PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of demographyspringer.comThis journalThis journalToc AlertsSubmit OnlineOpen Choice
 
Demography. 2009 August; 46(3): 469–492.
PMCID: PMC2831350

Employment Gains and Wage Declines: The Erosion of Black Women’s Relative Wages Since 1980

Abstract

Public policy initiatives in the 1950s and 1960s, including Affirmative Action and Equal Employment Opportunity law, helped mitigate explicit discrimination in pay, and the expansion of higher education and training programs have advanced the employment fortunes of many American women. By the early 1980s, some scholars proclaimed near equity in pay between black and white women, particularly among young and highly skilled workers. More recent policy initiatives and labor market conditions have been arguably less progressive for black women’s employment and earnings: through the 1980s, 1990s, and the first half of the 2000s, the wage gap between black and white women widened considerably. Using data from the Current Population Survey Merged Outgoing Rotation Group (CPS-MORG), this article documents the racial wage gap among women in the United States from 1979 to 2005. We investigate how demographic and labor market conditions influence employment and wage inequality among black and white women over the period. Although shifts in labor supply influence the magnitude of the black-white wage gap among women, structural disadvantages faced by black women help explain the growth in the racial wage gap.

Overall wage differences between black and white women were relatively small in 1980, and after some adjustments, black women outearned their white counterparts (Blau and Beller 1992; U.S. Commission on Civil Rights 1990). Relative wage gains among black men and women through the 1950s and 1960s have been attributed to progressive educational and occupational shifts and antidiscrimination law (Brown 1984; Darity and Mason 1998; Freeman 1973). Increases in the educational attainment of black Americans relative to their white counterparts through the 1950s and 1960s are a key explanation for relative wage gains among blacks (Freeman 1973). A movement away from domestic service and into better-paying occupations was particularly advantageous for the wages of older cohorts of black women workers (Blau and Beller 1992; Goldin 1990). Research also contends that affirmative action and antidiscrimination law fostered black wage gains by helping blacks secure employment in government jobs and professional occupations and through declines in overt wage discrimination (Darity and Mason 1998; Grodsky and Pager 2001).

In the face of optimistic accounts of black women’s economic standing in the early part of the 1980s, the racial wage gap between black and white women has grown dramatically since the 1980s. Data from the Current Population Survey Merged Outgoing Rotation Group (CPS-MORG), shown in Figure 1, indicate that while the white wage advantage among working-age women was just under 5% in 1979, it had nearly tripled by 2005. Dramatic growth in the black-white wage gap among young working women occurred in the 1980s. Among young working women, the wage gap hovered under 5% in 1979, widened considerably through the 1980s to reach a peak in 1990, and has fluctuated between 12% and 15% since the mid-1990s.

Figure 1.
Racial Difference in Average Hourly Wages of White and Black Women: 1979–2005

Large declines in the relative economic standing of black women since 1980 motivate a reexamination of the black-white wage gap among women and its explanations. A labor supply explanation suggests that the decline in the relative wages of black women results from racial differences in employment since 1980. For example, Neal (2004) speculated that declines in real wages among men and changes in public policy might have led to a disproportionate influx of well-skilled white women and poorly skilled black women into the paid labor force. This explanation implies that observed growth in the black-white wage gap among women reflects racial differences in the composition of the labor force and not widening wage inequality among workers with a given set of characteristics.

An alternative account of general trends in labor market inequality emphasizes growing disparities in wages within the labor market (Bernhardt et al. 2001). Structural explanations for growth in labor market inequality in the United States since the 1980s emphasize the importance of labor market conditions for the growth in inequality among men and women and between racial groups. Increasing returns to skill, economic restructuring, declines in government sector work, and the rise of part-time or nonstandard work have all been posited to influence wage inequality (e.g., Bernhardt et al. 2001; Grodsky and Pager 2001; Kalleberg, Reskin, and Hudson 2000; Tilly 1996).

In this article, we examine how shifts in labor supply and the wage allocation process within the labor market have affected inequality in wage outcomes between black and white women since 1980. We have two aims. First, we chart racial differences in employment to consider how labor force composition affects estimates of racial inequality in women’s wages. Women’s employment grew over the period for both blacks and whites, though racial differences in employment have fluctuated in relation to a host of macroeconomic and political conditions. Second, we examine how the wage allocation process differs between black and white women. We examine the extent to which demographic characteristics and labor market conditions affect the wages of employed women and how these influences may or may not have changed over the period to consider the alternative explanation for racial inequality in the labor market.

The racial wage gap compares the per capita wages of employed blacks to the per capita wages of employed whites and therefore represents a compendium measure of access to economic rewards among the employed. In this article, we question whether the wage gap is a good measure of the relative economic status of black women by examining alternative explanations for its growth. Dramatic shifts in women’s employment over the past few decades give reason to suspect that race differences in women’s employment are critical for understanding growth in the wage gap since 1980. At the same time, however, growing inequality within the labor market suggests that black women may be particularly disadvantaged by their location within the structure of the economy. Attention to differences in employment and the allocation of wages of black and white women since 1980 sheds light on the persistence—and extent—of racial inequality in the United States.

EMPLOYMENT AND THE RACIAL WAGE GAP

One of the most dramatic shifts in the U.S. labor market over the past 40 years has been the influx of women into the paid labor market. Employment rates of all women, especially mothers and women with young children, have risen fairly consistently through the last quarter of the twentieth century (Cohen and Bianchi 1999). In 1970, 41% of U.S. women over age 16 were employed. By 2000, 58% of American women were employed—an increase of over 40% (Bureau of Labor Statistics [BLS] 2005).

At the same time that women’s employment has grown, variability in employment rates has been related to overall macroeconomic conditions. Labor market conditions over the past few decades have been highly variable and characterized by alternating cycles of recession and growth. Most economic historians agree that between 1980 and 2005, the United States experienced three recessionary periods: 1980–1982, 1990–1991, and 2001–2003. Although the recession of the early 1980s was accompanied by significant increases in unemployment, the labor market generally withstood the economic declines of the 1990s and 2000s, exhibiting only small increases in unemployment. In contrast, the economic boom of the end of the 1990s brought about historic declines in unemployment, at least as measured by conventional labor force statistics (cf. Western and Pettit 2005).

Differentially increasing employment rates of both black and white women through the 1980s and 1990s have led to a concern that at least part of the widening of the racial wage gap among women may be attributable to changes in the composition of workers in the paid labor force over the period. Neal (2004) argued that economic conditions and social policy may have affected the composition of workers in ways that would be consistent with the observed growth in the racial wage gap among women. Economic conditions, such as declining real wages among men, may have led to the disproportionate influx of white women with high education and high potential wages into the paid labor market through the 1990s as they sought to compensate for the lost earnings of their spouses. At the same time, policy shifts that encouraged employment among low-skill women, including welfare reforms, may have led to increased involvement of black women with low levels of education and low potential wages into the paid labor force (Neal 2004). Neal (2004:S22) speculated that “a full analysis of selection patterns over the past decade might reveal that black-white gaps in wage offers among women actually narrowed over the decade, even though wage data from working women indicate a small increase in the gap” due to the influx into the paid labor market of highly skilled white women who would command high wage offers and poorly educated black women who would command low wage offers.

