Race and Social Problems
 Springer Science+Business Media, LLC 2011
College as an Investment: The Role of Graduation Rates in Changing Occupational Inequality by Race, Ethnicity, and Gender

Daniel H. KrymkowskiContact Information and Beth Mintz1

(1)  College of Arts and Sciences Dean’s Office, University of Vermont, 438 College St., Burlington, VT 05405, USA

Contact Information Daniel H. Krymkowski
Email: dkrymkow@uvm.edu

Published online: 24 March 2011

In this paper, we examine whether investments in higher education have contributed to changes in occupational inequality by focusing on the impact of college completion rates on movement into desirable occupations between 1983 and 2002. Since forces generating inequality vary by gender, race, and ethnicity, we examine trends for white, black, and Hispanic men and women in our study. Utilizing Ordinary Least Squares Regression on data from 20 Current Population Surveys, we find a modest decrease in both gender and racial inequality in access to desirable occupations and an increase in inequality between Hispanics and members of the other groups. College completion accounts for the progress made by white women and for the declines among Hispanic men. It does not explain changes for African Americans, either between men and women or when compared to whites.

Keywords  Occupational inequality – Higher education – Race – Ethnicity – Gender – Racial differences – Gender differences – Ethnic differences – Social stratification – United States – Social change – Occupational stratification – Inequality

Both the authors contributed equally to the paper.


In the last few decades, traditionally underrepresented groups have made substantial gains in higher education. Women, for example, are now more likely than men to earn college degrees, and this is true among African Americans, Hispanics, and whites (Bae et al. 2000; DiPrete and Buchmann 2006; England et al. 2007). Furthermore, the percentage increase in the postsecondary completion rates of black men and black, white, and Hispanic women have outpaced those of white men (US Bureau of the Census 2010).

This is good news: the groups that have made the most progress in this regard have typically earned less than white men, and a college degree may be an effective way to minimize such labor market disparities. In 2007, for example, college graduates earned about two times more than their high school counterparts (US Bureau of the Census 2010). Moreover, it is generally understood that education is the vehicle for upward mobility in our society (Blau and Duncan 1967) and that a college degree is a prerequisite for a middle-class life (Bennett and Lutz 2009).

The importance of educational attainment in explaining racial and ethnic differences in labor market outcomes is well documented, but the relationship between rising college graduation rates, in particular, and changes in labor market outcomes by race or ethnicity has not been extensively researched. Findings suggest that, on the general level, educational differences contribute both to racial disparities in earnings (Tomaskovic-Devey 1993; Jacobs and Blair-Loy 1996; Alon and Tienda 2005) and to the employment gap between white women, on the one hand, and Hispanic and black women on the other (England et al. 2004). Differences in education and cognitive skills help explain wage differentials between whites and Latinos (England et al. 1999), while educational attainment contributes both to wage disparities between Latinas and white women (Browne and Askew 2005) and to authority gaps between Hispanic and white men (Elliott and Smith 2004). The little research available on college degrees suggests that although earning differentials are attributable in part to educational differences, labor market disparities actually grow at higher levels of education. Among male college graduates, for example, life-course wage trajectories are visibly steeper for whites than for African Americans or Hispanics (Tomaskovic-Devey et al. 2005).

For gender, the literature has demonstrated that educational differences play a minimal role in workplace inequality (Tam 1997; Okamoto and England 1999), but the specific effect of higher education is not well understood. A small body of research has found that college has helped narrow the gender gap in earnings in the early career (Loury 1997; Gill and Leigh 2000). However, the wages of male college graduates remain higher than those of women (Averett and Burton 1996; Charles and Luoh 2003; Evertsson et al. 2009; Bobbitt-Zeher 2007), and this may limit the potential for education as an equalizer.1

We know, then, that educational attainment works differently for race or ethnicity than for gender. It is unclear, however, if this same pattern is true for the more specific case of a college degree, even though this may have important consequences for economic equality. Furthermore, we know little about the effect of a college education on occupational attainment, an important issue given that occupation remains a critical dimension of social stratification (Weeden and Grusky 2005).

In this paper, we address these issues by analyzing the relationship between increasing college completion rates of white women and minorities and movement into attractive occupations, examining the time period between 1983 and 2002. We use human capital theory as a framework for studying these relationships since its argument that individuals invest in skill development with the expectation of attractive returns in job and wage prospects (Becker 1994) is commonly offered to explain why groups might pursue educational opportunities. Using individual-level data on occupational earnings, authority, and prestige as indicators of occupational desirability, we include white, black, and Hispanic men and women in our study.2

More specifically, in order to assess the impact of the increased college graduation rates that underrepresented groups have experienced in recent years, we examine changes over time in occupational outcomes by: (1) race and ethnicity among men, on the one hand, and women, on the other; and (2) gender within the three racial and ethnic groups under investigation. In doing so, we contribute to the literature on the role of education in decreasing inequality by comparing the effectiveness of recent increases in human capital investments in college degrees as a vehicle for accessing better jobs for various racial, ethnic, and gender groups. In addition, by focusing on occupations, a crucial mechanism for allocating resources within labor markets, we assess the impact of college completion rates in ways that cannot be measured by individual earnings (Semyonov and Lewin-Epstein 2009).

