomposition developed by Juhn, Murphy, and Pierce (1991; hereafter JMP) to apportion change in the gender wage gap between compositional shifts, the changing mean differences in the individual and job characteristics of male and female workers, and price shifts, the changing wage returns to those characteristics. This study builds upon this research by documenting how changes in the composition and price of different occupational services help to explain the 1980s wage convergence and its subsequent slowing. More precisely, it analyzes the changes in the gender wage gap over the period 1979 to 2015 among all wage earners, among those employed in occupations providing nurturant and reproductive care, and among those employed in occupations that are not part of the care economy. By apportioning wage change into compositional and price components, the JMP decomposition distinguishes between the effects of changes in the gender composition of occupations and the effects of changes in the market returns to those occupations. In other words, the compositional component captures changes in job allocation that impact the wage gap, while the price component measures changes in job rewards that benefit or penalize both male and female workers, depending upon their placement in particular occupations.

The following section explains how changes in the gender composition and relative wages of different occupational groups are expected to affect the wage gap. Subsequent sections describe the data and methods and present the results of the analysis. In brief, the results indicate that women's changing occupational careers, especially their increased employment in managerial jobs that are not traditionally associated with female labor, account for much of the initial reduction in earnings inequality. The results also indicate that occupational career shifts are no longer narrowing the gender wage gap, that female workers continue to be disadvantaged by wage dispersion, and that most of the remaining wage gap arises within occupations. The concluding sections discuss these findings and argue that improving women's occupational status is still key to closing the pay gap.

Changes in the Gender Composition and Relative Wages of Occupational Services

The occupational shifts most clearly linked to the gender division of labor involve the expansion and feminization of market-based care. Care work can be broadly defined as services that maintain life and reproduce the next generation—that is, the services traditionally provided by women within families. Within this broad conception, feminist scholars often distinguish between nurturant and nonnurturant care and describe how each increasingly takes the form of market-based activities (e.g., Duffy, 2005, 2007, 2011; Dwyer, 2013). Nurturant care is inherently relational. It is increasingly provided through health care, teaching, and social work occupations and encompasses face-to-face services that maintain the physical and mental health, and develop the physical, cognitive, and emotional capabilities, of the care recipient (England, Budig, & Folbre, 2002). Nonnurturant, or reproductive, care includes cleaning, cooking, and laundry services, the dirty work that is more physical in nature and can be provided without a face-to-face interaction with the service recipient (Glenn, 1992). As market-based activities, both forms of care work are culturally associated with female labor and expanded over the course of the past century as women entered the paid workforce and as families purchased more care services.

In contrast to the long-standing overrepresentation of women in occupations providing care, the entry of a large and rapidly growing proportion of the female labor force into managerial and professional occupations that are not part of the care economy is a relatively recent phenomenon. As described by Goldin (1990, 2006), this occupational shift was the result of a “quiet revolution” in the career horizons, identity, and decision-making of the cohorts of young women who enter the labor market beginning in the 1970s. Goldin (2006) acknowledges that the changes in female labor supply behavior that accompanied this quiet revolution are broadly consistent with those predicted by human capital theory. But she also emphasizes how the expansion of women's career time horizons and the greater importance of finding individuality through work transformed jobs into potential careers while altering the occupational paths along which they pursued those careers. Occupational career choices shifted “from those considered traditional ones for women, such as teacher, nurse, librarian, and social worker, to a more varied group of professions” (Goldin, 2006, p. 13). In essence, the relatively recent and rapid entry of young women into a broader range of managerial and professional jobs was the result of a cultural transformation in women's work identities and career expectations that altered their labor supply behavior.

These compositional shifts affect women's relative earnings through both occupational expansion and feminization. To the extent that an occupation pays above or below the workforce average, and that women account for substantially more or less than half of that occupation's workforce, an increase or decrease in the occupation's share of total employment will impact gender wage inequality. Similarly, to the extent that an occupation pays above or below the workforce average, an increase or decrease in the female share of the occupational workforce will affect the gender pay gap. In other words, the expansion and feminization of particular occupations will impact the wage gap depending not only on the magnitude of those shifts but also on the extent to which the occupational wage differs from the workforce average. Conversely, an increase or decrease in the relative price, or market valuation, of particular occupational services will affect the gender pay gap depending not only on the occupation's employment share but also on the proportion of its services that are provided by female workers. As that proportion approaches 50% (i.e., as the ratio of female-to-male service providers approaches 1.0), the effects of a price change on male and female earnings cancel out. Beginning with nurturant care, the remainder of this section describes how changes in the composition and price of the different occupational services are expected to affect gender wage inequality.

Nurturant Care Occupations

The market provision of nurturant care increased over the course of the past century as a growing proportion of the adult female population entered the workforce, and as demand for expert care fueled the development of specialized knowledge and the professionalization of health care, education, and social services. Using U.S. census data and the nurturant care occupations listed in Table A1 in the Appendix, Duffy (2011) estimates that the number of workers providing nurturant care rose from 3% of the U.S. workforce in 1900 to more than 13% in 2007 and that the proportion of female care providers went from 55% to nearly 80% over the same period. This concurrent expansion and feminization of nurturant care occupations contributed to the growth of the female workforce. However, these compositional changes affect women's relative earnings only to the extent that the average wage of care providers differs from the workforce average, and the two averages did not differ greatly over the 1979 to 2015 period as a whole.¹

Because women are greatly overrepresented among nurturant care providers, an increase in the price of these occupational services will disproportionately benefit female workers and raise the gender wage ratio. Of course, increases in the price of an occupational service occur only if demand exceeds supply, and it appears that the demand for nurturant care has outpaced the supply of workers qualified to provide it (Liu & Grusky, 2013). The growing demand for nurturant care may be partly attributable to technological advance. In the revised version of the skill-biased technological change explanation for growing wage inequality, computers complement nonroutine cognitive work by increasing the supply of routine information inputs upon which the performance of cognitive tasks depends (Autor, Katz, & Kearney, 2006). Nurturant care services vary greatly in the extent to which they require nonroutine cognitive skills. However, the quality of care invariably depends on the information available to care providers. By increasing the availability of information and thereby improving the quality of nurturant care, the widespread adoption of computers may have boosted the demand for these service providers. The growing demand for nurturant care providers may also, and more simply, be explained by the marketization of care work. These occupational groups are concentrated in service industries that saw rapid employment growth during the 1980s and 1990s as the market provision of educational, social, and medical services expanded. The supply of workers qualified to provide these services would not have been sufficient to offset the increase in demand when access to training and credentials is restricted, and most nurturant care occupations have professionalized to varying degrees. An even greater number require formal educational degrees or licenses for entry, and Weeden (2002) has shown that the wage returns to these closure mechanisms are comparable across professional and nonprofessional occupations.² Whether attributed to technological change or to the marketization of nurturant care, increased demand for care providers is likely to outpace their supply when access to training and credential is restricted, thereby raising the relative wages of these service providers. In summary, it is expected that compositional shifts in the provision of nurturant care have had little effect on the wage gap and that increased demand and restricted supply raised the price of this care, disproportionately benefitting female care providers and reducing the pay gap.

Reproductive Care Occupations

The number of paid workers providing the cooking, cleaning, and laundry services that constitute reproductive care also increased over the course of the past century. However, the increase in the total employment share of these service occupations has been modest, and the female share of this occupational workforce has decreased. Based upon the list of reproductive care occupations in Appendix Table A1, Duffy (2011) estimates that the proportion of the workforce providing reproductive services increased from 6.8% in 1900 to roughly 8% in 2007. This modest overall growth masked divergent trends. The number of private household workers fell from roughly 1.3 million in 1900 to less than 570,000 in 1990, and at both time points more than 95% of these workers were women. In contrast, the number of workers providing reproductive care services outside of private households increased from 600,000 in 1900, or 2% of the workforce, to over 12 million in 2007, or nearly 8% of the workforce, and men and women entered these jobs in roughly equal proportions throughout the century (Duffy, 2007; Duffy, 2011). Together, these trends indicate that the proportion of reproductive care workers who are women decreased primarily because the number of women doing very low-wage private household work declined. The decrease in the female employment share will contribute to gender wage equality as long as reproductive care jobs pay considerably less than the workforce average. However, the impact of this compositional shift will be limited by the relatively small and slowly growing percentage of the workforce that provides reproductive care and will diminish over time as the proportions of female and male service providers become more equal.