Existing evidence reveals growing inequality and declining real wages among men in the middle and bottom of the earnings distribution at least through the mid-1990s (Bernhardt et al. 2001). Neal (2004) argued that as men’s wages fall, their wives are increasingly likely to enter the paid labor force. Qualitative evidence—for example, in Newman’s (1993) book Declining Fortunes—carefully illustrates how men’s job losses during the economic downturns in the 1980s and early 1990s undermined men’s household authority as women stepped in (sometimes reluctantly) to support families through increased involvement in waged work. To the extent that new labor market entrants are disproportionately white, highly educated, and would earn more than the prevailing average wage, the average wage of white women may increase simply because of changes in the composition of the employed during turbulent economic times.

Quantitative evidence on the extent to which women, especially high-earning white women, flooded the labor market as their husbands’ real earnings fell is less clear. Some research has found that women’s labor market involvement is relatively insensitive to the employment of their spouse (Juhn and Murphy 1997). Moreover, the men hit hardest by growing inequality through the 1980s and 1990s were from the middle to the bottom of the income distribution. Given educational homophily in marriage patterns (Kalmijn 1998), there is strong reason to suspect that the wives of men most hard hit by growing income inequality would also be drawn from near the middle of the income distribution, and so their increased participation would not increase the average wage of white women.

Accounts of the wage gap that hinge on shifts in employment among highly educated—and presumably high-earning—women are predicated on the assumption that there is significant variability in employment and joblessness in that group to move measures of average wages. This is consistent with claims of an “opt-out revolution” that emphasize swings in the employment of highly educated women, especially in relation to childrearing (Belkin 2003; Stone 2007). If highly educated white women exit the labor force in disproportionate numbers, an opt-out revolution would decrease the average wages of white women; as they return to the labor market—and employment—average wages would increase. However, while claims of opting out have garnered much public attention, evidence suggests that highly educated—and high-earning—women have strong and relatively stable attachments to the labor force (e.g., England, Garcia-Beaulieu, and Ross 2004; Goldin 1990; Percheski 2008). Cohen and Bianchi (1999) have shown increasing educational differences in women’s employment over the previous few decades; college-educated women worked increasingly more hours in the paid labor force relative to those without high school diplomas.

Public policy shifts in the mid-1990s explicitly encouraged welfare recipients, who were disproportionately black and likely to command low wages in the formal economy, to work in the paid labor force. For example, the 1996 Personal Responsibility and Work Reconciliation Act (PRWORA) mandates employment among a proportion of welfare recipients. PRWORA, and earlier state-level changes in work requirements associated with welfare receipt, may have led to the influx of large numbers of former welfare recipients into the paid labor force. Related research has found that the earned income tax credit (EITC) expansion between 1984 and 1996 led to a large increase in employment among single mothers during the period (Meyer and Rosenbaum 2001a, 2001b). To the extent that new labor market entrants are disproportionately black, are disproportionately low-skilled, and receive low wage offers, their inclusion in the paid labor force might drive down the average wages of black women and contribute to the appearance of a growth in the wage gap.

There are at least two important reasons to believe that the implementation of PRWORA would not generate significant changes in the racial composition of the labor force. First, work requirements do not guarantee employment or employment security. Welfare recipients face a number of barriers to employment and stable employment, and research suggests that black and Hispanic welfare recipients may be particularly disadvantaged when seeking employment in the formal economy (Holzer and Stoll 2002). Furthermore, it is important to consider that without strong growth in the economy and a real increase in the number of jobs, some fraction of the population will continue to be involuntarily unemployed or not in the labor force. Even at its trough in 1999, when only 4% of U.S. workers were unemployed, unemployment rates were much higher among black women. Black women’s unemployment was close to 10% in 1999, double the unemployment rate of white women (BLS 2005). Low-skilled women, who are disproportionately black, remain on the margins of the paid labor market.

Moreover, research has suggested that black female heads of household have fared relatively poorly in terms of employment and full-time employment through the 1980s and 1990s (Browne 1997, 2000; Reid 2002). Previous research has emphasized how industrial restructuring and the decline of public sector work negatively impact the employment fortunes of black women generally and black single mothers in particular (Browne 1997, 2000; Newsome and DoDoo 2002; Reid 2002).

In summary, the research is not clear on how shifts in employment within racial or ethnic groups may influence a cumulative measure of inequality like the racial wage gap. It is actually quite difficult to reconcile existing accounts of racial inequality in women’s employment with observed trends in women’s employment. Evidence suggests that although black and white women have both exhibited employment gains, racial inequality in employment has fluctuated in relation to both economic conditions and social policy. Moreover, generalized increases in women’s labor market involvement through the 1970s, 1980s, and 1990s may swamp the effects of smaller, but symbolically important, labor market and policy changes (e.g., welfare reform).

Recent economic and policy changes might affect the composition of the labor force in ways that change our interpretation of over-time comparisons of the black-white wage gap among women. Changing patterns of labor market involvement might lead estimates that rely on employed women to generate an under- or overestimate of the racial gap in wage offers among women. However, it is quite unclear the extent to which changes in labor supply might affect over-time estimates of racial inequality in wages. Detailed empirical tests are necessary to gauge whether and how shifts in women’s employment affect accounts of racial inequality in wages.

STRUCTURAL DISADVANTAGE OF BLACK WOMEN

Growing disparity among workers within the labor market provides an alternative explanation for the widening of the wage gap between black and white women. Accounts of the growth in black women’s economic disadvantage since 1980 have emphasized the significance of human capital and industrial restructuring. Black women are disadvantaged relative to white women in a number of respects that might affect their labor market outcomes: they have lower levels of educational attainment, they are more likely to be unmarried parents, and they are concentrated in nonprofessional/nontechnical jobs. Research has documented growing wage inequality by education among both men and women (Bernhardt et al. 2001; McCall 2001), important wage differences by family structure (Budig and England 2001), and growing inequality by occupation (Autor, Katz, and Kearney 2004; Katz and Autor 1999). Therefore, black women may be disadvantaged by differences in average social and economic characteristics compared with white women and by changing returns to these characteristics in the labor market.

Growth in the wage gap by education (e.g., Bernhardt et al. 2001; McCall 2001), coupled with black women’s concentration among the poorly educated segment of the population, can help explain relative wage declines of black women through the 1980s and 1990s. Black women are much less likely than white women to graduate from high school, attend college, or complete college. Depending on when it is measured, the race gap in high school graduation among young women can vary from as many as 5 to 15 percentage points. Although black women’s educational attainment has increased since 1980, by the close of the twentieth century, employed white women were over 13% more likely than employed black women to have attended some college. As the economic opportunities of the college educated diverge from those with less education, we should expect a growing wage gap between white and black women (McCall 2001).