Human Capital Theory, Education, and Occupational Attainment

The neoclassical economics theory of human capital has been used extensively in studies of labor force participation to explain why individuals might invest in their careers, arguing that workers pursue skill acquisition and training when they anticipate a positive return (financial or otherwise) on that investment (Becker 1994). The literature has identified three variables of particular relevance in this perspective: formal education, on-the-job-training, and years of job experience (England 1992). Of the three, education occupies a particularly important place in the popular imagination as a vehicle for upward mobility. Indeed, human capital theory is the theoretical foundation for the ideological assumption that everyone should strive for a college degree.

Human capital theory recognizes that decisions about labor market investments may be gendered suggesting, for example, that women who anticipate that family responsibilities will undermine employment gains may under invest in human capital development (Tomaskovic-Devey 1993). This does not necessarily affect decisions to attend college, however, since the perspective assumes that potential students weigh the costs of tuition and lost wages against either the promise of higher lifetime earning capacity or other types of benefits. It is clear that, in the aggregate, college graduates earn more than their high school counterparts, although the specific advantages of a higher education are not well understood. The little research available in this area suggests that a college degree produces a number of positive yields for women, including higher wages, increased chances of marriage, a higher standard of living, and insurance against poverty (DiPrete and Buchmann 2006). With the exception of earnings, women seem to have benefitted more on these dimensions than their male counterparts.

Earnings remain important, however, especially since women are likely to continue to work after having children. In 2005, 62.9% of college-educated women with infants remained in the labor force (Cohany and Sok 2007). Despite this continued commitment to employment, a gender gap in wages persists among college graduates, and it includes women who apparently have not under invested in education. Bobbitt-Zeher (2007) found that after controlling for relevant variables, including undergraduate major, young female college graduates earned $4,400 less per year than comparably credentialed men. Similarly, Shauman (2006) found that even with the same majors, female college graduates tended to enter less lucrative occupations.

Comparable information is not available by race or ethnicity, but research on educational attainment, more generally, suggests that the importance of education varies by race, ethnicity, and gender. In comparing African Americans and whites, for example, England et al. (1999) found that level of education accounts for only about half of the wage gap between men and around three-quarters for women. In addition, education accounts for a smaller proportion of the gender wage gap among African Americans than between black and white women or between black women and white men (England 1992). Note, however, that black and white women both pay a price for being female when it comes to returns on educational attainment (McGuire and Reskin 1993).

Other work has found that the relationship between educational attainment and layoffs varies by race (Wilson and McBrier 2005).3 The little research available on higher education suggests that a postgraduate degree insulates white women in white collar occupations from downward mobility more than American or Hispanic women (Wilson 2009).

We know less still about the role of higher education in gender, ethnic, and racial differences in occupational attainment, even though occupations remain fundamental mechanisms for allocating resources in labor markets. Earlier research documented trends toward increased inter-group equality in occupational status, although the rate of change seemed to be greater in the 1960s and 1970s than in the 1980s (see Featherman and Hauser 1978; DiPrete and Grusky 1990). Moreover, there is evidence that declines in educational inequality were responsible for some of this progress. For instance, most of the observed reduction between 1962 and 1973 in the occupational status differential between African American and white men was due to decreases in racial inequality in schooling (Featherman and Hauser 1978). However, neither the role of college degrees nor any gender differences were considered. These issues have become particularly important in recent years with the marked increase in college completion rates of underrepresented groups and our understanding that gendered outcomes vary by race and ethnicity.

In summary, we know that education, a classic example of a human capital investment, works differently by gender than by race or ethnicity, but we do not know specifically how in the case of higher education. We know less still about racial differences within genders and gender differences within racial and ethnic groups. And the relationship between college completion and occupational attainment by race/ethnicity and gender has not been studied.

Finally, it is clear that rewards for educational investments may take multiple forms (DiPrete and Buchmann 2006). For this reason, we use three variables to model occupational attractiveness—occupational earnings, occupational authority, and occupational prestige—each of which captures different dimensions of what desirable occupations may provide.

Research has emphasized the monetary rewards of labor force participation, and occupational earnings is a straightforward way of modeling this. A well developed body of literature has also identified workplace authority (e.g., the extent of a person’s supervisory responsibility) as an important dimension of social stratification. Smith (2002:511) underscored its significance when he noted that authority is “a highly coveted workplace resource,” while Wright et al. (1995) pointed out that it is a valued job attribute that confers status. Workplace authority may be accompanied by additional financial compensation, but it also often carries nonmonetary rewards of job satisfaction and job autonomy (Wilson 1997). Thus, authority carries with it valued workplace attributes that complement more traditional measures of financial remuneration.

Occupational prestige is well understood as a symbolic reward: Weber emphasized the importance of prestige in his concept of social honor, suggesting that the shared lifestyles and intra-group recognition among status group members is highly consequential (Gerth and Mills 1958). Zhou (2005) suggests it measures collective beliefs about the legitimacy of an occupation and is related to, but distinct from, other rewards, making it a “status ordering phenomenon” (p. 94). In these ways, our three measures of occupational attractiveness tap into different dimensions of occupational attainment, allowing us to examine the importance of educational investment by race, ethnicity, and gender with respect to a number of desirable outcomes.

Analytic Strategy and Hypotheses

We begin by examining college graduation rates at two points in time, 1983 and 2002, calculating differences in the probability of completing college for pairs of racial, ethnic, and gender groups in both years. For example, we ask whether differences in completion rates of white men and white women changed over time, and whether any identified change was statistically significant. We do this for African Americans and Hispanics as well. In the same way, we examine racial or ethnic differences among men and among women comparing, for example, changes over time in differences in graduation rates between white and African American men or between African American and Hispanic women. We use these data to formulate hypotheses on changing inter-group inequality patterns that are consistent with the assumptions of human capital theory.