As the female provision of market-based food preparation, cleaning, and laundry services declines, any increase in the price of those services will also have a diminishing effect on the gender wage gap. The evidence regarding the wage gains of low-wage service workers is mixed (cf., Autor & Dorn, 2013; Mishel, Schmitt, & Shierholz, 2013). On balance, the wage gains of reproductive care workers over the 1979 to 2015 period appear to lag those of the workforce as a whole, and any wage effect would have diminished as the ratio of male to female workers within these occupations approached parity. In summary, it is expected that the changing composition of reproductive care has had a modest and diminishing effect on the gender wage gap and that any change in the price of that care has had little or no effect.

Managerial and (Noncare) Professional Occupations

The overrepresentation of women in care-providing occupations has been a defining feature of the social organization of work for over a century. In contrast, women's entry into managerial and professional occupations that are not culturally associated with female labor increased dramatically beginning in the 1970s. As discussed earlier, Goldin (2006) attributes the rapid entry of young women into these occupations to civil rights era reforms that expanded educational and occupational opportunities, to young women's expectations that their future working lives would differ from those of their mothers and, ultimately, to the career choices that occurred in response to expanding opportunities and rising expectations. As the labor force attachment of female cohorts strengthened, their labor supply decisions became less elastic both with regard to their spouse's earnings (a long-term trend) and with regard to their own wages (Blau & Kahn, 2007). In essence, women's labor supply behavior became more like men's. This behavioral change is reflected in a compositional shift, the increasing entry of female workers into managerial and professional occupations, as well as in women's increased training, experience, and relative wage gains across a wide range of occupations.

Women's increased entry into both managerial and noncare professional occupations reflects a similar change in career aspirations, but these occupational shifts are not expected to have equivalent impacts on the gender wage gap. According to Bureau of Labor Statistics (BLS) figures, the percentage of the workforce employed in managerial occupations increased from 10.8% in 1979 to 14.5% in 2007, and the female share of managerial employment grew from 24.6% to 43.4%. The detailed occupational categories reported by the BLS changed considerably over this period, and estimates of the employment changes in traditionally male (i.e., noncare) professions are less precise. Still, it appears that these occupations represent a smaller, more slowly growing, and less feminized share of the workforce, accounting for approximately 8% of employed workers in 2007 of whom roughly a third were women.³ Managerial and professional occupations pay relatively high wages, and the expansion and feminization of both occupational groups should contribute to gender wage equality. However, it is expected that female entry into management has had the larger impact in narrowing the wage gap.

In contrast to changes in their gender composition, increases in the wage returns to managerial and noncare professional occupations are expected to widen the pay gap, although these price effects should diminish as female representation in these occupations increases. Whether or not they are part of the care economy, the price of professional services will rise to the extent that technological change improves their quality, market expansion boosts their demand, and closure mechanisms restrict their supply. There is also evidence that the lean and mean strategies adopted to squeeze labor costs over the past several decades had the paradoxical effect of heightening demand for managerial employees while an increasingly credential-based recruitment restricted their supply, thereby raising their relative wages (Goldstein, 2012; Gordon, 1996; Liu & Grusky, 2013). In summary, it is expected that the changing composition of professional and, especially, managerial occupations substantially narrowed the gender wage gap, that the increasing wage returns to these occupations widened the wage gap, and that both effects have diminished over time.

Summary

This section presents a set of research expectations regarding the impact on gender wage inequality of changes in the gender composition and relative wages of occupations traditionally associated with female labor and of managerial and professional occupations that women rapidly entered as their career horizons and choices expanded. The continuing expansion of the care economy and the rapid female entry into traditionally male managerial and professional occupations figure prominently in sociological accounts of the changing social organization of work. By focusing upon these occupational shifts, the present study examines one of the ways in which changes in job allocation and job rewards are affecting male–female earnings inequality. The key expectation is that the earnings effects of these occupational shifts have diminished, either because the changes in occupational placement have slowed or because the male–female ratio of service provision has become more equal, negating the effect of any price changes. Using a more detailed occupational breakdown may account for more of the change in the pay gap. It would not alter that expectation.

Data and Measures

The data used to analyze changes in the gender wage gap are from the Merged Outgoing Rotation Group (MORG; BLS, various years) files of the Current Population Survey (CPS). Both the descriptive analysis of wage trends and the formal decompositions of those trends use the survey files for the peak business cycle years of 1979, 1989, 1999, and 2007, and for 2015. The MORG samples are limited to noninstitutionalized, civilian wage earners between 18 and 65 years of age. The CPS does not collect comparable earnings data for self-employed workers, and they are excluded from the samples. In the following analyses, these samples are weighted using the BLS-constructed sampling weights.

The dependent variable is the logged hourly wage. Following convention, this variable was calculated for nonhourly workers by first dividing their weekly wages by the number of hours usually worked per week or, if this was missing, by the number of hours worked the week preceding the survey. All workers whose wages fell below $1/hour or above $100/hour in 1979 dollars were excluded, and wages that were top-coded in the CPS to preserve confidentiality were multiplied by 1.4 (Card & DiNardo, 2002). The log of the current wage is used throughout the analysis, although for descriptive purposes Table 1 presents real wage trends in inflation-adjusted 2009 dollars, based on the Bureau of Economic Analysis's Personal Consumption Expenditure Deflator.
Table 1.
Gender Wage Inequality by Occupational Group, 1979, 1989, 1999, 2007, and 2015.

1979 1989 1999 2007 2015 Change 1979– 2015

Part A. All wage earners:

Female employment share 43.4% 46.9% 48.2% 48.4% 48.2% 4.8

Gender log wage gap 0.396 0.288 0.226 0.162 0.165 −0.231

Implied female/male wage ratio 0.673 0.750 0.798 0.850 0.847 0.174

Mean wage (2009 $) $13.80 $13.62 $15.15 $15.96 $16.81 21.8%

Unweighted N^a 160,734 167,183 147,244 160,652 151,759

Part B. Occupational groups:

Nurturant care occupations

Percent of all wage earners 11.0% 11.7% 13.8% 15.6% 17.1% 6.1

Female employment share 74.0% 76.4% 77.6% 78.6% 78.5% 4.5

Female/male wage ratio 0.746 0.758 0.761 0.823 0.790 0.044

Mean occupational wage (2009 $) $13.66 $14.93 $16.69 $17.57 $18.21 33.3%

Reproductive care occupations

Percent of all wage earners 7.7% 7.8% 7.7% 8.2% 8.5% 0.8

Female employment share 64.4% 59.3% 58.4% 56.9% 54.8% −7.5

Female/male wage ratio 0.756 0.795 0.855 0.857 0.8795 0.124

Mean occupational wage (2009 $) $8.59 $7.67 $8.56 $9.50 $9.78 13.9%

Managerial occupations

Percent of all wage earners 10.6% 12.0% 13.7% 12.4% 14.3% 3.7

Female employment share 28.3% 44.3% 49.4% 49.6% 47.9% 19.6

Female/male wage ratio 0.645 0.695 0.729 0.820 0.790 0.145

Mean occupational wage (2009 $) $19.14 $20.05 $22.26 $23.70 $26.15 36.6%

Professional occupations (noncare)

Percent of all wage earners 4.4% 5.0% 6.0% 5.8% 6.6% 2.2

Female employment share 21.1% 28.2% 31.3% 34.1% 33.5% 12.4

Female/male wage ratio 0.620 0.717 0.776 0.788 0.822 0.202

Mean occupational wage (2009 $) $21.83 $23.12 $25.64 $25.34 $27.42 25.6%

All other occupations

Percent of all wage earners 66.3% 63.6% 58.8% 58.0% 53.6% −12.7

Female employment share 39.8% 41.9% 41.3% 40.3% 39.4% −0.4

Female/male wage ratio 0.703 0.768 0.803 0.838 0.840 0.137

Mean occupational wage (2009 $) $13.45 $12.82 $13.82 $14.68 $14.96 11.2%

Note. CPS = Current Population Survey.

a
Data are weighted using CPS earnings weights.

All of the results presented here are based on edited wage data. Missing wage data are imputed in the CPS using aggregated occupational categories. This imputation procedure can misallocate wages for detailed occupational categories, and this match bias may have increased over time as the percentage of cases with missing earnings increased (Hirsch & Schumacher, 2004; Mouw & Kalleberg, 2010). However, the occupational groupings used here are highly aggregated, and a sensitivity check revealed that the results obtained using unedited data did not differ substantively from those using the edited wage series.