Race differences in family structure and labor market inequalities by family structure might also influence racial inequality among women. The rising gap in marriage (and unmarried parenthood) between whites and blacks (Ellwood and Jencks 2004), combined with a marriage premium, may exacerbate racial wage gaps given the clear evidence of earnings penalties associated with childbearing and some evidence of a marriage premium (e.g., Budig and England 2001). At the same time, however, single parenthood often necessitates that women work in the paid labor force, especially in the absence of child support payments, welfare supports, or in an era of welfare-to-work programs. Single mothers in the labor market may be particularly disadvantaged because childbearing and childrearing places demands on women that are inconsistent with work in the paid labor force, especially full-time, well-paying jobs.

Occupational shifts away from low-paying domestic and service sector jobs meant that younger cohorts of black women workers fared better than their older black counterparts through the 1980s, primarily through cohort replacement (Blau and Beller 1992). More recently, however, research has noted how industrial restructuring has negatively impacted the employment fortunes of low-skill black women in particular (Browne 1997, 2000; McCall 2001). Recent trends in inequality by occupation (Autor et al. 2004; Katz and Autor 1999) and the concentration of black women in nonprofessional/nontechnical jobs and low-skill sectors with the weakest wage growth are key factors purported to affect growth in the wage gap between black and white women (Browne and Misra 2003; McCall 2005). Declining opportunities in the government sector and a shift away from affirmative action programs in professional/technical jobs may have served to displace blacks from highly paid positions and key avenues for advancement (Grodsky and Pager 2001).

Other economic shifts, including the expansion of part-time, temporary, or fixed-term employment (Kalleberg et al. 2000; Tilly 1996), may help explain growth in racial inequality in wages. The growth of part-time employment, and women’s concentration in it, helps explain why a focus only on full-time workers indicates little change in the black-white wage gap among women through the 1990s (Browne and Askew 2005). Just including women who work part-time into over-time estimates of racial wage inequality presents a dramatically different picture of the black-white wage gap through the late 1980s and 1990s (Moore 2004; and see Figure 1). While black women are less likely to work part-time than white women, they are overrepresented in occupations that are more likely to be both part-time and poorly paid. As a consequence, black women populate an increasingly marginalized segment of the economy, and the persistence of gender and race-typing of jobs may interact to limit the occupational scope—and earnings—for black women (Gibson, Darity, and Myers 1998; King 1995, 1998). To the extent that there is an increasing gap in the wages of full-time and part-time workers and that occupational inequality is growing, we might expect growth in the racial wage gap among women.

Finally, there are lingering questions about the salience of unobserved characteristics and conditions for the growth of the racial wage gap. Dramatic wage declines among young blacks in the 1980s in the face of educational improvements led some scholars to emphasize the relevance of large differences in the returns to unobservable skills (Blau and Beller 1992; Card and LeMieux 1994) and previously unmeasured indicators of earning power, such as cognitive skills (Farkas and Vicknair 1996) and soft skills (Moss and Tilly 1996). Researchers have also claimed that discrimination can help explain the relatively poor economic outcomes for blacks (e.g., Cancio, Evans, and Maume 1996; Kim 2002).

In summary, previous research on wage inequality within the U.S. labor market suggests that the growth in the black-white wage gap among women is not an artifact of shifts in labor supply. Instead, black and white women enter the labor market with different skills and characteristics, the rewards of which have changed quite dramatically over time. Moreover, black and white women occupy largely different segments of the labor market, in occupations and jobs that distribute increasingly unequal rewards. Little research has examined the cumulative effects of race differences in social and demographic characteristics—and their returns in the labor market—on recent trends in the wage gap between black and white women. This article examines alternative explanations for the growth in the racial wage gap among American women since 1980. A better understanding of the mechanisms undergirding racial inequality in employment and wages is critical to gauge the economic standing of black women and may have important implications for policy and practice both within the labor market and outside of it in domains ranging from affirmative action and antidiscrimination enforcement to education.

METHOD

In this article, we investigate how changes in employment and racial differences in social and demographic conditions and their returns in the paid labor force influence our understanding of the black-white gap in wages among women. First, we construct a counterfactual wage gap to test the labor supply explanation. That is, do changes in the composition of the labor force account for the growing black-white wage gap among women? The counterfactual wage gap is a weighted average of observed wages for employed women and hypothetical wage offers for nonemployed women.1 The difference between the observed gap (Figure 1) and an adjusted gap depends on two key quantities: the size of the employed population relative to the size of the jobless population, and differences in the wage offers that employed and jobless women would receive. To the extent that either or both of these change over the period of observation, there may be larger or smaller wage adjustments over time.

Second, we conduct a decomposition of the black-white wage gap among employed women to explore alternative explanations for growing racial inequality. We examine how the social and demographic factors that influence wages have changed over time and differ between blacks and whites, and how shifts in the labor market returns to observable characteristics and the effects of unobservable characteristics and conditions may have changed over time. Results from a decomposition point to the conditions that contribute to growing black disadvantage.

To study wage inequality between black and white women from 1979 through 2005, we analyze data from the Current Population Survey Merged Outgoing Rotation Group (CPS-MORG). This survey contains employment and wage information for CPS households. Estimates of inequality in this article are based on log hourly wages, although similar results were found for weekly earnings. Wages are adjusted for inflation. The analysis is restricted to non-Hispanic, nonfarm, civilian women, including part- and full-time workers. We report results for white and black women aged 22–65 and 22–35. We compare the results for 22- to 65-year-old women and 22- to 35-year-old women to examine whether overall patterns hold for young women. Employment-population ratios are calculated using the CPS survey weights.

ESTIMATING THE EFFECTS OF CHANGING LABOR MARKET INVOLVEMENT

If log wages of white and black women are written yw and yb, respectively, then the difference in mean wages is given by d = ywyb. Because hypothetical wage offers of the jobless are not observed and the jobless are hypothesized to come from the tails of the wage distribution, d is a potentially biased estimate of the differential in wage offers. To adjust the wage differential for selective attrition from employment, we calculate

d^=y^wy^b,

where the adjusted means ŷi include imputed mean wages for nonemployed women. Omitting the race subscripts, the adjusted mean wage is the weighted average,

y^=(1pN)y¯W+pNy¯N,

where the subscript W denotes the mean calculated for the employed from observed wages, yN is the mean wage for nonemployed, and the weights pN are proportions of the population not employed.

There is not a well-established strategy for assigning wage offers to the jobless (Darity 1980). Therefore, we adopt three strategies to impute wages to the jobless: (1) a regression adjustment; (2) a regression adjustment that includes a 20% wage penalty for time out of the labor market; and (3) an assignment of zero wages to the jobless.

In the regression adjustment, wages of the jobless are generated from mean wages of the employed. The predicted mean wage for the nonemployed can be written as a regression equation, y¯N=X¯NbW, in which bW are the coefficients generated from wage regressions for the employed, and XN is a vector of covariate means for the nonemployed. Mean wages are predicted accurately if, given included covariates, the mean wage offers of the nonemployed are identical to the mean wage of the employed. With this regression adjustment, women do not incur a wage penalty for time out of the labor force and so wage differences between the employed and the jobless are due only to differences in the composition of the groups, XWXN , and their relative sizes.