Table 1 presents data taken from the 1983 and 2002 Current Population Surveys (King et al. 2010) summarizing recent trends in educational attainment, which provide the foundation for our hypotheses. Every group under investigation increased its college completion rates; however, as a percentage of the baseline figures, men lagged behind women and whites lagged behind African Americans. The percentage of white men who earned bachelor’s degrees rose from about 26 to 33% between 1983 and 2002, but for white women, the figure grew from 20 to 32%. African American men and women experienced increases from 11 to 18% and 12 to 19%, respectively, with Hispanic completion rates rising from 8 to 10% for men and 9 to 13% for women. Note that the gender differences in these rates were markedly smaller among minorities than whites, which reflects why college campuses are feeling the impact of rising attendance rates of white women, in particular.
Table 1 College graduation rates by gender, race/ethnicity, and year from current population survey data (respondents in labor force)




Whites (%)

African Americans (%)

Hispanics (%)



















These relationships are summarized by the statistics presented in Table 2, which indicate the differences between pairs of groups in college completion rates at the beginning and end of the time period under investigation. When the change between groups in a pair is statistically significant, we hypothesize a corresponding change in the proportion of each group in attractive occupations. The difference in college completion rates between white men and white women, for example, was 6.46% in 1983, declining to .83% by 2002. This decrease is statistically significant, so assuming that women’s investment in higher education will be reflected in movement into attractive occupations, we predict that occupational inequality between these groups has decreased over time.
Table 2 College completion rates and human capital hypotheses by race, ethnicity, and gender

Comparison groups

Trends in differentials in college completion rates

Difference in rates in 1983 (%)

Difference in rates in 2002 (%)

Significant change?

Inequality prediction

White men and white women





African American men and African American women




No change

Hispanic men and Hispanic women




No change

White men and African American men




No change

White men and Hispanic men





African American men and Hispanic men





White women and African American women





White women and Hispanic women





African American women and Hispanic women




No change

Since gender differences in college graduation rates among African Americans and Hispanics did not change over time, we predict no changes in occupational inequality in these groups by gender. Similarly, we predict no changes in access to desirable occupations when comparing white to African American men or when comparing African American to Hispanic women. We do hypothesize, however, that differences in access to desirable occupations between Hispanic and other men and between white women and women of color will increase over time.

Note that increasing equality in higher education does not necessarily imply a reduction in occupational inequality; that is why we view these predictions as hypotheses. Whether such a reduction actually occurs depends on two factors: (1) the exact proportions of college graduates in the groups being compared; (2) the size of the college “premium” in terms of occupational attainment for the groups.

Although we focus on the impact of college completion, it is important to note that there have been changes in inter-group differences at other levels of education as well. We address this issue below.

Data and Methods

We use data on individuals, drawn primarily from all twenty March Current Population Surveys (CPS) between 1983 and 2002 (King et al. 2010). This time frame is attractive for two reasons. First, the occupational classification schemes employed by the Census Bureau in 1980 and 1990 are virtually identical. We can thus be sure that any trends we discover are not artifacts of changed occupational classification schemes.4 Second, this time period saw statistically significant changes in college completion rates for more than half of the race/ethnic and gender pairs under investigation and is, therefore, very well suited to test our hypotheses about the impact of recent changes in this form of educational attainment. To prevent the inclusion of the same respondents from surveys of contiguous years, we utilize outgoing-rotation respondents only. Since the unit of analysis is the individual, we weight the data using the CPS variable “perwt.” The total sample size after listwise deletion of missing data is 684,928.

We use occupational earnings to evaluate the monetary aspect of occupational attainment, which are computed using aggregated individual-level data from the 1990 Decennial Census; this year represents the approximate mid-point of our 1983–2002 analytical period.5 For each detailed occupation, we convert the mean earnings of all wage-earning workers to constant 2007 dollars. The authority level of an occupation is measured using data from the 1989–2006, pooled General Social Survey (GSS). For individuals in the GSS, authority is measured using a three-point scale: 2 denotes someone who supervises but is not supervised, 1 refers to someone who supervises and is supervised, and 0 indicates an individual who does not supervise anyone. We compute the mean authority level for each detailed occupation in the GSS by aggregating these data on individuals. To measure occupational prestige, we utilize the Nakao–Treas Occupational Prestige Scores (Nakao and Treas 1994). These scores were constructed using data from the 1989 GSS and were meant to provide an update to the older Hodge-Siegel-Rossi scores from the 1960s. In addition, they were developed specifically for the 1980 Census Occupational Classification.

College completion is measured using a dummy variable. Between 1983 and 1991 the CPS recorded individuals’ years of primary school, secondary school, and college completed; between 1992 and 2002 it recorded years completed up to grade 12 and the highest credential received beyond that. Thus, for the 1983–1991 data, we code respondents who have completed four or more years of college as having graduated from college, and for the 1992–2002 data, we code respondents with a Bachelor’s Degree or more as having graduated from college.6

We also use several variables as controls. To make sure that we are not tapping other dimensions of human capital theory, we include job experience, experience squared, marital status, the number of children at home, and the number of children at home under 5 years of age in our analysis. In addition, we consider occupational growth, region, and whether the respondent works in the public sector, which are often used variables in this type of analysis. Growth is measured by the ratio of the number of incumbents in an occupation in the 1980 and 1990 Decennial Censuses. Job experience is computed using the well-known proxy, age-years of education-6, and marital status is a dummy variable coded married or not. Region refers to the respondent’s Census region of residence from the Current Population Survey using the following categories: New England, Middle Atlantic, South Atlantic, East North Central, West North Central, East South Central, West South Central, Mountain, and Pacific.