The CPS uses the Census occupational coding scheme that is adjusted periodically to capture changes in the occupational structure. To obtain consistent measures of caregiving and noncaregiving occupations, I used the BLS crosswalk developed by Meyer and Osborne (2005) to combine the 1970, 1980/1990 and 2000 census classification codes into a common set of occupational categories (see Dwyer, 2013). Nurturant and reproductive care workers were then identified based on the occupational classifications developed by England (1992; see also England et al., 2002) and Duffy (2005), while noncaregiving professionals and managers were identified by excluding those working in nurturant care occupations (Appendix Table A1).

With few exceptions, the covariates used in the wage decompositions correspond to those in Cha and Weeden's (2014) analysis of the MORG files.⁴ Standard cut points of fewer than 35 hours, 35 to 49 hours, and 50 hours or more a week define part-time, full-time, and overworkers, respectively. Other covariates include gender; race/ethnicity (non-Hispanic White, Black, non-Black Hispanic, and other); age; age squared; education (less than high school, high school, some college, college, and advanced degree); potential years of work experience (i.e., age − years of schooling − 6); potential work experience squared; and region. To control for industry sector, I coded industries into 23 categories based on the standard 22-category scheme, following Dwyer (2013) in splitting “business and repair services” into separate sectors. Marital status is excluded from the wage-change analysis because its price effects cannot be assumed to be consistent for men and women. Other known correlates of wages such as actual work experience, job tenure, and union status were either omitted from the MORG files or were available for only part of the coverage period. The means and standard deviations of the variables used in the wage-change analyses are presented in Table A2 in the Appendix.

Methods

The analysis of the effect of a changing social division of labor on the gender wage gap proceeds in two steps. First, measures of the female employment share, the gender wage gap, and the inflation-adjusted mean wage (2009 $) are calculated for the business cycle peaks of 1979, 1989, 1999, and 2007, as well as 2015. These measures are reported both for the entire sample and for each occupational group. In addition, the total employment share of the occupational groups is reported at each time point.

This descriptive overview sets the stage for the formal wage decomposition. In the present context, the JMP decomposition estimates the effects on the wage gap of changes in the gender composition and wage returns of the different occupational groups while controlling for other determinants of change in the gender wage ratio. This analytic framework apportions change in the gender wage gap into four components. The observed characteristics component shows the effect of changes in the mean male–female difference in observed characteristics, including occupational placement, while the observed prices component captures the effect of changes in the wage returns to those characteristics. The unobserved changes component represents the change in the relative position, or percentile rank, of female workers in the residual male wage distribution. Reductions in the wage gap measured by this component could be due to relative improvements in unobserved characteristics such as work effort and commitment, reduced discrimination, or an unobserved increase in demand that benefits female workers (e.g., an increase in demand for interactive or relational skills relative to the demand for physical strength). Finally, the residual inequality component measures the effect of change in the residual wage distribution, holding constant women's average percentile rank within that distribution. That is, it measures the increase in the gender wage gap that occurs when the wage distribution becomes more unequal. The latter two components equal the change in the unexplained portion of the wage gap in a conventional regression analysis.

The JMP decomposition begins by estimating the male wage equation:
$Y_{it} = X_{it} B_{t} + σ_{t} Θ_{it} .$
(1)
where Y_it is the log wage, X_it is the set of explanatory variables, B_t is the vector of regression coefficients, σ_t is the standard deviation of the male residual wage distribution for a given year, and Θ_it is a standardized residual with mean of zero and variance of one for each year. Based on this equation, the gender pay gap between any 2 years can be decomposed into the
$ObservedCharacteristicsComponent = (Δ X_{1} - Δ X_{0}) B_{1} .$
(2)

$ObservedPricesComponent = Δ X_{0} (B_{1} - B_{0}) .$
(3)

$UnobservedChangesComponent = (Δ Θ_{1} - Δ Θ_{0}) σ_{1} .$
(4)

$ResidualInequalityComponent = Δ Θ_{0} (σ_{1} - σ_{0}) .$
(5)
where Δ signifies the average male–female difference in the variable it precedes.⁵ The equations for observed characteristics and prices are the primary focus of inquiry because they estimate the reductions in the wage gap that can be attributed to changes in the gender composition and relative wages of the different occupational groups, adjusted for changes in all other observed characteristics.

Overview of Gender Wage Gap Trends

Part A of Table 1 displays the trend change in the female share of all wage earners, in the overall gender wage gap, and in the inflation-adjusted average wage. Consistent with the findings of previous studies (e.g., Autor et al., 2006; Blau & Kahn, 1997, 2006), the relative earnings of female workers improved substantially during the 1980s, rising from 67.3% to 75.0% of the male average between 1979 and 1989. This relative wage gain continued at a slower pace during the 1990s and 2000s, reaching 85% of the male average in 2007. Since then there has been no additional wage gain. The gender wage ratio remained at 0.847 in 2015, 7 years after the recent business cycle peak. Progress toward gender earnings equality ceased during the Great Recession and its aftermath.

Part B of Table 1 documents the changes in the composition and relative wages of the occupational subgroups. As these figures show, the expansion and feminization of market-based nurturant care continued throughout the coverage period. Nurturant care providers accounted for 11% of all wage earners in 1979 and for 17.1% in 2015, and the relative number of female care providers increased by 4.5 percentage points, from 74% to 78.5%. However, the mean wage of these occupations is close to the workforce average, so the effect of these occupational changes on the gender pay gap would have been small. What is likely to have narrowed the wage gap is the rising price of nurturant care. Measured in 2009 dollars, the mean hourly wage of nurturant care providers rose by 33.3% between 1979 and 2015 (from $13.66 to $18.21), considerably more than the 21.8% real wage gain ($13.80 to $16.81) of all wage earners.

Reproductive care occupations employ a smaller and more slowly growing percentage of the workforce, one that increased from only 7.7% to 8.5% between 1979 and 2015, and the female share of these low-wage jobs fell by nearly ten percentage points. Because these occupations pay well below the workforce average, the decline in their female employment share would have contributed to the narrowing of the overall wage gap. By comparison, changes in the relative price of reproductive care would have had little or no impact. The real wages of reproductive care providers increased by 13.9% over the coverage period (from $8.59 to $9.78), less than the workforce average, and the effect of this price change would have diminished as the ratio of female-to-male workers within these occupations approached parity.

The compositional shifts most likely to have contributed to the narrowing of the gender pay gap occurred in managerial and, to a lesser extent, male-dominated professional occupations. The total employment share of managerial employees increased by 3.7 percentage points and that of professionals by 2.2 percentage points, over the coverage period. Female representation within these occupations also increased substantially during the 1980s and 1990s, before leveling off after 1999. Over the entire coverage period, the relative number of female employees increased within managerial occupations by 19.6 percentage points (i.e., from 28.3% to 47.9%) and within noncare professional occupations by 12.4 percentage points (i.e., from 21.1% to 33.5%). Because these occupations pay well above the workforce average, their concurrent expansion and feminization would have contributed to the initial narrowing of the gender wage gap, while the slowing of these occupational shifts helps to explain the subsequent lack of progress. Partly offsetting these inequality-reducing effects, the relatively large increase in the wage returns to professional and managerial employment would have widened the wage gap. Between 1979 and 2015, the mean hourly wage of managerial employees increased by 36.6% while that of noncare professionals went up 25.6%. Although this price effect would diminish as women increasingly entered these occupations, the overrepresentation of male employees means they would disproportionately benefit from these relative wage gains.

Decompositions of the Gender Wage Gap

Table 2 presents a more accurate assessment of the wage-equalizing effects of occupational career shifts by decomposing the change in the gender pay gap over the entire coverage period. The figures at the top indicate that the gender wage differential decreased by 0.231 log points between 1979 and 2015, a figure equivalent to the changes in the log wage gap and implied female/male wage ratio shown in Table 1. The coefficients in each column (Models 1 and 2) show the portion of this overall wage change that can be attributed to observed compositional and price changes and to unobserved changes. Model 1 shows the decomposition of the overall wage change when logged hourly wage is regressed on the four occupational groups (i.e., “other” occupations is the residual category), and Model 2 presents the decomposition when control variables are added to the regression equations. (Tables A3 and A4 in the Appendix report the regression equations upon which the wage decompositions are based.)
Table 2.
Decomposition of Changes in Gender Pay Gap, 1979 to 2015.