As in previous research, we predict the wages of the jobless given age and education (Blau and Beller 1992; Neal 2004), which capture the main human capital differences in wages. We also include marital status and region. Age is measured continuously, education is measured in years of school completed, and marital status is divided into (1) never married, (2) married, and (3) separated, divorced, or widowed. We ran the analyses separately for married and nonmarried women to test whether the racial wage gap operates differently for these groups. The substantive results did not change, so we report only the results of the pooled analyses. There are four regional categories: South, Midwest, West, and Northeast. The covariates are used to help sharpen predictions for the wages of the jobless.2

The regression adjustment assumes that the wage offers received by the jobless, given observed covariates, are equal to those of the employed, although there is reason to believe this is unrealistic. Those not employed may earn less or more than those employed depending on the sample selection mechanism. The regression adjustment may understate wage offers if a loss in income causes or is otherwise related to joblessness. On the other hand, wage offers are likely to be relatively low among many jobless, leading to an overestimate of mean wages in the nonemployed population. In order to test the sensitivity of results to assumptions about the wage offers of the jobless, our second adjustment strategy assigns a 20% wage penalty to the jobless.3

Our 20% wage penalty is motivated by previous research investigating the wage penalty associated with periods of joblessness. Chowdhury and Nickell (1985) found that returning workers earned 13%–14% less immediately following an unemployment spell. Fallick’s (1996) review of the research on displaced workers revealed inconclusive evidence regarding the effects of unemployment due to displacement on earnings. Research suggests that displaced workers face wage losses of 11% to 16% immediately after finding work and wage losses of 14% to 25% several years later compared with continually employed individuals (Fallick 1996). To the extent that displaced workers differ from other workers experiencing periods of unemployment, the wage penalties faced by displaced workers may be more or less compared to the wage penalties faced by workers after unemployment spells due to other causes (Kletzer 1998). Our 20% wage penalty falls in the middle range of possible wage losses of long-term displaced workers and may represent the high end of wage penalties faced by other workers immediately following an unemployment spell.

Our third adjustment strategy assigns zero wages to the jobless (Darity and Myers 1998). It is unrealistic to think that the nonemployed would be offered zero wages if they sought employment. However, assigning zero wages to the jobless represents an “upper bound” on economic inequality between black and white women by incorporating differences in employment along with differences in wages. The zero-wage adjustment might be thought of as a representation of the average wages of all black women (not just the employed) compared with those of white women. When racial inequality in employment is high, the difference between the observed gap and the zero-wage adjustment will be greatest. When racial inequality in employment is low, there will be little difference between the regression adjustment and the zero-wage adjustment.

The three adjustment strategies make different assumptions about the wage-offer process. The regression adjustment incorporates information about employment rates and compositional differences between the employed and the jobless. The 20% penalty adjustment is similar to the regression adjustment but updated with the expectation that women will be penalized for spending time out of work. The zero-wage adjustment includes information only about employment rates. We don’t have strong expectations about the wage offers the nonemployed would receive, and we hesitate to make claims that any one of these adjustments is more realistic than another. However, we are interested in how each of the adjustments differs from the observed racial wage gap; comparing them will reveal important differences in the relative effects of the size and the composition of the jobless for estimates of racial inequality in wages.

Because predicted wages may be too high or too low, there is uncertainty about how well the observed wage data predict wage offers. In this article, this uncertainty is incorporated in the calculation of standard errors and confidence intervals. Following Western and Pettit (2005), we construct a subjective probability interval for adjusted wages. This involves constructing a Bayesian distribution for the predictive mean, yj = Xjbj (Little and Rubin 1987; Rubin 1977).

There are no sample data on wages for the nonemployed, but prior information is given by wage data in the CPS. Uncertainty about the predictive mean is a function of prior uncertainty about the coefficient vector bN. Prior uncertainty about the counterfactual wages of the jobless depends partly on sampling uncertainty in the CPS wage estimates and subjective uncertainty about the utility of the CPS for predicting counterfactual wages of the jobless. Prior means for the coefficients, bN, are given by the mean wages for the employed, bW, yielding predictive distributions centered at y¯N=X¯NbW.

Prior variances for the coefficients are set to multiples of the least-squares variances, ψ12V^W (for the jobless). The prior parameter ψ1 describes confidence that the wage offers of the nonemployed match the CPS wages for the employed. If the generating assumption is true and the conditional wages of workers accurately describe the counterfactual wages of the jobless, then ψ12=1.0. Throughout this analysis, ψ12=2.0, reflecting skepticism that the survey data on which prior variances are based accurately predict the unobserved wages. Uncertainty about the true location of the coefficients vector bN is set to twice the conventional sampling uncertainty usually calculated from the CPS.

DECOMPOSING RACE DIFFERENCES

This article also presents a decomposition of the black-white wage gap among employed women in an effort to understand changes in wage allocations that may have differentially affected blacks and whites through 2005. The decomposition strategy employed here follows Blau and Beller (1992) to evaluate the impact of race differences in the means of the variables on the log earnings differential. In this analysis, we include covariates for age, marital status, and region as coded in the previous analysis of selection effects. We code education into four categories: less than a high school diploma, a high school diploma, some college, and a college degree or higher. We also include indicators of part-time status based on the CPS code for part-time work. The portion of the race differential that is not explained by differences in the means or effects of observed covariates may be thought of as the impact of unmeasured indicators of productivity or, possibly, discrimination.

We estimate separate wage regressions for whites and blacks and decompose the wage gap into the total differential (T), mean effect (M), and unexplained differential (U). Following Blau and Beller (1992:283),

T=ibiwxiwibibxib

M=ibiw(xiwxib)

U=TM=ixib(biwbib)

where bij and xij are the regression coefficient and weighted mean, respectively, of the ith variable for the jth race group (w = whites, b = blacks), and year subscripts are omitted. We report results using the white wage function, but we also conducted analyses using the black wage function.4 The unadjusted (RX) and adjusted (RA) ratios are equal to

RX=eT

and

RA=eU.

It is important to consider, however, that analyses based on the employed may lead to biased estimates of skill effects. To the extent that poorly educated women, and poorly educated black women in particular, are left out of analyses designed to examine the relationship between education and wages, we may be misstating the true effects of education. Moreover, if we fail to observe the most disadvantaged blacks in conventional analyses, and to the extent that education and race are tightly coupled, we also may be misstating the true effects of race on economic outcomes.

RESULTS

Employment and Labor Force Composition

The proportion of women who are employed has grown dramatically among both whites and blacks between 1979 and 2005. The top panel of Table 1 shows that the fraction of both white and black jobless women aged 22–65 fell between 1979 and 2005. Two-fifths of white women were not employed at the beginning of the period, but only 27.7% were not employed by the end of the period. Jobless rates are slightly higher among blacks: the percentage of women not employed fell from 43.1% to 32.6% over the period. Employment rates dipped—especially for black women—in the early 2000s.