We begin our analysis by presenting some descriptive information on the following: (1) the 1983 baseline values occupational income, authority, and prestige for each group; and (2) how these values have changed over time. Then, using Ordinary Least Squares, we regress each occupational characteristic on survey year, a dummy variable for each of nine intergroup comparisons, and the interaction effect between this dummy variable and survey year to identify trends in group differences over time. The comparisons include a gender contrast within each of the three race/ethnic groups and three racial/ethnic contrasts within each gender. In the gender analyses, the dummy variables are coded 1 = male/0 = female. For the race/ethnic analyses, they are coded 1 = white/0 = African American, 1 = white/0 = Hispanic, and 1 = African American/0 = Hispanic.7 We then add the control variables, followed by a dummy variable for college completion to evaluate the extent to which changes in college graduation rates explain trends in intergroup differences. To test whether gains at levels of education other than college explain any changes that we find, we include additional dummy variables for the completion of some college, a high school diploma, and some high school. We discuss these results only when the additional dummy variables explain a statistically significant portion of the change in inequality.

We present mean values of the occupational characteristics that we use to measure attractiveness in our baseline year of 1983 in Table 3. Historic, inter-group differences on these dimensions are well known in the stratification literature, but it is interesting to note how little within gender differentiation there was in access to attractive occupations between African Americans and Hispanics. Note also that there were only small gender differences in occupational prestige within racial and ethnic groups.
Table 3 Occupational characteristics by race, ethnicity, and gender 1983

White men

White women

African American men

African American women

Hispanic men

Hispanic women






















Figure 1 displays trend lines for occupational income, authority, and prestige for our six groups, graphically illustrating the changes over time that we are examining. These lines come from within-group regressions of each occupational characteristic on survey year and, in the analyses that follow, we ask whether changes in intergroup college completion rates can explain the changes captured in these figures. For comparative purposes, we also include an average trend line, which is the unweighted mean of the trends for the groups. Using Cohen and Cohen’s (1983) effect coding procedure, we performed statistical tests for the difference between the average trend and the trend for each group; all of these differences were statistically significant at the .01 level. We also performed a global test for the interaction between group and survey year by comparing R-Squared with and without this interaction. This test was statistically significant at the .001 level for each of the three occupational characteristics.
Fig. 1 Trends in occupational characteristics by group. Notes: a Slopes of all trend lines are statistically different from the average trend at the .01 level. The average trend is the unweighted mean of the trends for the six groups. R-Squared without all interactions between group and time is .088. R-Squared with the interactions is .089. The R-Squared change is statistically significant at the .001 level. b Slopes of all trend lines are statistically different from the average trend at the .01 level. The average trend is the unweighted mean of the trends for the six groups. R-Squared without all interactions between group and time is .0395. R-Squared with the interactions is .0400. The R-Squared change is statistically significant at the .001 level. c Slopes of all trend lines are statistically different from the average trend at the .01 level. The average trend is the unweighted mean of the trends for the six groups. R-Squared without all interactions between group and time is .0339. R-Squared with the interactions is .0343. The R-Squared change is statistically significant at the .001 level (Color figure online)

Although the increments to R-Squared from the interactions are not large there are nonetheless some interesting differences in these trends. For occupational income, the slope of the lines for both Hispanic men and women are noticeably lower than the average, indicating less growth than other groups over time. In contrast, the slope for African American women was greater than the overall trend, demonstrating more progress for this group.

For occupational authority, white and African American women experienced the most progress while African American men saw less than average improvement. And for occupational prestige, white and African American women again enjoyed the most improvement over time, with white women overtaking white men by 2002. However, in this case, as with occupational income, Hispanic men and women experienced very little growth over time.

In order to explain the role of college graduation rates in these and other changes, we examine nine different intergroup comparisons in detail. Table 4 presents trends in gender differences in access to desirable occupations by race and ethnicity for the period 1983–2002.8 The top panel indicates the magnitude of the gross changes over time and whether they are statistically significant. In the case of occupational earnings among African Americans, for example, the negative coefficient indicates that the gender advantage for men has declined by about $80 per year.
Table 4 Gender differences in access to desirable occupations: average yearly change from 1983 to 2002


African Americans


Gross trend

Occupational earnings




Occupational authority




Occupational prestige




Trend with control variables

Occupational earnings




Occupational authority




Occupational prestige




Trend with control variables and college

Occupational earnings




Occupational authority




Occupational prestige




Difference between trend w/controls and trend controlling for college significant at .05?

Occupational income



Yes (@.10)

Occupational authority




Occupational prestige




+ p < .01; * p < .05
aConsistent with predictions based on human capital theory

We see that women made progress vis--vis men in all groups.9 For occupational earnings, white women narrowed the gap with their male counterparts by about $144 per year, amounting to a decline of $2,730 over the 19-year period and representing 24% of the initial difference reported in Table 3. The gender gap declined by $1,526 (23%) among African Americans and $1,682 (33%) for Hispanics during this period. Occupational authority gaps decreased more markedly: 39%, 62%, and 39% among whites, African Americans, and Hispanics, respectively. For occupational prestige, on the other hand, women in all groups enjoyed small advantages over men in 1983, and these differences increased slightly over time for whites and African Americans.