Decomposition components Model 1
Model 2

Percent change Percent change

Change in differential −.231 −.231

All observed characteristics −.046 19.9% −.064 27.7%

Nurturant care −.006 2.6% −.001 0.4%

Reproductive care −.018 7.8% −.008 3.5%

Managers −.021 9.1% −.012 5.2%

Professionals (noncare) −.000 0.0% −.000 0.0%

Overwork .003 −1.3%

Part-time variables −.010 4.3%

Potential experience variables −.027 11.7%

Education variables −.005 2.2%

Race variables −.002 0.9%

Region variables .001 −0.4%

Industry variables −.003 1.3%

All observed prices −.039 16.9% −.031 13.4%

Nurturant care −.046 19.9% −.007 3.0%

Reproductive care .000 0.0% −.001 0.4%

Managers/professionals .001 −0.4% .000 0.0%

Professionals (noncare) .006 −2.6% .002 −0.9%

Overwork .017 −7.4%

Part-time variables .006 −2.6%

Potential experience variables −.007 3.0%

Education variables −.005 2.2%

Race variables .002 −0.9%

Region variables .000 0.0%

Industry variables −.037 16.0%

Unexplained differential −.146 63.2% −.136 58.9%

Unobserved changes −.173 74.9% −.158 68.4%

Residual inequality effect .028 −12.1% .022 −9.5%

Unweighted N^a 312,493 311,853

Note. CPS = Current Population Survey.

a
Data are weighted using CPS earnings weights.

The summary figures (All observed characteristics and All observed prices) in Model 1 indicate that changes in the composition and relative wages of these occupational groups account for nearly 37% (i.e., 0.085/0.231) of the total wage convergence, a figure nearly equal to the percent explained by more detailed occupational breakdowns.⁶ As expected, the changes in managerial employment reduced the wage gap by 0.021 log points, more than 9% of the overall reduction, but surprisingly, the employment shifts among traditionally male (noncare) professions did not contribute to the overall wage convergence. Compositional shifts in the occupations providing reproductive and nurturant care are associated with smaller wage-gap reductions of 0.018 and 0.006 log points, respectively. And as expected, the most important price change involved the highly feminized occupations providing nurturant care. Judging from the uncontrolled wage change equations, the increasing wage returns to these occupations reduced the pay gap by 0.046 log points, nearly a fifth of the overall convergence. Changes in the relative wages of workers providing reproductive care had no effect on the wage gap, while the relative wage gains associated with managerial and professional occupations in which women continue to be underrepresented increased the pay gap by 0.001 and 0.006 log points, respectively.

Although these observed changes account for a substantial portion of the overall wage convergence, most of the 0.231 log point reduction in the gender wage gap is due to unobserved changes. The 0.173 log point reduction attributable to unobserved factors indicates that the mean percentile rank of female workers within the residual wage distribution rose substantially, presumably due to a relative improvement in their unobserved earnings-related characteristics, to a decline in gender discrimination, or to an increase in the demand for female labor that is not captured by the occupational shifts or other covariates. The effect of these unobserved changes was partially offset by a 0.028 log point increase in residual wage inequality. This residual inequality effect measures the extent to which the gender wage gap widened due to the increased dispersion of the wage distribution.

The coefficients in the Model 2 column show the decomposition of the overall change in the gender pay gap when the control variables are added to the regression equations. Adding these measures reduces the impact of the occupational shifts. The reduction in the pay gap attributable to changes in managerial employment decreases from 0.021 to 0.012 log points, or 5.2% of the total reduction, while the wage gap reductions associated with the changing composition of reproductive and nurturant care occupations fall to 0.008 and 0.001 log points, respectively. Adding control measures reduces the impact of occupational price changes to an even greater extent. In particular, the wage gap reduction directly attributable to the earnings gains of nurturant care providers decreases from 0.046 to 0.007 log points. The large coefficient measuring the price effect of the industry variables (−0.037 log points) indicates that the relative wage gains benefitting female workers are occurring in the industries in which nurturant care services are concentrated and are not limited to the workers providing those occupational services.⁷ It is also important to note that the beneficial effects of these price changes were partly offset by increasing returns to overwork, as Cha and Weeden (2014) have shown. That is, because women are underrepresented among those who work 50 or more hours a week, they were disadvantaged by the increase in the wage returns to long work hours. The estimates in Table 2 indicate that this price increase, combined with the much smaller compositional change in overwork, added 0.020 log points to the gender pay gap between 1979 and 2015.

Controlling for individual variation in work hours, human capital, race and ethnicity, and regional and industrial location reduces, but does not eliminate, the wage-equalizing effects of occupational career change. The expansion of managerial employment, and the increase in the relative number of female managers, clearly contributed to the overall reduction in the gender pay gap. To a lesser extent, compositional changes associated with the decline of private household work also contributed to wage convergence. The direct effects of these occupational changes are outweighed by the reduction in the pay gap associated with the potential experience gains of female workers and, to an even greater extent, by the reduction resulting from unobserved changes. However, it is reasonable to assume that both experience gains and unobserved changes reflect a strengthening job and labor force attachment that developed in response to expanding career opportunities. In other words, the occupational changes examined here contributed indirectly as well as directly to gender wage convergence.

The direct and likely indirect effects of occupational change become somewhat clearer when we examine the slowing and eventual cessation of wage convergence after the 1980s. Table 3 presents the wage change decompositions for the subperiods 1979 to 1989, 1989 to 1999, 1999 to 2007, and 2007 to 2015. The “change in differential” figures in the top row indicate that the gender wage gap narrowed by 0.109 log points during the 1980s and by 0.062 and 0.063 log points, respectively, during the 1990s and early 2000s. It then widened by 0.003 log points between 2007 and 2015. Broken down by subperiod, the occupation employment shifts that help to explain the 1980's wage convergence and its subsequent slowdown involve managerial and reproductive care jobs. Changes in the female share of higher wage managerial jobs accounted for 0.010 log points (9.2%) of the wage convergence in the 1980s and 0.005 log points (8.1%) in the 1990s, had no effect in the period 1999 to 2007, and added 0.002 log points to the wage gap after 2007. Similarly, the decrease in the proportion of lower wage reproductive care jobs held by women reduced the wage gap by 0.005 log points (4.6%) in the 1980s, by 0.002 log points (3.2%) in the 1990s, and had negligible effects thereafter. Employment shifts within nurturant care and professional occupations had negligible effects upon the wage gap across subperiods. Broken down by subperiod, the only clear price trend involves the relative wage gains and losses of workers in different industries. This industry effect accounted for 0.016 log points (14.7%) of the wage convergence in the 1980s, 0.013 log points (21%) in the 1990s, 0.004 log points (6.3%) in the period 1999 to 2007, and had no effect on the wage gap from 2007 to 2015. It appears that the concentration of female workers in industries (e.g., hospitals and health services) with relatively rapid wage growth, and the concentration of male workers in industries (e.g., durable goods manufacturing) with relatively little or no wage growth, reduced the wage gap during the 1980s and 1990s. It is also apparent that differences in industry wage growth are no longer contributing to gender wage convergence.
Table 3.
Decomposition of Changes in Gender Pay Gap by Subperiods, 1979 to 2015.