Table 1.
Percentages of Black and White Women Who Are Jobless: 1979–2005

The bottom panel of Table 1 shows joblessness among young women (aged 22–35). Among young white women, the fraction of women not employed is slightly lower than for all white women but exhibits similar declines through the 1980s and 1990s. Young black women’s jobless rates fluctuate over the period, showing greater variability than jobless rates of all black women. Young black women exhibit particularly high levels of joblessness during the early to mid-1980s and again in the early to mid-1990s. There is an uptick in joblessness among both white and black young women in the early 2000s, though the decline in employment is more acute among blacks; during 2003–2005, the jobless rate of young black women matches that found among all black women.

Trends in joblessness between whites and blacks are shown in the final column of Table 1. These results indicate growing racial inequality in employment through the early 1990s, strong employment gains for blacks through the early 2000s, and growth in racial inequality in employment in the last three years of the period.

Growth in women’s employment among both blacks and whites might lead us to expect that the gap between the wages of the employed and nonemployed should narrow over time as the size of the jobless population gets smaller. However, increasing labor inactivity among young black women appears to be driving periods of widening racial employment gaps, especially between 2003 and 2005. The wage adjustment may actually get larger as the predicted wages of young jobless women diverge even more dramatically from those of older women employed in the paid labor force.

The size of the adjustment between observed wages among employed women and hypothetical wage offers among all women depends not only on the proportion of employed women relative to jobless women, but also on how the composition of the employed and jobless compare, especially with respect to characteristics that are systematically related to wage offers. To the extent that the composition of those employed relative to the non-employed changes over time, the observed racial wage gap may conflate labor supply effects with shifts in the wage allocation process.

Comparing the age and education distributions of employed and jobless women in Table 2 indicates some similarities and several important differences between employed and jobless whites and blacks in 2005. Among white women in 2005, jobless women are older than employed women. Among black women, joblessness is concentrated among both the youngest and oldest age groups. Within each race, more jobless women failed to complete high school or attend college than employed women. However, the proportion of nonemployed black women who did not complete high school is more than twice as large as the proportion of nonemployed white women without a high school diploma. Similarly, while 53% of jobless white women attended some college, only 37% of jobless black women attended college.

Table 2.
Percentages of Black and White Women Aged 22–65 Who Are Employed and Jobless, by Age and Education: 2005

Race differences in the composition of the employed relative to the jobless suggest that wage offers of the jobless (both white and black) are likely to be lower than wage offers among the employed. It isn’t clear, a priori, how much wage offers of the employed and jobless will differ between blacks and whites. Lower levels of education among the jobless suggest that they may have lower wage offers than more highly educated employed women. However, the negative effects of education may be partially offset by returns to age for older jobless women.

There are some important race differences in social and demographic characteristics of employed women over time that may also help explain persistent racial inequality as measured by the black-white wage gap. Table 3 shows the average social and demographic characteristics for employed black and white women (aged 22–65) from 1985 to 2005 in 10-year increments. Employed white women are, on average, increasingly likely to be older than employed black women; the difference in average age between employed black and white women grew by 0.52 years between 1985 and 1995, and by 0.95 years between 1995 and 2005. The average level of education has grown slightly faster among employed whites than it has among employed blacks. Perhaps most dramatically, employed blacks are increasingly likely to be unmarried (or never married) compared with whites. Involvement in professional jobs has increased more among white women over the period, while involvement in sales occupations has grown more among black women. Adverse trends in the social and demographic characteristics of employed blacks compared with employed whites certainly deserve further scrutiny.

Table 3.
Average Characteristics of Employed White and Black Women Aged 22–65: 1985–2005

Accounting for Labor Market Selectivity

We construct counterfactual wage gaps between black and white women to examine whether changes in labor supply drive the wage gap between black and white women. Estimating wage offers for jobless women reveals that conventional labor force statistics dramatically overstate black women’s relative economic standing. Among young women, black women’s relative wages may be overstated by as much as 110% due to high rates of labor inactivity among young, low-skilled black women.

The top panel of Table 4 shows observed and adjusted differences in hourly wages between white and black women aged 22–65 in 2005 using our three adjustment strategies. The first adjustment assumes that jobless women would earn wages comparable to those of similar women employed in the paid labor force; not accounting for racial differences in labor market participation leads to an 11.7% overestimate of black relative wages. To the extent that jobless women garner lower wages upon returning to the labor force due to less experience and other factors, this first adjustment underestimates the magnitude of the black-white wage gap. Assuming the jobless would incur wage penalties leads to larger differences between the adjusted wages of black and white women. Assuming a wage penalty of 20% leads to a 37.9% adjustment. Assigning zero wages to the jobless represents the upper bound of the black-white wage gap; the zero wage adjustment indicates that conventional estimates of the wage gap overstate black relative wages by 71%.

Table 4.
Observed and Adjusted Log Hourly Wages for Black and White Women: 2005

The bottom panel of Table 4 indicates that among young women aged 22–35, the black-white wage gap is smaller than among all women, but the size of the adjustment for the nonemployed depends on the wage penalty assessed to the jobless. Estimating counter-factual wage offers without penalties for jobless young women generates a regression-adjusted wage gap 8.3% larger than the observed gap. Although the observed wage gap is smaller for young women than for all women, adjustments assigning a 20% wage penalty and zero wages result in larger differences between the observed and adjusted wage gaps for young women compared with all women. Assuming the jobless would incur a 20% wage penalty suggests that conventional estimates overstate black relative wages by 50%. Assigning zero wages to the jobless inflates the gap by 110% for young women.

Time-series analyses of adjustments to the gap presented in Figure 2 show that differential selectivity into the labor market contributes to consistent underestimates of the black-white wage gap among women. The top panel of Figure 2 plots observed wage gaps from 1979 to 2005 for women aged 22–65 along with the regression adjustment and the zero wage adjustment. The figure clearly demonstrates an increase in the black-white wage gap over the period, and adjustments to the gap indicate that the wage gap among workers is consistently smaller than the counterfactual gaps in wage offers.

Figure 2.
Time Trends of Hourly Wages, Wage Adjustments, and Selection Effects for Black and White Women: 1979–2005

The regression adjustment (shown in the bottom panel of Figure 2) does not suggest that recent compositional changes in the labor force explain the widening wage gap between black and white women; the difference between the regression-adjusted and observed wage gap does not increase over the period. These results suggest that changes in labor market selectivity do not explain the fundamental observation that the relative wages of black women are at best stagnating, and at worst lagging further behind wages of white women through the 1980s, 1990s, and early 2000s. The composition of the labor force affects the magnitude of the black-white wage gap among women; however, compositional changes do not explain the observed increase in the gap.