In the second panel of the table, we present trends in gender differences once the control variables have been taken into account. These results are similar to those just presented, except in the case of whites: the control variables here account for about one-third of the decline in the gap in occupational earnings. The reason for this is twofold. First, women were more likely than men to move into occupations that were growing. Second, the proportion of men who were married decreased over time, and marriage is known to be positively associated with social status, especially among men.

The third panel of the table examines the role of educational attainment in these results by adding college completion to the control variables, while the bottom panel indicates the following: (1) whether the changes in the coefficients, after taking college credentialing into account are statistically significant; and (2) if the changes are consistent with predictions based on human capital theory.

Results indicate that higher education matters for white women: college graduation rates contribute to the narrowing of the gender gap in occupational earnings, authority, and prestige. However, college completion rates are most important with respect to prestigious occupations. In percentage terms, graduation rates account for about 96% of the small increase in the gender gap in prestige over time, compared to 37% for earnings and about 22% for authority.10 Thus, although the striking increase in college graduation rates of white women has reduced gender inequalities to some extent, its effect on the occupational earnings and authority gaps has been rather modest.

For Latinas, higher college completion rates contribute to increased equality in occupational prestige, and if accepting a significance level of p < .10, to increased equality in earnings to as well. For African Americans, women made progress, as predicted, when compared to men in occupational earnings, authority, and prestige, but this success is not related to disproportionate increases in college completion rates. Thus, for Latinas, the impact of a college degree is mixed, but African American women’s progress is not due to investments in higher education.

In only one situation did the addition of dummy variables for educational levels other than college completion explain a statistically significant portion of the decline in gender inequality: when “some college” was taken into account for the white male–white female prestige comparison.

Table 5 examines changes over time in access to desirable occupations by race and ethnicity for men and for women. Looking at gross trends, we find that African American men narrowed the gaps with their white counterparts in occupational earnings and prestige; white and African American men and women increased their advantages over Hispanics in all three characteristics; and African American women narrowed the occupational prestige gap with white women. Comparing African American and white men, for example, changes in earnings and prestige were about $54 and three-fourths of a point per year. This amounts to a modest $1,029 (11%) and 1.4 prestige points during the 1983–2002 period. Among their female counterparts, the annual decline in the racial prestige difference was less than two-thirds of a point (the change in the occupational earnings differential between African American and white women was not statistically significant).
Table 5 Ethnic/racial differences in access to desirable occupations: average yearly change from 1983 to 2002



Whites vs. African American

Whites vs. Hispanics

African American vs. Hispanics

Whites vs. African American

Whites vs. Hispanics

African American vs. Hispanics

Gross trend

Occupational earnings







Occupational authority







Occupational prestige







Trend with control variables

Occupational income







Occupational authority







Occupational prestige







Trend with control variables and college

Occupational income







Occupational authority







Occupational prestige







Difference between trend w/controls and trend controlling for college significant at .05?

Occupational earnings







Occupational authority


Yes (@.06)a

Yes (@08)a

Yes (@.11)a

Yes (@.07)a


Occupational prestige







p < .05; + p < .01
aConsistent with predictions based on human capital theory

The occupational earnings gap between white and Hispanic men increased by about $136 per year. This represents $2,582 over the 19-year period or an increase of approximately 25%. The corresponding increases in differences in occupational authority and prestige were rather small, however. For Latinas, the growth in the earnings gap was larger, about $191, or about $3,629 over the period (a 93% increase). The authority and prestige differences grew as well, by about .04 and 2.1, respectively.

Of particular interest is the disparity that emerged between African Americans and Hispanics. In 1983, Hispanic men and women had roughly equal access to desirable occupations when compared to their African American counterparts (see Table 3), but this situation changed by 2002. For example, the gap in occupational earnings grew by $190 per year for men and $182 per year for women to about $3,500 for each gender. The differences in occupational authority and prestige grew as well. In the case of occupational prestige, the ethnic gap increased by about 3 points over time among both men and women.

As in Table 4, the third panel presents results controlling for college completion rates, while the bottom section indicates whether these changes are (1) statistically significant and (2) consistent with human capital theory predictions.11 For men, a college degree is not responsible for the modest decreases that we found in racial occupational inequality. Since racial differences in college graduation rates did not change over time, this is not surprising. For women, we see an unexpected decrease in the racial gap in occupational prestige and after taking higher education into account, the gap decreased even further. College degrees are not responsible for the gross decline, however, since racial inequality in education in the time period under investigation increased between white and African American women. Furthermore, the result when controlling for college completion indicates that if white women had not disproportionately increased their college completion rates, the decline in inequality would have been larger still.

Occupational earnings worked a little differently, but with the same result: we found no change in the occupational earnings gap between African American and European American women. However, when we controlled for college, the original gap unexpectedly declined. Thus, white women’s increasing advantage in college graduation rates is masking within-educational level declines in the earnings gap, suggesting that, again, something other than higher education is driving these results. We see the same pattern for occupational authority (at p < .11, at least).

For Hispanic men, lack of college degrees contributes to the growing difference in access to attractive occupations when compared to both blacks and whites, as predicted by human capital theory. Thus, higher education has been more important for Hispanic men than for their African Americans counterparts.

When we compare Latinas and white women, our results are consistent with the statistics presented in Table 2: we see increasing gaps in access to attractive occupations on all three measures. Latinas also increased the gap with African American women, but educational differences as measured by college completion rates were not responsible for this. Here too, then, African American women made progress that was not driven by educational advances.