Decomposition components 1979–1989
1989–1999
1999–2007
2007–2015

Percent change Percent change Percent change Percent change

Change in differential −.109 −.062 −.063 .003

All Observed Characteristics −.038 34.9% −.016 25.8% −.030 47.6% .002 66.7%

Nurturant .000 0.0% .000 0.0% −.001 1.6% .000 0.0%

Reproductive −.005 4.6% −.002 3.2% −.001 1.6% −.001 −33.3%

Managers −.010 9.2% −.005 8.1% .000 0.0% .002 66.7%

Professionals (noncare) −.000 0.0% .001 −1.6% −.002 3.2% .001 33.3%

Overwork .000 0.0% .000 0.0% −.001 1.6% .000 0.0%

Part-time variables −.004 3.7% −.002 3.2% −.003 4.8% −.002 −66.7%

Potential experience variables −.012 11.0% −.007 11.3% −.016 25.4% .001 33.3%

Education variables −.001 0.9% −.002 3.2% −.005 7.9% .001 33.3%

Race variables −.002 1.8% .001 −1.6% −.002 3.2% .002 66.7%

Region variables .001 −0.9% .001 −1.6% .000 0.0% .000 0.0%

Industry variables −.005 4.6% −.001 1.6% .000 0.0% −.002 −66.7%

All Observed Prices .006 −5.5% −.017 27.4% −.002 3.2% .000 0.0%

Nurturant .003 −2.8% −.009 14.5% .007 −11.1% −.007 −233.3%

Reproductive .001 −0.9% .001 −1.6% −.001 1.6% −.001 −33.3%

Managers .001 −0.9% .000 0.0% .000 0.0% .000 0.0%

Professionals (noncare) .003 −2.8% .000 0.0% −.001 1.6% .000 0.0%

Overwork .009 −8.3% .009 −14.5% −.004 6.3% .006 200.0%

Part-time variables .006 −5.5% −.004 6.5% .001 −1.6% .004 133.3%

Potential experience variables −.001 0.9% .000 0.0% .002 −3.2% −.001 −33.3%

Education variables .000 0.0% .000 0.0% −.002 3.2% −.002 −66.7%

Race variables .000 0.0% .000 0.0% .001 −1.6% .001 33.3%

Region variables .000 0.0% −.001 1.6% .000 0.0% .000 0.0%

Industry variables −.016 14.7% −.013 21.0% −.004 6.3% .000 0.0%

Unexplained differential −.077 70.6% −.029 46.8% −.031 49.2% .001 33.3%

Unobserved changes −.094 86.2% −.029 46.8% −.032 50.8% −.009 −300.0%

Residual inequality effect .017 −15.6% .000 0.0% .001 −1.6% .010 333.3%

Unweighted N^a 327,525 313,181 306,352 311,865

Note. CPS = Current Population Survey.

a
Data are weighted using CPS earnings weights.

Most of the 1980 wage convergence was due to changes in unobserved factors, and most of the slowdown in wage convergence since then is accounted for by their diminishing effect. Unobserved changes accounted for 0.094 of the 0.109 log point reduction in the pay gap during the 1980s, and the wage-equalizing effect of these changes diminished to 0.029 and 0.038 log points, respectively, during the 1990s and early 2000s, and to only 0.009 log points after 2007. In their analysis of wage trends in the Panel Study of Income Dynamics, Blau and Kahn (2006) similarly found that unobserved changes account for most of the 1980's wage convergence, and that the 1990s slowdown in wage convergence was entirely accounted for by their diminished effect. We can only speculate as to the sources of this improvement in women's relative earnings, but these authors point to several changes that help to explain women's rapid wage gains in the 1980s.⁸ During that decade, there was a large decline in the gender gap in housework and a rapid rise in female labor force participation, especially among better educated women, indicating that the work commitment and effort of female workers is likely to have strengthened considerably. Employers' perceptions of this increased work commitment and effort would, in turn, have undermined the rationale for statistical discrimination, which may have weakened accordingly. In brief, the contention that the work commitment and effort shown by female workers strengthened, and that the discrimination they encounter lessened, is consistent with observed changes in female labor force participation and in the household division of labor during the 1980s. But it is also reasonable to conclude that the changes in female labor force participation and household labor, as well as in women's work commitment and effort, developed in conjunction with, and partly in response to, improved career opportunities. As the components of change in Table 3 show, the wage-equalizing effects of both occupational employment shifts and unobserved behavioral changes are most pronounced during the 1980s and diminish thereafter.

Whatever their source, changes in unobserved factors do not fully account for the virtual cessation of wage convergence after 2007. As the “change in differential” at the top of Table 3 shows, the overall pay gap decreased by 0.063 log points between 1999 and 2007 and then increased by 0.003 log points during the most recent period. Nearly half of the cessation in wage convergence after 2007 is accounted for by the diminished effect of observed characteristics. Changes in measured characteristics narrowed the pay gap by 0.030 log points in the early 2000s and then widened it by 0.002 log points during the most recent period. In particular, women's relative gains in education and potential experience have little or no impact on the pay gap after 2007. This reduced effect does not indicate an absolute or even a relative decline in the education or potential experience of female wage earners. It does suggest that any relative improvement in these earnings-related characteristics has either slowed or is having little effect on the remaining wage gap in the absence of further occupational career gains.

Discussion

The accounting framework of formal wage decompositions is used to apportion the convergence in the gender wage gap between changes in the occupational placement of female workers and changes in the wage returns of different occupations. As expected, female entry into high-wage managerial occupations, and their decreasing representation in low-wage reproductive care jobs, contributed substantially to the narrowing of the pay gap during the 1980s. The wage-equalizing effects of these employment shifts have since diminished and ceased. These occupational employment shifts coincide with or precede the changes in unobserved factors and in other observed characteristics that account for most of the wage convergence and its subsequent slowdown. Although the different components of wage change are statistically independent, they are not unrelated. Occupational career change, educational and experience gains, and strengthening work commitment and effort have a common origin in the expanded horizons and work-based identity of the female cohorts that began entering the labor market in the 1970s. The slowing and recent cessation of wage convergence suggests that the occupational career shifts that accompanied this quiet revolution will not be sufficient to achieve gender earnings equality.

Changes in the relative earnings of different occupational groups impact the pay gap to the extent that women are over- or underrepresented in those occupations. Because women are increasingly overrepresented among nurturant care providers, it was also expected that increases in the price of that care would contribute to wage convergence. The results of the wage decompositions are only partly consistent with that expectation. In the decomposition that included only occupational measures, the relative wage gains of nurturant care providers accounted for nearly a fifth of the overall wage convergence. However, this price effect was greatly reduced when measures of industry location were added to the equations. The wage gains of nurturant care providers were mainly due to relatively rapid wage growth in the industries in which female workers, including nurturant care providers, are concentrated. Examining this industry wage gain across subperiods reveals that it too is no longer contributing to gender wage convergence. Previous studies have shown that nurturant care providers incur a substantial wage penalty (e.g., England et al., 2002; Folbre, 2001), and it should be noted that the relative wage gains of these occupational groups, documented in Table 1, do not contradict this finding. Most nurturant care providers work in professional and semiprofessional occupations; yet, they continue to earn little more than the workforce average. The results of the wage decompositions indicate only that the size of any penalty was reduced, primarily due to industry differences in wage growth during the 1980s and 1990s.

It is important to acknowledge the limitations of this analysis. As noted earlier, the MORG files contain relatively few measures of the individual characteristics and labor market conditions known to affect wages. In addition, the analysis of occupational employment and price changes is limited to very broad groupings. These groupings account for most of the wage convergence that can be attributed to occupational change, and the use of more detailed categories would not alter the finding that occupational employment shifts are no longer narrowing the wage gap. Nevertheless, research using more detailed occupational breakdowns may contribute to our understanding of how specific changes in job allocation and rewards affect earnings inequality. Apart from the limitations imposed by data and measures, the analytic strategy constrained the nature of the inquiry and the extent of its findings. Decompositions provide an accounting of the components of wage change; they do not account for the remaining pay gap. However, viewed in the light of recent studies, the findings presented here help to identify some of the factors sustaining gender wage inequality and indicate possible directions of future research.

The first of these findings is the continuing effect of wage dispersion on the gender pay gap. Previous studies showed that female workers have been “swimming upstream,” raising their average earnings within a widening distribution (Blau & Kahn, 1997; Datta Gupta et al., 2006). The residual inequality component of the wage decompositions indicates that wage dispersion continues to impede progress toward earnings equality. Swimming against this current is particularly difficult in the United States. Without centralized collective bargaining, the United States relies upon legislated increases in the minimum wage to raise the wage floor and compress the earnings distribution. In the absence of regular increases, the declining real value of the minimum wage and the erosion of legal and normative labor market standards result in both a widening gender wage gap and greater within-group inequality (McCall, 2000). Eroding labor market standards and a falling wage floor particularly disadvantage female workers because they are more likely than males to hold part-time, contingent, and temporary jobs that pay very low wages (Bernhardt, Boushey, Dresser, & Tilly, 2008). Immigrant female workers also experience more employment law violations and wage theft than their male counterparts (Petrescu-Prahova & Spiller, 2016). The linkages between a falling wage floor, nonstandard work, and gender earnings inequality remain an important topic for future research, but the immediate effect of raising the wage floor is easily demonstrated. Recoding hourly earnings in the 2015 MORG file to a minimum of $12 an hour raised the female–male wage ratios in Table 1 from 0.847 to 0.871 among all wage earners, and from 0.880 to 0.939 among those in lower wage reproductive care occupations.⁹ These static estimates do not take into account the behavioral responses of employers and workers to such a wage increase and, therefore, do not show how the resulting wage gains and job losses would impact the female–male wage ratio over the longer term. They do indicate that the concentration of female workers at the bottom of a widening wage distribution accounts for a substantial portion of the remaining pay gap.