There is some evidence that the size of the selection effect declined through the late 1990s and then increased after 2000, particularly in relation to economic conditions that influence joblessness. We can see this clearly in Figure 2, where we assign zero wages to the jobless. The zero wage adjustment generates the largest gap and shows a great deal of variability over time and in relation to economic cycles. This adjustment, which takes into consideration size but not composition of the labor force, represents the ceiling of the wage gap. The difference between the gap including zero wages and the observed gap grows most steeply during the early 1990s, declines during the last half of the 1990s, and grows again after 2000. This figure implies that the employment of black women is more susceptible than the employment of white women to macroeconomic conditions.

Sensitivity analyses to test alternative assumptions about sampling variance and the similarity between the wages of employed and jobless women indicate the robustness of results presented above. The vertical lines in the bottom panels of Figure 2 show 95% confidence intervals for the size of the adjustment effect. Despite the relatively small size of the adjustments, the confidence intervals never include zero. We can be quite confident in these results even assuming twice the usual sampling variance.

The general patterns observed among women aged 22–65 hold for young women aged 22–35. However, assuming larger wage penalties for the jobless among young women leads to even larger selection effects, strong growth in selection effects in the early 1980s and early 1990s, declines in selection effects in the late 1990s, and growth in selection effects after 2000. If wage offers for the jobless are only 80% of wages for employed women, black relative wages for young women are overestimated by approximately 50% in the early 1990s and again in 2005. Figure 2 shows that the zero wage adjustment results in the largest difference between the observed and adjusted wage gap in the early 1990s and early 2000s during times of economic recession.

In summary, there is clear evidence of growth in the black-white wage gap among women workers between 1979 and 2005. After taking into consideration the composition of the labor force, we see a much larger black-white wage gap among women. This article demonstrates that the black-white wage gap among employed women overstates the relative economic standing of black women in the United States, since low-earning blacks are disproportionately unemployed or not in the labor force. There is little evidence that the growth in the gap through the 1990s and early 2000s is attributable to changes in labor market composition.

Differences in Means, Returns, and Unobservable Variables

Table 5 presents a decomposition of the log wage differential every five years between 1980 and 2005 to investigate the extent to which differences in average characteristics and changing labor market returns to these characteristics explain the growth in the wage gap between whites and blacks aged 22–65 over the period. The top section of the table documents the portion of the black-white wage gap that is due to differences in average characteristics, including age, education, marital status, part-time employment, and region between black and white women. The table includes a summary of the total portion of the black-white wage gap that is due to differences in average characteristics. The next section of the table documents the portion of the wage gap that is due to differences in returns to those same characteristics.

Table 5.
Decomposition of the Log Earnings Differential (white wage function) for Women Aged 22–65

Table 5 indicates the growing importance of differential average socioeconomic characteristics between blacks and whites in explaining race differences in wages for women aged 22–65. For each year presented from 1985 onward, the proportion of the total wage differential due to means, or average characteristics, was larger than the previous year reported. The last five columns of the table reflect this growth in importance of average characteristics: the change in proportion attributable to total mean differences across five-year increments was positive for every period after 1985. Therefore, adverse trends in average socioeconomic characteristics slowed wages for employed black women aged 22–65 relative to white women. Results further indicate that differences between black and white women in average education contributed the most to the growing black disadvantage. However, race gaps in marriage and region also help to explain growth in the black-white wage gap among employed women.

Labor market returns to age, education, and marital status became increasingly important determinants of race differences in wages toward the end of the period under study. The proportion of the wage gap due to differences in returns to age grew for every year reported after 1990, resulting in positive changes shown in the last three columns of the table. The importance of returns to education grew intermittently throughout the period, and the proportion of the racial wage gap attributable to differences in returns to education was larger at the end of the period in 2005 than it was at the beginning of the period in 1980. The importance of returns to marital status in explaining the racial wage gap decreased during the 1980s but increased since 1990. The importance of returns to region on the racial wage gap fluctuated across the period. By 2005, returns to region of the country were actually a less important determinant of race differences in wages than they were at the beginning of the period.

The table also indicates a growing role for unexplained differences between black and white women. Although the proportion of the wage gap that is due to unexplained differences decreased overall from .114 in 1980 to .036 in 2005, the part due to unexplained differences grew since 1995, indicating a trend toward increasing importance. Although these unexplained differences could reflect unmeasured differences in productivity, such as cognitive skill (Farkas and Vicknair 1996) or racial inequality in soft skills (Moss and Tilly 1996), they might also reflect a growing salience of discrimination or the consequence of a retreat from affirmative action programs (e.g., Dobbin et al. 1993).

Table 6 reports a decomposition of the log wage differential between young black and white women aged 22–35 over the period. The results are quite consistent with those presented in Table 5 and confirm the salience of differences in means, returns, and unobservable variables for the black-white wage gap among young women. It is quite striking that 50 years after Brown v. Board of Education, educational differences between white and black women play an increasingly important role in generating wage inequalities even among young women. The wage gap among young women is also acutely affected by growing returns to age, marriage, and the proportion of variance unexplained. For example, the last five columns of the table show that the proportion of the wage gap due to returns to age grew overall across the period, with positive changes in each five-year increment reported. Returns to marital status for young women grew in importance since 1990 and explained a greater proportion of the wage gap at the end of the period in 2005 than at the beginning in 1980.

Table 6.
Decomposition of the Log Earnings Differential (white wage function) for Women Aged 22–35

Table 6 also indicates growth in the unexplained component of race differences in wages for young women across the period. By 2005, labor market returns to age and education explained a greater proportion of the racial wage gap for young women than for all women combined. Furthermore, the unexplained component of race differences in wages was greater for young women than for women aged 22–65 at the end of the period. However, because the effect is not limited to young workers, as documented in previous research, we suspect this effect is related to general labor market conditions that disadvantage black workers rather than to returns to cognitive skills, which have particularly disadvantageous effects for young workers.

CONCLUSION

Conventional labor force statistics indicate a marked increase in the racial wage gap among women through the 1980s, 1990s, and early 2000s. The racial wage gap has more than doubled among all working-age women and tripled among young workers in the two-and-a-half decades since 1980. Our analyses show that estimates of racial inequality in wages that rely on the employed consistently underestimate the growing cleavage between black and white women’s economic fortunes. Accounting for racial inequality in employment generates consistently larger estimates of the gap, though it fails to account for growth in the gap since 1980.

While there is some concern that growth in employment among women since 1980 undermines the utility of the black-white wage gap as a measure of the relative economic standing of black women, racial inequalities in employment obscure the full extent of racial inequality in wage offers. Black and white women have both experienced large employment gains since 1980, yet black women continue to be at greater risk of joblessness, especially in recessionary periods. Our analyses demonstrate that including the jobless in estimates of racial differences in wage offers generates race differences in offer wages 11%–71% larger than those observed among employed women; similar effects are found for young women. These effects are robust to alternative assumptions about the wage offers of the jobless and are consistent over the life course. Black women’s wages—both observed and expected—continue to lag well behind the wages of white women.