In only one instance did the addition of dummy variables for educational levels other than college completion explain a statistically significant portion of the decline in ethnic or racial inequality: when “some college” was taken into account in the African American/Hispanic prestige comparison among men.


In this paper, we examined the impact of college graduation on access to desirable occupations by race, ethnicity, and gender. We were particularly interested in whether human capital theory’s emphasis on education as an investment would predict changes in access to attractive occupations when we use college completion as a measure of educational attainment. We have found that higher education decreased the gender gap among whites and helped us understand the growing difference between whites and Hispanics. It was not useful in explaining changes for African Americans, either when compared to whites or with regard to the narrowing of the gender gap. From these results, we conclude that human capital theory is resilient because it explains some very important occupational outcomes. Its weakness is that it has not theorized differential outcomes by race, gender, and ethnicity and is therefore unable to explain racialized and gendered differences in the way that college graduation contributes to occupational attainment.

Gender differences among whites in rates of college completion nearly disappeared over time, but this led only to modest decreases in occupational earnings inequality. On the other hand, it is well known that men and women often pursue different college majors, and this contributes to earnings discrepancies (Bobbitt-Zeher 2007; Bradley 2000). Although we do not have data to test the impact of major on movement into desirable occupations, recent research demonstrates that even among college graduates with the same major, women are more likely than men to enter stereotypically female jobs; they are also attracted to occupations with flexible work schedules, opportunities for intrinsic rewards, and the need for social and\or caring skills (Shauman 2006). Our findings are consistent with this research and underscore the limitations of education, alone, as a vehicle for changing the gendered nature of earnings differentials.

Monetary rewards are not the only measure of occupational attractiveness, however. College investment among white women leads to declines in the prestige and authority gaps with white men. Earlier research has found that college education leads to a higher probability of marriage (DiPrete and Buchmann 2006), and we wonder if the increases in occupational prestige and authority for women with degrees might also translate into more attractive options within marriage markets. Thus, we suggest that rewards for college completion for white women in the form of increased occupational prestige and authority may be very worthwhile.

We have also found that, over time, occupational inequality between African Americans and whites declined. For men, our results point to increased equality in occupational earnings and prestige. African American women also experienced gains when compared to white women, but only in terms of occupational prestige. Importantly, these results are not explained by changes in college graduation rates. There was essentially no change over time in racial differences in college completion for men, and white women increased their educational advantage over African American women. This suggests that African American men and women are benefitting from other processes.

What might be responsible for the increased equality between African Americans and whites? One possibility is that educational gains at levels other than college may explain the changes. But, as mentioned earlier, we tested this possibility by adding dummy variables for the completion of some college, a high school diploma, or some high school. In no case did these additional dummy variables explain a statistically significant portion of the decline in inequality.

Another contributing factor to this progress may be state action. Recent research has begun to examine the impact of state policy on processes of stratification, and this work has found that public policies can have important implications for labor force participation, both on the national (Western and Pettit 2005) and international levels (Pettit and Hook 2005). Many have speculated, for example, that federal policies banning segregation in the work place have played an important role in countering the discrimination that contributed to occupational segregation (Reskin 1993). Moreover, empirical research has found that laws flowing from the Civil Rights Act of 1964 have, indeed, had positive effects on the employment outcomes of white women and people of color (DiPrete and Grusky 1990; Stainback et al. 2005).

A number of studies, however, have demonstrated that the largest declines in racial occupational segregation occurred in the mid-to-late 1960s, during and in the immediate aftermath of the Civil Rights Movement (Featherman and Hauser 1978; Hout 1984) but before government policies had any real legal authority. Moreover, Stainback et al. (2005) suggest that over the past 25 years or so, policy changes and budget reductions in the Equal Employment Opportunity Commission have limited the agency’s ability to monitor employment practices. The progress prior to formalized legislation, combined with the weakened enforcement in recent years, makes us question whether Affirmative Action policies were strong enough to account for the changes that we found.

A third possibility is changes in discrimination patterns. Many argue that despite the decline of the most overt forms of discrimination, more subtle forms of differential treatment in the labor market remain (Bertrand and Mullainathan 2004) and that discrimination, more broadly, helps explain both racial and gender differences in occupational outcomes (Huffman and Cohen 2004a, b; McBrier and Wilson 2004; Petersen and Saporta 2004; Reid and Padavic 2005). Recent research, however, has found changing racial attitudes among whites particularly in relation to stereotyping, and this may be accompanied by changes in behavior. Between 1990 and 2006, for example, the percentage of whites surveyed who believed that whites work harder than blacks declined from about 66 to 43%, and the percentage expressing beliefs that blacks are less intelligent than whites dropped from 57% to about 25% (Krysan 2008). This raises two questions: first, do these changes reflect fundamental change rather than sensitivity to what is acceptable in a social situation (Krysan 2008) and, second, following Reskin’s (2003) lead, what are the mechanisms through which attitudes might be translated into outcomes? This is clearly a fruitful area for future research.

Our findings for Hispanics follow a different pattern. For men and women, college graduation rates explain much of the gap with whites in access to attractive occupations, suggesting that higher education is more consequential for Hispanics than African Americans. This is consistent with earlier research that has found a racioethnic hierarchy in access to workplace authority (Smith 2001).

We also note that at the beginning of our analytical period, Hispanic men and women and their African American counterparts displayed similar rates in access to desirable occupations. By the end of the period under investigation, however, Hispanics had fallen behind in occupational authority, earnings, and prestige. Given our analysis, we believe that this disparity is the product of increased immigration rates of Hispanics of relatively low socioeconomic status. Moreover, our findings suggest that these gaps will continue to widen.