A second finding concerns the increasing returns to long work hours, a form of wage dispersion that exacerbates female–male earnings differences throughout the workforce. Controlling for detailed occupational location, Cha and Weeden (2014) demonstrated how the increasing returns to overwork are widening the pay gap, especially among professional and managerial employees, and the wage decompositions reported earlier partly replicate that finding. Similarly, Goldin (2014) showed that the gender wage disparity among employees in high-wage occupations is largely explained by the tendency of some work organizations to reward individuals who work long and continuous hours and claimed that the pay gap will close as competitive forces reward employers who offer work-time flexibility and equal hourly pay rates to workers in the same occupations. The increasing nonlinearity of pay with respect to work hours is exacerbating gender earnings disparities, but it should be clear that eliminating that nonlinearity will not equalize pay within occupations or close the gender pay gap. Using the 2015 MORG data, I calculated the mean female and male wage for those working 35 to 40 hours a week within each of the detailed categories of the Standard Occupational Classification (SOC) and then assigned those respective averages to all fulltime male and female workers in each three-digit occupational category. This raised the overall female–male earnings ratio in Table 1 from 0.847 to 0.858 and had similarly modest effects on the wage ratios within each broad occupational group. Eliminating nonlinearity in the detailed occupational pay of all full-time workers thus reduces the overall pay gap while leaving most of the gender earnings disparity unexplained. In most occupations, women's hourly earnings are considerably less than those of men regardless of the number of hours worked.

A final and related finding with implications for future research concerns the relative importance of intra- and interoccupational components of the gender wage gap. With the exception of reproductive care occupations, the female–male wage ratios within the broad occupational groups in Table 1 are lower than the workforce average of 0.847. Each of these broad groups includes numerous occupations with different average pay, and most of the gender pay gap may still be attributable to differences in the occupational placement of female and male workers (i.e., interoccupational wage differences). Yet, in a comparison based upon highly detailed occupational categories, Goldin (2014) reports that within-occupation variation accounts for most of the gender wage gap. Analysis of the 2015 MORG file similarly revealed that at least three-quarters of the gender wage difference resides within both the 320 detailed categories of the BLS occupational classification and the 483 categories of the SOC.¹⁰ Comparing within- and between-occupation components tells us how much of the pay gap would be eliminated if we were to equalize men's and women's average earnings within each occupation as opposed to equalizing the distributions of male and female workers across occupations. It appears that continued progress in closing the gender wage gap will depend primarily on equalizing women's and men's average occupational earnings.

This conclusion may be disputed. In what they term the fractal hypothesis, Levanon and Grusky (2016) assert that essentialist beliefs—the view that women and men differ in their fundamental interests and abilities and are therefore better suited for jobs that differ correspondingly in their intrinsic occupational traits—not only sustain occupational segregation but also generate an ever more finely grained occupational division of labor. Support for this hypothesis is adduced from studies showing that as growing numbers of women enter traditionally male occupations and those occupations integrate at an aggregate level, segregation reemerges at a more disaggregate level in the form of male- and female-dominated occupational specialties (e.g., Cohen, Huffman, & Knauer, 2009; Ku, 2011). The finding that gender wage inequality arises primarily within occupations may therefore be a statistical artifact, a failure of even detailed occupational breakdowns to capture wage differences constantly being generated by occupational specialization. Determining the extent to which specialization and a more finely grained division of labor sustain gender wage inequality is an important goal for future research. However, essentialist beliefs do not always predispose men to enter jobs that pay well, and women to enter jobs that pay poorly. As Levanon and Grusky (2016) emphasize, physical essentialism, the widely accepted view that men are better suited to jobs requiring physical strength, channels men into occupations that now pay less than the female average. The one essentialist belief that sustains occupational segregation and appears to contribute to wage inequality concerns sociability. These authors confirm that women are much more likely than men to enter occupations that embody sociability, and these occupations incur a wage penalty, an outcome that helps to explain the relatively low earnings of many nurturant care providers. With this notable exception, it is not clear that essentialist beliefs and occupational specialization account for the persistence of gender wage inequality or preclude the closure of the wage gap.

An alternative explanation for intra-occupational wage differences emphasizes the allocative and evaluative processes that adversely affect the quality of women's job matches. For example, several recent studies document the tendency of couples to prioritize men's earnings in making (re)location decisions. Sorenson and Dahl's (2016) study of blue-collar and lower level white-collar Danish couples found that the mismatch of women to potential employers following a move accounted for as much as 36% of the gender wage differential, and studies of U.S. couples similarly find that husbands' human capital better predicts relocation decisions than that of wives (Shaumann, 2010; Tenn, 2010). The tendency of couples to give priority to men's earnings in location decisions, like the more thoroughly investigated motherhood wage penalty, is rooted in family role differences. These allocative processes systematically disadvantage female workers and violate the liberal egalitarian principle that women have a right to compete on equal terms with men for better paying jobs. In the final analysis, these discriminatory processes are justified by gender status beliefs and the lesser value accorded female labor. Studies based upon reward expectations theory show that both women and men internalize the existing status inequalities within the workplace and, as a consequence, perceive the lower wages paid to female workers as fair (Auspurg, Hinz, & Sauer, 2017; Jasso & Webster, 1999). And as Auspurg et al. (2017) conclude, only the widespread attainment of high-status positions by female role models will transform gender status beliefs and overcome discriminatory evaluations.

Conclusions

The narrowing of the earnings gap is an important indicator of progress toward gender equality, and studies have long pointed to slowing wage convergence as evidence of a stalled gender revolution (e.g., Blau & Kahn, 2016; Cotter, Hermsen, & Vanneman, 2004; England, 2010). The wage decompositions presented here contribute to this literature by investigating how women's occupational career shifts help to account both for the narrowing of the wage gap and for the subsequent slowing and cessation of wage convergence. They highlight how the rapid entry of female workers into management and, to a lesser degree, their exit from private household work, reduced the pay gap during the 1980s and 1990s, and how these wage-equalizing effects ended as female employment in high-status occupations leveled off. Female entry into management in particular has attracted research attention both because of the rewards these occupations offer and because the female exercise of managerial authority can promote gender equality within the workplace and the larger society (Cohen et al., 2009; Cohen & Huffman, 2007). Wage decompositions cannot assess the broader implications of the female exercise of authority, but the analysis does reveal that the gender wage ratio rose substantially within managerial occupations as the female employment share increased. In other words, female entry into management was accompanied by real wage gains and cannot be simply attributed to occupational reclassification (cf., Cohen et al., 2009; Jacobs, 1992).

The results also shed light on the sources of the remaining pay gap and point out possible directions for future research. Wage decompositions quantify the effect on group inequality of changes in the earnings distribution, and the results indicate that female workers continue to be disadvantaged by a widening wage distribution. The detrimental effects of wage dispersion are greatest at the bottom where a falling wage floor and eroding workplace standards result in the proliferation of nonstandard jobs and employment law violations. Consistent with previous studies (Cha & Weeden, 2014; Goldin, 2014), the results also show that wage dispersion in the form of increasing returns to long work hours is widening the gender wage gap, and eliminating the nonlinearity of earnings with respect to work hours modestly reduces wage inequality. Finally, the results indicate that most of the remaining wage gap resides within occupations. Improving women's occupational placement would therefore have only a modest direct effect on the remaining wage gap, just as changes in women's occupational careers directly account for a limited amount of the wage convergence that occurred over the past several decades. However, the presence of growing numbers of women in high-wage, high-status positions raises the status and reward expectations of female workers. Occupational career gains thereby improve women's job matches and help to equalize female–male earnings throughout the workforce.