We find evidence in support of explanations for the growth in the racial wage gap that emphasize changes in average characteristics of employed black and white women as well as those that emphasize shifts in the allocation of wages within the labor market. Racial differences in observed social and demographic characteristics help to explain at least part of the shift in the magnitude, and perhaps the trend, in the racial wage gap. Widening gaps in education, marriage, and region all contribute to growing inequality. However, they don’t fully account for the fundamental observation that black women’s wages lag farther behind those of their white counterparts. Increasing returns to age, education, and marriage help explain a growing portion of the black-white wage gap among employed women.

In addition, a growth in the proportion of unexplained variance suggests the continued relevance of wage discrimination or differences in unmeasured indicators of productivity. Since the salience of unmeasured indicators is large for both older and younger workers, we conjecture that it reflects overall labor market conditions that disadvantage blacks (e.g., discrimination) rather than returns to cognitive skills, which have had particularly disadvantageous effects for young workers. Our results are consistent with studies that emphasize how the decline of government jobs and the movement away from affirmative action programs may have disadvantaged black women (e.g., Dobbin et al. 1993; Grodsky and Pager 2001).

Estimates of relative economic standing based on data from employed workers alone understate the extent of disadvantage that black women face. Despite growth in employment among both black and white women, there is clear evidence of continued economic marginalization of black women, and low-skilled black women in particular. Moreover, race differences in social and demographic characteristics like education and single parenthood have served only to further cleave the economic opportunities of black women from those of whites. As a consequence, not only is the black-white gap in inequality among women understated in conventional labor force statistics, but a comprehensive understanding of the significance of race for wage outcomes is obscured.

These results emphasize how the labor market both reflects and reinforces inequality in other social domains. Despite decades of educational expansion, employed black women continue to lag behind employed whites in the educational qualifications that are increasingly relevant in the contemporary workplace. Premarket educational inequalities are magnified by a labor market that increasingly rewards education. Widening racial gaps in marriage—combined with growing returns to marriage—also disadvantage African American women. These factors, combined with a retreat from affirmative action programs and weak enforcement of employment discrimination law, may have uniquely disadvantaged the economic fortunes of black women.

Near parity in black and white women’s wages in the late 1970s drew academic attention away from the economic concerns of black women. However, recent changes in both the level and trend in the black-white wage gap among women demand further attention and research. Our results have begun to illustrate how both patterns of employment and within-market allocation processes are associated with widening wage inequality. A full accounting of racial inequalities in the economy demands greater attention to the conditions both within and outside the marketplace that contribute to enduring racial inequalities in wages and the conditions that produce them.

Footnotes

This work has been supported by a National Institutes of Health K-01 Mentored Research Development Award (K01-HD049632-01A1) and the article was completed while the first author was a visiting scholar at Northwestern University and the American Bar Foundation.

1.We use the term nonemployed synonymously with jobless to characterize women who are not currently employed or working for pay. The jobless population includes the unemployed and those not in the labor force.

2.We do not include information on number of children because of poor data quality over the period. While the CPS-MORG extracts are large and contain detailed information on employment and earnings, they do not contain information on the presence or number of biological children for many of the years covered by our analysis (Feenberg and Rothl 2007). Specifically, the CPS-MORG files do not include variables related to children for years 1979–1983 or 1994–1998. We examined the influence of children specifically using data from the June CPS. The results are consistent with those presented in this article and are available from the authors.

3.Previous studies combining observed wages from the employed with imputed wages from the nonemployed estimated wages for the jobless using a range of assumptions. Blau and Beller (1992) presented a range of counter-factual estimates by assuming that the jobless earned between 0% and 40% less than the employed with the same observed characteristics. Neal (2004) used imputation methods that varied across subgroups, typically assigning relatively low wages to black jobless and high wages to white jobless—a preliminary form of the median regression technique employed in previous work (Neal and Johnson 1996).

While predictions that acknowledge the low earning power of the jobless may be more realistic than predictions that assume that the nonemployed earn the same as their employed counterparts, there is very little evidence to suggest that any group of nonemployed women would experience much of a wage premium above similarly skilled employed women when they entered the paid labor market. Evidence suggests that time out of the labor force is a consistent predictor of wage declines, and there are both theoretical and empirical reasons to believe that highly skilled workers would experience the largest wage penalties when returning to the paid labor force (Mincer 1974). An analysis of previously unemployed workers using the National Longitudinal Survey of Youth (NLSY) found that the wage gap between continuously employed women and women with recent periods of unemployment is largest among highly educated white women and smallest among highly educated black women (Dozier 2007). There is no evidence to suggest that poorly educated blacks experience a greater wage penalty than poorly educated whites (Dozier 2007).

4.Results using the black wage function are available from the authors.