Finally, although we have treated college completion as nonproblematic, we take seriously Tomaskovic-Devey et al.’s (2005) point that educational acquisition is itself a social process, rather than an isolated individual investment choice. Thus, the limitations of human capital theory in understanding the racial and gender components of labor market outcomes are exacerbated by the underlying processes that generate differential college attendance itself.


Alon, S., & Tienda, M. (2005). Job mobility and early career wage growth of white, African-American, and Hispanic women. Social Science Quarterly, 86, 1196–1217.
Averett, S., & Burton, M. (1996). College attendance and the college wage premium: Differences by gender. Economics of Education Review, 15, 37–49.
Bae, Y., Choy, S., Giddens, C., Sable, J., & Snyder, T. (2000). Trends in educational equity of girls and women. US Department of Education, National Center for Educational Statistics. Washington, DC: US Government Printing Office.
Becker, G. (1994). Human capital. Chicago: University of Chicago Press.
Bennett, P., & Lutz, A. (2009). How African American is the net black advantage? Differences in college attendance among immigrant blacks, native blacks, and whites. Sociology of Education, 82, 70–100.
Bertrand, M., & Mullainathan, S. (2004). Are Emily and Greg more employable than Lakisha and Jamal? A field experiment on labor market discrimination. American Economic Review, 94, 991–1013.
Blau, P., & Duncan, O. D. (1967). The American occupational structure. New York: Wiley.
Bobbitt-Zeher, D. (2007). The gender income gap and the role of education. Sociology of Education, 80, 1–22.
Bradley, K. (2000). The incorporation of women into higher education: Paradoxical outcomes. Sociology of Education, 73, 1–18.
Browne, I., & Askew, R. (2005). Race, ethnicity and wage inequality among women: What happened in the 1990s and early 21st century? American Behavioral Scientist, 20, 1275–1292.
Charles, K., & Luoh, M.-C. (2003). Gender differences in completed schooling. Review of Economics and Statistics, 85, 559–577.
Cohany, S., & Sok, E. (2007). Trends in labor force participation of married mothers of infants. Monthly Labor Review. www.bls.gov/opub/mlr/2007/02/art2full.pdf.
Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum.
DiPrete, T., & Buchmann, C. (2006). Gender-specific trends in the value of education and the emerging gender gap in college completion. Demography, 43, 1–24.
DiPrete, T., & Grusky, D. B. (1990). Structure and trend in the process of stratification for American men and women. American Journal of Sociology, 96, 107–143.
Elliot, J., & Smith, R. (2004). Race, gender, and workplace power. American Sociological Review, 69, 365–386.
England, P. (1992). Comparable worth: Theories and evidence. New York: Aldine de Gruyter.
England, P., Allison, P., Li, S., Mark, N., Thompson, J., Budig, M., et al. (2007). Why are some academic fields tipping toward female? The sex composition of US fields of doctoral degree receipt, 1971–2002. Sociology of Education, 80, 23–42.
England, P., Christopher, K., & Reid, L. (1999). Gender, race, ethnicity, and wages. In I. Browne (Ed.), Latinas and African-American women at work: Race, gender, and economic inequality. New York: Russell Sage Foundation.
England, P., Garcia-Beaulieu, C., & Ross, M. (2004). Women’s employment among blacks, whites, and three groups of Latinas: Do more privileged women have higher employment? Gender & Society, 18, 494–509.
Evertsson, M., England, P., Reci, I., Hermsen, J., de Bruijn, J., & Cotter, D. (2009). Is gender inequality greater at lower or higher educational levels? Common patterns in the Netherlands, Sweden and the US. Social Politics, 16, 210–241.
Featherman, D. L., & Hauser, R. M. (1978). Opportunity and change. New York: Academic Press.
Gerth, H. H., & Wright Mills, C. (1958). From Max Weber: Essays in sociology. New York: Oxford University Press.
Gill, A., & Leigh, D. (2000). Community college enrollment, college major, and the gender wage gap. Industrial and Labor Relations Review, 54, 163–181.
Hout, M. (1984). The occupational mobility of black men: 1962–1973. American Sociological Review, 49, 308–322.
Huffman, M., & Cohen, P. N. (2004a). Occupational segregation and the gender gap in workplace authority: National versus local labor markets. Sociological Forum, 19, 121–147.
Huffman, M., & Cohen, P. N. (2004b). Racial wage inequality: Job segregation and devaluation across US labor markets. American Journal of Sociology, 109, 902–936.
Jacobs, J. A., & Blair-Loy, M. (1996). Gender, race, local labor markets and occupational devaluation. Sociological Forum, 29(3), 209–230.
King, M., Ruggles, S., Trent Alexander, J., Flood, S., Genadek, K., Schroeder, M. B., Trampe, B., & Vick, R. (2010). Integrated public use microdata series, current population survey: Version 3.0 [machine-readable database]. Minneapolis: University of Minnesota.
Krysan, N. (2008). Data update to racial attitudes in America. An update and website to complement, Racial attitudes in America: Trends and interpretations, Revised Edition. Howard Schuman, Charlotte Steeh, Lawrence Nono, and Maria Krysan, 1997, Harvard University Press.
Loury, L. (1997). The gender earnings gap among college-educated workers. Industrial and Labor Relations Review, 50, 580–593.
McBrier, D., & Wilson, G. (2004). Going down? Race and downward occupational mobility for white-collar workers in the 1990s. Work and Occupations, 31, 283–322.
McGuire, G., & Reskin, B. (1993). Authority hierarchies at work: The impacts of race and sex. Gender and Society, 7, 487–506.
Nakao, K., & Treas, J. (1994). Updating occupational prestige and socioeconomic scores: How the new measures measure up. Sociological Methodology, 24, 1–72.
Okamoto, D., & England, P. (1999). Is there a supply side to occupational sex segregation? Sociological Perspectives, 42, 557–582.
Petersen, T., & Saporta, I. (2004). The opportunity structure for discrimination. American Journal of Sociology, 109, 852–901.
Pettit, B., & Hook, J. (2005). The structure of women’s employment in comparative perspective. Social Forces, 84, 779–801.
Reid, L., & Padavic, I. (2005). Employment exits and the race gap in young women’s employment. Social Science Quarterly, 86, 1242–1260.
Reskin, B. (1993). Gender segregation in the workplace. Annual Review of Sociology, 19, 241–270.
Reskin, B. (2003). Including mechanisms in our models of ascriptive inequality. American Sociological Review, 68, 1–21.
Semyonov, M., & Lewin-Epstein, N. (2009). The declining racial earnings gap in the united states: multi-level analysis of males’ earnings, 1960–2000. Social Science Research, 38, 296–311.
Shauman, K. A. (2006). Occupational sex segregation and the earnings of occupations: What causes the link among college-educated workers. Social Science Research, 35, 577–619.
Smith, R. A. (2001). Particularism in control over monetary resources at work: An analysis of racioethnic differences in the authority outcomes of black, white, and Latino men. Work and Occupations, 28, 447–468.
Smith, R. A. (2002). Race, gender, and authority in the workplace. Annual Review of Sociology, 28, 509–542.
Stainback, K., Robinson, C., & Tomaskovic-Devey, D. (2005). Race and workplace integration: A politically mediated process? American Behavioral Scientist, 48, 1200–1228.
Tam, T. (1997). Sex segregation and occupational gender inequality in the United States: Devaluation or specialized training? American Journal of Sociology, 102, 1652–1692.
Tomaskovic-Devey, D. (1993). Gender and racial inequality at work: The Sources and consequences of job segregation. Ithaca, NY: ILR Press.
Tomaskovic-Devey, D., Thomsa, M., & Johnson, K. (2005). Race and accumulation of human capital across the career: A theoretical model and fixed-effects application. American Journal of Sociology, 111, 58–89.
US Census Bureau. (2005, 2010). Statistical abstract of the United States. Washington, DC, 2010. http://www.census.gov/statab/www/.
Weeden, K. A., & Grusky, D. B. (2005). The case for a new class map. American Journal of Sociology, 111, 141–212.
Western, B., & Pettit, B. (2005). Black-white wage inequality, employment rates, and incarceration. American Journal of Sociology, 111, 553–578.
Wilson, G. (1997). Pathways to power: Racial differences in the determinants of job authority. Social Problems, 44, 38–54.
Wilson, G. (2009). Downward mobility of women from white collar employment: Determinants and timing by race. Sociological Forum, 24, 382–401.
Wilson, G., & McBrier, D. (2005). Race and loss of privilege: African American/white differences in the determinants of job layoffs from upper-tier occupations. Sociological Forum, 20, 301–321.
Wright, E., Olin, B., Janeen, G., & Birkelund, E. (1995). The gender gap in workplace authority: A cross-national study. American Sociological Review, 60, 407–435.
Zhou, X. (2005). The institutional logic of occupational prestige ranking: Reconceptualization and reanalyses. American Journal of Sociology, 111, 90–140.