	1979	1989	1999	2007	2015	Change 1979– 2015
Part A. All wage earners:
Female employment share	43.4%	46.9%	48.2%	48.4%	48.2%	4.8
Gender log wage gap	0.396	0.288	0.226	0.162	0.165	−0.231
Implied female/male wage ratio	0.673	0.750	0.798	0.850	0.847	0.174
Mean wage (2009 $)	$13.80	$13.62	$15.15	$15.96	$16.81	21.8%
Unweighted N^a	160,734	167,183	147,244	160,652	151,759
Part B. Occupational groups:
Nurturant care occupations
Percent of all wage earners	11.0%	11.7%	13.8%	15.6%	17.1%	6.1
Female employment share	74.0%	76.4%	77.6%	78.6%	78.5%	4.5
Female/male wage ratio	0.746	0.758	0.761	0.823	0.790	0.044
Mean occupational wage (2009 $)	$13.66	$14.93	$16.69	$17.57	$18.21	33.3%
Reproductive care occupations
Percent of all wage earners	7.7%	7.8%	7.7%	8.2%	8.5%	0.8
Female employment share	64.4%	59.3%	58.4%	56.9%	54.8%	−7.5
Female/male wage ratio	0.756	0.795	0.855	0.857	0.8795	0.124
Mean occupational wage (2009 $)	$8.59	$7.67	$8.56	$9.50	$9.78	13.9%
Managerial occupations
Percent of all wage earners	10.6%	12.0%	13.7%	12.4%	14.3%	3.7
Female employment share	28.3%	44.3%	49.4%	49.6%	47.9%	19.6
Female/male wage ratio	0.645	0.695	0.729	0.820	0.790	0.145
Mean occupational wage (2009 $)	$19.14	$20.05	$22.26	$23.70	$26.15	36.6%
Professional occupations (noncare)
Percent of all wage earners	4.4%	5.0%	6.0%	5.8%	6.6%	2.2
Female employment share	21.1%	28.2%	31.3%	34.1%	33.5%	12.4
Female/male wage ratio	0.620	0.717	0.776	0.788	0.822	0.202
Mean occupational wage (2009 $)	$21.83	$23.12	$25.64	$25.34	$27.42	25.6%
All other occupations
Percent of all wage earners	66.3%	63.6%	58.8%	58.0%	53.6%	−12.7
Female employment share	39.8%	41.9%	41.3%	40.3%	39.4%	−0.4
Female/male wage ratio	0.703	0.768	0.803	0.838	0.840	0.137
Mean occupational wage (2009 $)	$13.45	$12.82	$13.82	$14.68	$14.96	11.2%

Decomposition components	Model 1	Model 2
Change in differential	−.231		−.231
All observed characteristics	−.046	19.9%	−.064	27.7%
Nurturant care	−.006	2.6%	−.001	0.4%
Reproductive care	−.018	7.8%	−.008	3.5%
Managers	−.021	9.1%	−.012	5.2%
Professionals (noncare)	−.000	0.0%	−.000	0.0%
Overwork			.003	−1.3%
Part-time variables			−.010	4.3%
Potential experience variables			−.027	11.7%
Education variables			−.005	2.2%
Race variables			−.002	0.9%
Region variables			.001	−0.4%
Industry variables			−.003	1.3%
All observed prices	−.039	16.9%	−.031	13.4%
Nurturant care	−.046	19.9%	−.007	3.0%
Reproductive care	.000	0.0%	−.001	0.4%
Managers/professionals	.001	−0.4%	.000	0.0%
Professionals (noncare)	.006	−2.6%	.002	−0.9%
Overwork			.017	−7.4%
Part-time variables			.006	−2.6%
Potential experience variables			−.007	3.0%
Education variables			−.005	2.2%
Race variables			.002	−0.9%
Region variables			.000	0.0%
Industry variables			−.037	16.0%
Unexplained differential	−.146	63.2%	−.136	58.9%
Unobserved changes	−.173	74.9%	−.158	68.4%
Residual inequality effect	.028	−12.1%	.022	−9.5%
Unweighted N^a	312,493		311,853

Decomposition components	1979–1989	1989–1999	1999–2007	2007–2015
Change in differential	−.109		−.062		−.063		.003
All Observed Characteristics	−.038	34.9%	−.016	25.8%	−.030	47.6%	.002	66.7%
Nurturant	.000	0.0%	.000	0.0%	−.001	1.6%	.000	0.0%
Reproductive	−.005	4.6%	−.002	3.2%	−.001	1.6%	−.001	−33.3%
Managers	−.010	9.2%	−.005	8.1%	.000	0.0%	.002	66.7%
Professionals (noncare)	−.000	0.0%	.001	−1.6%	−.002	3.2%	.001	33.3%
Overwork	.000	0.0%	.000	0.0%	−.001	1.6%	.000	0.0%
Part-time variables	−.004	3.7%	−.002	3.2%	−.003	4.8%	−.002	−66.7%
Potential experience variables	−.012	11.0%	−.007	11.3%	−.016	25.4%	.001	33.3%
Education variables	−.001	0.9%	−.002	3.2%	−.005	7.9%	.001	33.3%
Race variables	−.002	1.8%	.001	−1.6%	−.002	3.2%	.002	66.7%
Region variables	.001	−0.9%	.001	−1.6%	.000	0.0%	.000	0.0%
Industry variables	−.005	4.6%	−.001	1.6%	.000	0.0%	−.002	−66.7%
All Observed Prices	.006	−5.5%	−.017	27.4%	−.002	3.2%	.000	0.0%
Nurturant	.003	−2.8%	−.009	14.5%	.007	−11.1%	−.007	−233.3%
Reproductive	.001	−0.9%	.001	−1.6%	−.001	1.6%	−.001	−33.3%
Managers	.001	−0.9%	.000	0.0%	.000	0.0%	.000	0.0%
Professionals (noncare)	.003	−2.8%	.000	0.0%	−.001	1.6%	.000	0.0%
Overwork	.009	−8.3%	.009	−14.5%	−.004	6.3%	.006	200.0%
Part-time variables	.006	−5.5%	−.004	6.5%	.001	−1.6%	.004	133.3%
Potential experience variables	−.001	0.9%	.000	0.0%	.002	−3.2%	−.001	−33.3%
Education variables	.000	0.0%	.000	0.0%	−.002	3.2%	−.002	−66.7%
Race variables	.000	0.0%	.000	0.0%	.001	−1.6%	.001	33.3%
Region variables	.000	0.0%	−.001	1.6%	.000	0.0%	.000	0.0%
Industry variables	−.016	14.7%	−.013	21.0%	−.004	6.3%	.000	0.0%
Unexplained differential	−.077	70.6%	−.029	46.8%	−.031	49.2%	.001	33.3%
Unobserved changes	−.094	86.2%	−.029	46.8%	−.032	50.8%	−.009	−300.0%
Residual inequality effect	.017	−15.6%	.000	0.0%	.001	−1.6%	.010	333.3%
Unweighted N^a	327,525		313,181		306,352		311,865

Footnotes

Appendix

Acknowledgments

I would like to thank the anonymous reviewers for their helpful comments. I especially want to thank Scott Drewianka for comments clarifying the math and logic of the JMP wage decomposition.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Thomas S. Moore is retired from the University of Wisconsin-Milwaukee. His recent work has appeared in Sociological Forum, American Journal of Sociology, and the Journal of Sociology & Social Welfare.

References

Auspurg

Hinz

Sauer

(2017) Why should women get less? Evidence on the gender pay gap from multifactorial survey experiments. American Sociological Review 82: 179–210.

Autor

D. H.

Dorn

(2013) The growth of low-skill service jobs and the polarization of the U.S. labor market. American Economic Review 103: 1553–1597.

Autor

D. H.

Katz

L. F.

Kearney

M. S.

(2006) The polarization of the U.S. labor market. American Economic Review: Papers and Proceedings 96: 189–194.

Bernhardt

Boushey

Dresser

Tilly

(2008) The gloves-off economy: Workplace standards at the bottom of the labor market, Ithaca, NY: Cornell University Press.

Blau

F. D.

Kahn

L. M.

(1997) Swimming upstream: Trends in the gender wage differential in the 1980s. Journal of Labor Economics 15: 1–42.

Blau

F. D.

Kahn

L. M.

(2006) The U.S. gender pay gap in the 1990s: Slowing convergence. Industrial and Labor Relations Review 60: 45–66.

Blau

F. D.

Kahn

L. M.

(2007) Changes in the labor supply behavior of married women: 1980-2000. Journal of Labor Economics 25: 393–438.