REFERENCES

  • Autor D, Katz L, Kearney M. National Bureau of Economic Research; Cambridge, MA: 2004. “Trends in US Wage Inequality: A Re-Assessment of the Revisionists” NBER Working Paper No 11627.
  • Belkin L. “The Opt-Out Revolution” The New York Times. 2003. October 26.
  • Bernhardt A, Morris M, Handcock M, Scott M. Divergent Paths: Economic Mobility in the New American Labor Market. New York: Russell Sage Foundation; 2001.
  • Blau F, Beller A. “Black-White Earnings Over the 1970s and 1980s: Gender Differences in Trends” Review of Economics and Statistics. 1992;74:276–86.
  • Brown C. “Black-White Earnings Ratios Since the Civil Rights Act of 1964: The Importance of Labor Market Dropouts” Quarterly Journal of Economics. 1984;99:31–44.
  • Browne I. “Explaining the Black-White Gap in Labor Force Participation Among Women Heading Households” American Sociological Review. 1997;62:236–52.
  • Browne I. “Opportunities Lost? Race, Industrial Restructuring, and Employment Among Young Women Heading Households” Social Forces. 2000;78:907–29.
  • Browne I, Askew R. “Race, Ethnicity, and Wage Inequality Among Women” American Behavioral Scientist. 2005;48:1275–92.
  • Browne I, Misra J. “The Intersection of Race and Gender in the Labor Market” Annual Review of Sociology. 2003;29:487–513.
  • Budig M, England P. “The Wage Penalty for Motherhood” American Sociological Review. 2001;66:204–25.
  • Bureau of Labor Statistics (BLS) Bureau of Labor Statistics; Washington, DC: 2005. “Women in the Labor Force: A Databook.”
  • Cancio AS, Evans TD, Maume D. “Reconsidering the Declining Significance of Race: Racial Differences in Early Career Wages” American Sociological Review. 1996;61:541–56.
  • Card D, Lemieux T. National Bureau of Economic Research; Cambridge, MA: 1994. “Changing Wage Structure and Black-White Wage Differentials Among Men and Women: A Longitudinal Analysis” NBER Working Paper No 4755.
  • Chowdhury G, Nickell S. “Hourly Earnings in the United States: Another Look at Unionization, Schooling, Sickness, and Unemployment Using PSID Data” Journal of Labor Economics. 1985;3:38–69.
  • Cohen P, Bianchi S. “Marriage, Children, and Women’s Employment: What Do We Know?” Monthly Labor Review. 1999;122:22–31.
  • Darity W. “Illusions of Black Economic Progress” The Review of Black Political Economy. 1980;10:153–68.
  • Darity W, Mason P. “Evidence on Discrimination in Employment: Codes of Color, Codes of Gender” Journal of Economic Perspectives. 1998;12:63–90.
  • Darity W, Myers S. Persistent Disparity: Race and Economic Inequality in the United States Since 1945. New York: Edward Elgar Publishing; 1998.
  • Dobbin F, Sutton J, Meyer J, Scott R. “Equal Opportunity Law and the Construction of Internal Labor Markets” American Journal of Sociology. 1993;99:396–427.
  • Dozier R. University of Washington; 2007. “Accumulating Disadvantage: The Growth in the Black-White Wage Gap Among Women” Unpublished dissertation.
  • Ellwood D, Jencks C. “The Spread of Single-Parent Families in the United States Since 1960.” In: Moynihan D, Smeeding T, Rainwater L, editors. The Future of the Family. New York: Russell Sage Foundation; 2004. pp. 25–65.
  • England P, Garcia-Beaulieu C, Ross M. “Women’s Employment Among Blacks, Whites, and Three Groups of Latinas: Do More Privileged Women Have Higher Employment?” Gender and Society. 2004;18:494–504.
  • Fallick B. “A Review of the Recent Empirical Literature on Displaced Workers” Industrial and Labor Relations Review. 1996;50:5–16.
  • Farkas G, Vicknair K. “Appropriate Tests of Racial Wage Discrimination Require Controls for Cognitive Skill: Comment on Cancio, Evans, and Maume” American Sociological Review. 1996;61:557–60.
  • Feenberg D, Rothl J. National Bureau of Economic Research; Cambridge, MA: 2007. “CPS Labor Extracts 1979–2006.”
  • Freeman R. “Changes in the Labor Market for Black Americans, 1948–1972” Brookings Papers on Economic Activity. 1973;1:67–131.
  • Gibson K, Darity W, Myers S. “Revisiting Occupational Crowding in the United States: A Preliminary Study” Feminist Economics. 1998;4(3):73–95.
  • Goldin C. Understanding the Gender Gap. New York: Oxford University Press; 1990.
  • Grodsky E, Pager D. “The Structure of Disadvantage: Individual and Occupational Determinants of the Black-White Wage Gap” American Sociological Review. 2001;66:542–67.
  • Holzer H, Stoll M. Assessing the New Federalism. Urban Institute; Washington, DC: 2002. “Employer Demand for Welfare Recipients by Race.” Discussion Paper No. 01-07.
  • Juhn C, Murphy K. “Wage Inequality and Family Labor Supply” Journal of Labor Economics. 1997;15:72–97.
  • Kalleberg A, Reskin B, Hudson K. “Bad Jobs in America: Standard and Non-standard Employment Relations and Job Quality in the United States” American Sociological Review. 2000;65:256–78.
  • Kalmijn M. “Intermarriage and Homogamy: Cases, Patterns, and Trends” Annual Review of Sociology. 1998;24:395–421. [PubMed]
  • Katz L, Autor D. “Changes in the Wage Structure and Earnings Inequality.” In: Ashenfelter O, Card D, editors. Handbook of Labor Economics. 3A. New York: North-Holland; 1999. pp. 1463–55.
  • Kim M. “Has the Race Penalty for Black Women Disappeared in the United States?” Feminist Economics. 2002;8:115–24.
  • King M. “Black Women’s Labor Market Status: Occupational Segregation in the United States and Great Britain” Review of Black Political Economy. 1995;24(1):23–43.
  • King M. “Are African-Americans Losing Their Footholds in Better Jobs?” Journal of Economic Issues. 1998;32:641–69.
  • Kletzer L. “Job Displacement” Journal of Economic Perspectives. 1998;12:115–37.
  • Little R, Rubin D. Statistical Analysis With Missing Data. New York: Wiley; 1987.
  • McCall L. “Sources of Racial Inequality in Metropolitan Labor Markets: Racial, Ethnic, and Gender Differences” American Sociological Review. 2001;66:520–41.
  • McCall L. “Managing the Complexity of Intersectionality” Signs: Journal of Women in Culture and Society. 2005;30:1771–800.
  • Meyer B, Rosenbaum D. “Making Single Mothers Work: Recent Tax and Welfare Policy and Its Effects.” In: Meyer B, Holtz-Eakin D, editors. Making Work Pay: The Earned Income Tax Credit and Its Impact on America’s Families. New York: Russell Sage Foundation; 2001a. pp. 69–115.
  • Meyer B, Rosenbaum D. “Welfare, the Earned Income Tax Credit, and the Labor Supply of Single Mothers” Quarterly Journal of Economics. 2001b;116:1063–2014.
  • Mincer J. Schooling, Experience, and Earnings. New York: Columbia University Press; 1974.
  • Moore Q. Department of Economics, University of Michigan; 2004. “Is the Measured Black-White Wage Gap Among Women Still Too Small?” Unpublished manuscript.
  • Moss P, Tilly C. “Soft Skills and Race: An Investigation of Black Men’s Employment Problems” Work and Occupations. 1996;23:252–76.
  • Neal D. “The Measured Black-White Gap Among Women Is Too Small” Journal of Political Economy. 2004;112:S1–S28.
  • Neal D, Johnson W. “The Role of Pre-market Factors in Black-White Wage Differences” Journal of Political Economy. 1996;104:869–95.
  • Newman K. Declining Fortunes: The Withering of the American Dream. New York: Basic Books; 1993.
  • Newsome Y, DoDoo FN-A. “Reversal of Fortune: Explaining the Decline in Black Women’s Earnings” Gender and Society. 2002;16:442–64.
  • Percheski C. “Opting Out? Cohort Differences in Professional Women’s Employment Rates From 1960 to 2005” American Sociological Review. 2008;73:497–517.
  • Reid L. “Occupational Segregation, Human Capital, and Motherhood: Black Women’s Higher Exit Rates From Full-Time Employment” Gender and Society. 2002;16:728–47.
  • Rubin D. “Formalizing Subjective Notions About the Effect of Nonrespondents in Sample Surveys” Journal of the American Statistical Association. 1977;72:538–43.
  • Stone P. Opting Out? Why Women Really Quit Careers and Head Home. Berkeley: University of California Press; 2007.
  • Tilly C. Half a Job: Bad and Good Part-Time Jobs in a Changing Labor Market. Philadelphia: Temple University Press; 1996.
  • U.S. Commission on Civil Rights . The Economic Status of Black Women: An Exploratory Investigation. Washington, DC: 1990.
  • Western B, Pettit B. “Black-White Wage Inequality, Employment Rates, and Incarceration” American Journal of Sociology. 2005;111:553–78.

Articles from Demography are provided here courtesy of The Population Association of America