1 Increasing gender equality in higher education does not necessarily imply a reduction in gender wage inequality. We discuss this issue later in the paper.
2 Ideally, a study of this sort would include Asian Americans but we were unable to do so because of data limitations. Asian Americans were not included in one of our major data sources: the 1983–1987 CPS surveys, and in later years, the sample sizes were very small.
3 Similarly, the human capital variable, on-the-job-training varies by both race and gender. England (1992) reports that while job tenure accounts for a portion of the wage gaps between white men and women, white men and black women and black and white men, the impact of on-the-job-training was extremely minor for differences between black men and women and black and white women.
4 In contrast, the 2000 Census Occupational Classification Scheme is very different.
5 An alternative given the skewed nature of earnings data would be the median. However, the skewness of earnings within occupations is less than at the individual level. In addition, the correlation between mean and median occupational earnings in the 1990 Census data is .97.
6 This is what is recommended by the staff of the Integrated Public Use Microdata Series at the University of Minnesota Population Center. It has been employed by a number of other researchers (e.g., DiPrete and Buchmann 2006).
7 “Hispanic” includes all individuals who self-identified as Hispanic. Races other than white and black are excluded from the analysis.
8 We tested for nonlinearities by adding a quadratic term to each regression equation. In no case did the addition of this term increase R-Squared by more than .0004.
9 Of course, changes in inequality can result from a number of combinations of groups making progress, staying the same, or regressing. With one exception, we found that all groups made progress on our three measures of occupational attractiveness. Reductions (or increases) in inequality occurred because some groups made more progress than others. The one exception involved analyses of the occupational earnings of Hispanic men: they declined slightly over time.
10 Using bootstrapping techniques, we found the difference between 96% and the other two percentages to be statistically significant. We used 1,000 replications.
11 The results in the second panel are very similar to those in the first, so we do not comment on them here.