Blau, F. D., & Kahn, L. M. (2016, January). The gender wage gap: Extent, trends, and explanations (Working Paper No. 21913). Cambridge, MA: National Bureau of Economic Research.

Bureau of Labor Statistics. (various years). Merged outgoing rotation group of the current population survey. Retrieved from National Bureau of Economic Research website: http://www.nber.org.

10.

Card

DiNardo

J. E.

(2002) Skilled-biased technological change and rising wage inequality: Some problems and puzzles. Journal of Labor Economics 20: 733–783.

11.

Cha

Weeden

K. A.

(2014) Overwork and the slow convergence in the gender gap in wages. American Sociological Review 79: 457–484.

12.

Cohen

P. N.

Huffman

M. L.

(2007) Working for the woman? Female managers and the gender wage gap. American Sociological Review 72: 681–704.

13.

Cohen

P. N.

Huffman

M. L.

Knauer

(2009) ‘Stalled progress’: Gender segregation and wage inequality among managers, 1980-2000. Work and Occupations 36: 318–342.

14.

Cotter

D. A.

Hermsen

J. M.

Vanneman

(2004) Gender inequality at work, New York, NY: Russell Sage.

15.

Datta Gupta

Oaxaca

R. L.

Smith

(2006) Swimming upstream, floating downstream: Comparing women's relative wage progress in the United States and Denmark. Industrial and Labor Relations Review 59: 243–266.

16.

Duffy

(2005) Reproducing labor inequalities: Challenges for feminists conceptualizing care at the intersection of gender, race, and class. Gender & Society 19: 66–82.

17.

Duffy

(2007) Doing the dirty work: Gender, race, and reproductive labor in historical perspective. Gender & Society 21: 313–336.

18.

Duffy

(2011) Making care count: A century of gender, race, and paid care work, New Brunswick, NJ: Rutgers University Press.

19.

Dwyer

R. E.

(2013) The care economy? Gender, economic restructuring, and job polarization in the U.S. labor market. American Sociological Review 78: 390–416.

20.

England

(1992) Comparable worth: Theories and evidence, New York, NY: Aldine de Gruyter.

21.

England

(2010) The gender revolution: Uneven and stalled. Gender & Society 24: 149–166.

22.

England

Budig

Folbre

(2002) Wages of virtue: The relative pay of care work. Social Problems 49: 455–473.

23.

Folbre

(2001) The invisible heart: Economics and family values, New York, NY: New Press.

24.

Glenn

E. N.

(1992) From servitude to service work: Historical continuities in the racial division of paid reproductive labor. Signs: Journal of Women in Culture and Society 18: 1–43.

25.

Goldin

(1990) Understanding the gender gap: An economic history of American women, Oxford, England: Oxford University Press.

26.

Goldin

(2006) The quiet revolution that transformed women's employment, education, and family. American Economic Association: Papers and Proceedings, 96, 1–21. .

27.

Goldin

(2014) A grand gender convergence: Its last chapter. American Economic Review 104: 1091–1119.

28.

Goldstein

(2012) Revenge of the managers: Labor cost-cutting and the paradoxical resurgence of managerialism in the shareholder value era, 1984 to 2001. American Sociological Review 77: 268–294.

29.

Gordon

(1996) Fat and mean: The corporate squeeze of working Americans and the myth of managerial downsizing, New York, NY: Free Press.

30.

Granovetter

(1981) Toward a sociological theory of income differences. In: Berg

(ed.) Sociological perspectives on labor markets, New York, NY: Academic Press, pp. 11–47.

31.

Hirsch

B. T.

Schumacher

E. J.

(2004) Match bias in wage gap estimates due to earnings imputation. Journal of Labor Economics 22: 689–722.

32.

Jacobs

J. A.

(1992) Women's entry into management: Trends in earnings, authority, and values among salaried managers. Administrative Science Quarterly 37: 282–301.

33.

Jasso

Webster

Jr. (1999) Assessing the gender gap in just earnings and its underlying mechanisms. Social Psychology Quarterly 62: 367–380.

34.

Juhn

Murphy

K. M.

Pierce

(1991) Accounting for the slowdown in black-white wage convergence. In: Kosters

(ed.) Workers and their wages, Washington, DC: AEI Press, pp. 107–143.

35.

M. C.

(2011) When does gender matter? Gender differences in specialty choice among physicians. Work and Occupations 38: 221–262.

36.

Leicht

K. T.

(2008) Broken down by race and gender? Sociological explanations of new sources of earnings inequality. Annual Review of Sociology 34: 237–255.

37.

Levanon

Grusky

D. B.

(2016) The persistence of extreme gender segregation in the twenty-first century. American Journal of Sociology 122: 573–619.

38.

Liu

Grusky

D. B.

(2013) The payoff to skill in the third industrial revolution. American Journal of Sociology 118: 1330–1374.

39.

McCall

(2000) Explaining levels of within-group wage inequality in U.S. labor markets. Demography 37: 415–430.

40.

Meyer, P. B., & Osborne, A. M. (2005, September). Proposed category system for 1960-2000 census occupations (BLS Working Paper No. 383). Washington, DC: Bureau of Labor Statistics.

41.

Mincer

Polachek

(1974) Family investments in human capital. Journal of Political Economy 82: S76–S108.

42.

Mishel, L., Schmitt, J., & Shierholz, H. (January, 2013). Assessing the job polarization of growing wage inequality (Working Paper). Washington, DC: Economic Policy Institute.

43.

Morris

Western

(1999) Inequality in earnings at the close of the twentieth century. Annual Review of Sociology 25: 623–657.

44.

Mouw

Kalleberg

A. L.

(2010) Occupations and wage inequality in the United States. American Sociological Review 75: 402–431.

45.

O'Neill

(2003) The gender gap in wages, circa 2000. American Economic Review 93: 309–314.

46.

Petrescu-Prahova

Spiller

M. W.

(2016) Women's wage theft: Explaining gender differences in violations of wage and hour laws. Work and Occupations 43: 371–400.

47.

Polachek

(2012) A human capital account of the gender pay gap. In: Grusky

D. B.

Kricheli-Katz

(eds) The new gilded age, Stanford, CA: Stanford University Press, pp. 161–188.

48.

Shaumann

K. A.

(2010) Gender asymmetry in family migration: Occupational inequality or inter-spousal comparative advantage. Journal of Marriage and Family 72: 375–392.

49.

Sorenson

Dahl

M. S.

(2016) Geography, joint choices, and the reproduction of gender inequality. American Sociological Review 81: 900–920.

50.

Tenn

(2010) The relative importance of the husband's and wife's characteristics in family migration, 1960-2000. Journal of Population Economics 23: 1319–1337.

51.

Weeden

K. A.

(2002) Why do some occupations pay more than others? Social closure and earnings inequality in the United States. American Journal of Sociology 108: 55–101.

Decomposition components	Model 1		Model 2
Decomposition components		Percent change		Percent change
Change in differential	−.231		−.231
All observed characteristics	−.046	19.9%	−.064	27.7%
Nurturant care	−.006	2.6%	−.001	0.4%
Reproductive care	−.018	7.8%	−.008	3.5%
Managers	−.021	9.1%	−.012	5.2%
Professionals (noncare)	−.000	0.0%	−.000	0.0%
Overwork			.003	−1.3%
Part-time variables			−.010	4.3%
Potential experience variables			−.027	11.7%
Education variables			−.005	2.2%
Race variables			−.002	0.9%
Region variables			.001	−0.4%
Industry variables			−.003	1.3%
All observed prices	−.039	16.9%	−.031	13.4%
Nurturant care	−.046	19.9%	−.007	3.0%
Reproductive care	.000	0.0%	−.001	0.4%
Managers/professionals	.001	−0.4%	.000	0.0%
Professionals (noncare)	.006	−2.6%	.002	−0.9%
Overwork			.017	−7.4%
Part-time variables			.006	−2.6%
Potential experience variables			−.007	3.0%
Education variables			−.005	2.2%
Race variables			.002	−0.9%
Region variables			.000	0.0%
Industry variables			−.037	16.0%
Unexplained differential	−.146	63.2%	−.136	58.9%
Unobserved changes	−.173	74.9%	−.158	68.4%
Residual inequality effect	.028	−12.1%	.022	−9.5%
Unweighted N^a	312,493		311,853